On Tuesday Americans elect a new House and a third of the Senate. On the House, the market has made up its mind: prediction markets give Democrats a 94% chance of taking it, and about seven-in-ten of taking the whole Congress. The uncertainty sits in the Senate.

And yet, in the week before the vote, the people moving credit spreads weren’t candidates. On Wednesday the Fed held at 3.75–4.00%, with two dissents in favour of a hike. On Thursday the ECB held at 2.50%, said it had debated a hike, and its president announced she would leave early; the same day, US core inflation came in at 3.1% and growth at 3.4%. On Friday Intel lost 29% in a day and took the chip sector down 7% with it. Brent is at $113. High-yield spreads widened 12bp in a session.

Every midterm brings the same question from clients and PMs: what does the result do to my book? At index level, less than you’d think ! And I’ll show you seventy years of data that say so. At sector and issuer level, a lot ! A Democratic Congress, a split one and a Republican hold produce different winners in healthcare, power, telecom, banks, packaging, housing and gaming. Some of that is already in the price. Some isn’t.

This piece covers both. The first part is free: where markets stand this week, the calendar, the race, and what Congress can and can’t change. The second part is the market work: the history, the Fed-versus-politics regression, the five channels from Congress to spreads, data centres, eighteen sectors, 35 names, the cross-asset read, three scenarios and what I’d do. Charts throughout; skip to whatever you trade.

⚡ The short version

• The market has priced the House. Democrats are 94% to win it and about 71% to sweep Congress. The surprise outcomes are a split Congress (~24%) and a Republican hold (~6%). In the four Senate toss-ups, markets are more confident than the polls.

• Midterms barely move the index. Since 1954, high-grade spreads have moved anywhere from −68bp to +73bp in the three months after a midterm. Average: about zero. Ex-2002, the post-2000 average is 7bp tighter for high grade and +2.9% for high yield.

• The Fed does move it. Regress seventy years of spread changes on Fed policy and on who controls Washington: the Fed is significant (p=0.006), the party isn’t (p=0.411).

• This year’s run-in was rough. High-yield spreads widened 33bp in the three months to mid-October, the worst run-in of any midterm since 2002. Some of it retraced, then tech gave part of it back on Thursday.

• Tech had a bad week. Intel, one of the high-grade picks, lost 29% on Friday after a weak guide; Alphabet and Amazon fell after results. Hyperscaler capex plans held, so the supply story in tech credit is intact.

• Congress can’t do much on its own. The veto, executive tariffs and appointees who stay in place limit what a Democratic Congress can change. What it gets is hearings, subpoenas, the budget and the debt ceiling.

• Five channels, uneven. Fiscal is the broadest, regulation and M&A the most potent, trade the clearest winners and losers, affordability the most political, AI infrastructure the most durable.

• Data centres are the one thing both parties agree on. 58% of voters say they hurt the country. The pain concentrates in high-yield developers, merchant power and fibre, where there’s no balance sheet to wait out a delay.

• The sector map is where the vote bites: payors under a Democratic Congress, regional bank M&A under any Democratic gain, defence budgets, packaging and housing on the other side.

• 35 names. Twelve high-grade picks that work whoever wins, six high-grade pans, ten high-yield picks and seven pans.

• What I’d do: don’t trade the index on Tuesday’s result. Trade the Fed on 9 December. Own carry that works in all three outcomes, use relative-value pairs that need no view on the vote, and stay away from data-centre credits that can’t wait out a delay. Fuel beta is an Iran trade, not a midterm trade.

What’s inside

• This week: the Fed, then tech

• Part I — The calendar

• Part II — The race

• Part III — What Congress can and can’t change

• — paid subscribers from here —

• Part IV — Seventy years of midterms in high grade

• Part V — High yield and loans: the rally has a macro name

• Part VI — The Fed is the bigger Washington institution

• Part VII — Five channels from Congress to credit

• Part VIII — Data centres: the bipartisan backlash

• Part IX — Eighteen sectors

• Part X — The midterm book: 35 names

• Part XI — Cross-asset: Treasuries, the dollar, oil, equities, Europe

• Part XII — Three scenarios, every asset

• Part XIII — What I’d do

• Part XIV — The dashboard

📌 This week: the Fed, then tech

The vote is the fourth most important thing that happened to credit this week. Here’s where things stood at Friday’s close.

The Fed held, and two dissenters wanted more. The FOMC kept the target range at 3.75–4.00% by ten votes to two; Lisa Cook and Beth Hammack dissented in favour of another 25bp, on top of September’s hike, the first in three years. Chair Warsh didn’t pre-commit to anything. He doesn’t believe in forward guidance “absent a crisis”, and he said financial conditions have tightened but that he is “unwilling and unable to call them tight”. Growth is the reason. First-half GDP was revised up to 2.4%, the third quarter is tracking above 3%, business investment is growing at double digits and jobless claims are at their lowest “in generations”. Asked about Tuesday, he said politics played no role in the decision. The next morning’s data made his point: third-quarter GDP grew at a 3.4% annualised pace, and core PCE, the Fed’s preferred inflation measure, rose 3.1% on the year in September (headline 3.6%). By Friday, other Fed officials were signalling further hikes and futures priced a December move at about 76%.

The ECB held too, and talked about hiking. The deposit rate stays at 2.50%. Lagarde said a hike was debated, the decision to hold was unanimous, and the “direction of travel is quite clear” if the October assessment holds into December. Euro-area headline inflation rose to 3.8% in September, energy inflation to 18.9%. Then the surprise: she will step down after the February 2027 meeting, about nine months before her term ends. For a euro credit book, the December ECB and the succession race matter more than the Ohio Senate seat.

Credit tightened, then followed tech. US investment-grade spreads went from 91bp on 23 October to 86bp on Wednesday, then back to 89bp on Thursday, 10bp wider than at the start of the year. High yield went from 340bp to 314bp and back to 326bp, 45bp wider on the year; the last available print predates Friday’s sell-off. The high-yield run-in to this vote has been the weakest of any midterm since 2002; more on that in Part V.

Tech had its worst day in months. Intel fell 29% on Friday, its worst session since 1974, after guiding fourth-quarter revenue to $15–16bn against $17.1bn expected: an accident at its Israeli fab means an extended shutdown and fewer server chips, and management expects the PC slump to last through the first half of 2027. The chip index lost 7.2%, the Nasdaq 2.9%, the S&P 500 1.8%. Alphabet had already dropped 12% after its results, and Amazon fell after reporting negative free cash flow while keeping its 2026 capex plan at about $220bn. For credit, two things matter: Intel is one of the high-grade picks (Part X), and the hyperscalers aren’t slowing their spending, so the bond supply that comes with it isn’t slowing either.

Oil is the wild card, and it now has a date attached. Brent’s January contract closed at $113.43 on Friday, against about $72.5 before the Iran war began in February; the front month rose about 15% in October. The President has said there will be no strikes before the vote, and markets are bracing for what comes after. On Thursday night a tanker leaving Hormuz was hit by mines attached to its hull, the first successful attack of that kind since the war began. At the same time, the Vice President has written to Tehran and Qatari mediators are back there with a draft deal. Whatever Congress looks like on Wednesday, the oil curve will set the tone for airlines, chemicals, packaging and the Fed

💭 My View: The vote is a sector event that the market will try to trade as a macro event on Wednesday morning. I’d let it. The macro this autumn is a Fed that hiked in September and is debating another hike, an ECB that wants to follow, $113 oil with a strike threat dated for after the election, and a tech sector that has just reminded everyone how much of the index is AI. That’s what decides where the index is in March. The result on Tuesday decides which sectors outperform inside it.

Part I — The calendar 📅

Start with the dates. For credit, the vote is the beginning of a sequence, not the end of one.

Election night will give us the House, probably not the Senate. If the House goes the way the market expects, it should be called overnight or on Wednesday. The Senate may take longer. Alaska and Maine count with ranked-choice voting, which can add days, and in Georgia a Senate race goes to a run-off on 1 December if nobody clears 50%. The market will trade a probable Senate before it has a certified one.

Then the lame duck. The outgoing Congress has until 11 December to fund the government again; the stopgap it passed before the campaign runs out that day, and so do the highway programme’s authorisations. The FOMC meets on 8–9 December with new projections and a dot plot, and the ECB on 16–17 December.

Then 2027. The new Congress is sworn in on 3 January. The housing law’s limits on large institutional buyers of single-family homes start on 7 January. The US–China trade truce runs to January. Lagarde chairs her last meeting on 3–4 February. And somewhere between late winter and mid-summer, the Treasury runs out of room under the $41.1tn debt ceiling, by the Bipartisan Policy Center’s estimate. That is the channel through which a Democratic House can move rates volatility.

💭 My View: If I had to circle one date for credit spreads, it would be 9 December, not 3 November. If I had to circle a second, it would be whatever date in 2027 the debt ceiling becomes binding with a Democratic House holding the pen.

Part II — The race 🏁

The national numbers point one way. Democrats lead the generic congressional ballot by 8.3 points in the RealClearPolitics average and by about 9 in Silver Bulletin’s, within half a point of the cycle high. If that holds, it would be the biggest generic-ballot margin since 2008. The President’s approval is 37–38%, and only a quarter of voters approve of his handling of the cost of living. Gasoline averages $4.27 a gallon, against $3.00 on election day last year. That doesn’t help the party in power.

The House looks done. Democrats need a net three seats; Republicans hold 220–215. Prediction markets put their chances at 94% on both Kalshi and Polymarket, up from about 90% in mid-October. The forecasting models agree on the direction, less on the confidence: 95% at Silver Bulletin, 80% at Decision Desk HQ.

The Senate is the real election. Republicans hold 53 seats. Democrats need 51, a net gain of four, because the Vice President breaks ties. Twelve races decide it.

North Carolina, an open Republican seat, is the likeliest flip: Roy Cooper leads by ten points or more and trades at 97%. Ohio is the one I’d watch for credit. Sherrod Brown, trying to return to the Senate two years after losing his seat, has run against data centres and tied Jon Husted to them; Husted’s answer was the Ratepayer Protection Act, which would make data centres pay for their own power and grid upgrades. It passed the House in September with a big bipartisan majority and failed in the Senate. Brown leads by 2.6 points in the RealClearPolitics average. In Texas, Ken Paxton ousted John Cornyn for the nomination and now trails James Talarico by 2.1. Both candidates now say the data-centre industry should pay for its own infrastructure and lose its tax breaks. Maine (Troy Jackson +1.4 against Collins) and Alaska (Peltola +1.2 against Sullivan) lean towards the Democrats, Kansas is a tie, Iowa leans Republican by half a point, and Nebraska has an independent, Dan Osborn, whom the market gives about one chance in four. In South Carolina, the Republican, Graham Nordone, leads by 2.7. Democrats have to hold Georgia, New Hampshire and Michigan, and they lead in all three.

Markets have moved with the polls, and then some. In mid-October a Blue Wave was priced at 60–65%. It’s about 71% now on Polymarket, with a split Congress (Democratic House, Republican Senate) at 24% and a Republican sweep at 6%. Senate control is 71% on both platforms; the models range from 60% (Decision Desk HQ) to 78% (Silver Bulletin).

Down the ballot, two sets of races matter for credit. Thirty states elect attorneys general, sixteen of them currently Democratic and fourteen Republican. State AGs have become a real brake on media mergers even under a Republican federal government: Paramount Skydance needed a settlement before it could close on Warner Bros, and Nexstar’s integration of TEGNA is under a preliminary injunction after legal action started by a group of AGs. More Democratic AGs, more court challenges to consolidation. Then the governors. In Pennsylvania, Josh Shapiro, once a data-centre booster, signed an executive order in August making developers pay the full cost of their power and obtain local approval; his challenger Stacy Garrity has spoken in favour of a pause.

💭 My View: Seven-in-ten odds for a Blue Wave means the market’s surprise scenario is a split Congress, at roughly one in four. That’s the outcome I’d want to have thought through, because it’s the one that would move sector spreads most on Wednesday. Look at the gap between polls and prices: in Ohio, Texas, Maine and Alaska, Democrats lead by one to three points and trade at around 70% in each. Polling misses tend to run in the same direction across states. A three-point miss towards the Republicans in those four would turn a Blue Wave into a split. Watch Ohio and Texas early in the night. If Brown and Talarico win, a Blue Wave is close to confirmed and the market will start pricing a 2027 of hearings, subpoenas and data-centre scrutiny.

One more thing. This campaign has been light on policy for an election of this size. Populism runs through both parties, and on data centres they sound almost the same. A smaller policy gap between the parties means less market impact from either result.

Part III — What Congress can and can’t change 🏛️

Before pricing anything, it helps to remember how little a midterm changes on its own.

The veto stays. The President keeps his pen whatever happens on Tuesday. Overriding a veto takes two-thirds of both chambers, which no realistic outcome delivers. A Democratic Congress can pass bills that reverse the healthcare cuts in last year’s tax-and-spending law, restore ACA subsidies or tighten drug pricing. Expect them to be vetoed. They are messages for 2028, not law.

Tariffs run through the executive. The Supreme Court struck down the emergency (IEEPA) tariffs on 20 February, and importers had claimed more than $94bn in refunds by early June. The administration rebuilt the wall on firmer statutes: a 12.5% Section 301 tariff on Chinese goods from 24 July, Section 301 tariffs of 10–12.5% on 60 countries once the temporary Section 122 tariffs lapsed, a 50% Section 338 tariff on Canadian wood, pulp and paper, and Section 232 duties on steel, aluminium, copper and vehicles that the court ruling never touched. None of these needs Congress. Narrowing them would take new legislation and, again, a veto-proof majority.

The agencies stay too. Antitrust sits with the FTC and the Justice Department, and both are run by confirmed appointees. So is the FCC. Bank deregulation is administered by the Fed, which has just finalised its stress-test changes, and a new Congress won’t undo those.

What a Democratic House does get is real, just slower: hearings and subpoenas, the power of the purse (the defence budget request is $1.5 trillion, and new housing programmes are authorised but not funded), leverage on the debt ceiling, and a louder voice on everything from FCC oversight to prediction markets. A Democratic Senate adds confirmations. Nominees, including a pending FCC commissioner, would move more slowly.

And a lot of what matters isn’t federal. Zoning, permits, water and utility rates sit with states and local governments. On data centres, that’s where the timeline is decided.

💭 My View: If you remember one thing from the free part: the index trades the Fed; the sectors trade Congress. A Blue Wave changes the agenda for hearings and the odds of M&A approvals. It doesn’t change the tariff regime or the Fed’s reaction function. That’s why seventy years of spreads barely care who wins, and why I spend most of the paid part on sectors and names.

Read more

At the end of September the 10-year OAT-Bund spread went through 140bp for the first time since 2012. This week it went through 170bp. On Wednesday the 10-year OAT yielded more than 5% for the first time since 2002; on Thursday the spread touched 178bp, and it closed the week at 172bp, levels last seen in the euro crisis. France now pays about 50bp more than Italy to borrow for ten years. In January the two were level. On Friday night Moody’s gave France a reprieve. The market hasn’t, yet.

And yet. Over the month in which OAT-Bund widened by +61bp, the euro investment-grade corporate index made money: +0.5%. Through early October, French corporate credit had lost about a third of what OATs had lost, and a fifth of what the CAC had. French companies borrow more cheaply than their own government beyond six years. The French names in the main euro CDS index still trade inside it.

That gap — between a sovereign the market is visibly nervous about and a corporate sector it mostly isn’t — is the reason I wrote this. Over the next six months every one of you will be asked some version of the same question: what does the French election do to my book? I wanted one document that answers it properly, asset by asset, with the numbers.

It’s long. The first part (where markets stand this Friday, the calendar, the race, the Assembly problem) is free and is the context everyone needs. The second part is the market work: the fiscal maths, the programmes costed, the rating path, OATs and the ECB backstop, then credit, banks, covered bonds, sectors and equities, a scenario matrix and what I’d do. Charts throughout; skip to whatever you trade.

⚡ The short version

• The calendar is longer than people think. Two rating reviews, a budget deadline and a manifesto before Christmas; first round 18 April, second round 2 May, and very likely a snap legislative election in June. The election that decides France’s fiscal path may be the second one.

• The polls point one way, history says be careful. Le Pen leads every first-round scenario (31–38%) and gained after the 17 October riots. Glucksmann’s primary win has turned second place into a three-way tie with Mélenchon and Philippe. Le Pen wins every run-off tested. Polls six months out have missed the eventual winner more often than not !

• Nobody is likely to get a working majority. A fragmented Assembly is the base case whoever wins. That limits radical policy, and it also limits fiscal consolidation.

• The arithmetic doesn’t care who wins. Deficit above 5% for a fifth year, debt above 120% of GDP by end-2027, interest heading from 2.2% towards 4% of GDP. No realistic scenario stabilises debt before 2030. 😓️

• The programmes are big on targets, light on costings. The most detailed so far, the RN’s €140bn roadmap, leans on optimistic growth and a third of its savings look legally fragile. The centrists promise 2% deficits by 2032 without the detail yet. LFI publishes in November.

• Ratings: Moody’s blinked, the single-A middle still isn’t priced. Moody’s kept Aa3 with a negative outlook on 23 October; a cut to A1 was in the price, so this is a reprieve, not a re-rating. S&P on 27 November is the review that matters.

• OATs: two regimes. A fiscal repricing the market can live with, or a fragmentation story that, paradoxically, brings the ECB closer. We’re still in the first, but this week’s 34bp move had some of the second’s features: thin liquidity, forced selling, no single trigger.

• What turns it. Five catalysts: a budget near 5%, a Bund rally that spreads follow, the ECB talking, the domestic bid coming back, cleaner technicals. On 23 October none was fully in place.

• Credit is the calm way to own France. The France premium in euro credit sits almost entirely in financials and subordination. Non-financials have barely moved.

• French banks are a France proxy, not a France problem. Strongest fundamentals in years, 1.1% of assets in OATs, and spreads that have widened three times as much as Europe’s. The relative value is in senior non-preferred and short-call AT1s.

• Covered bonds keep their AAA. A downgrade costs public-sector programmes collateral, not ratings.

• The sector risks are policy, not budgets. Power-price regulation for utilities, foreign policy for defence exporters, the postcode for domestic REITs.

• What I’d do: own diversified French credit over OATs, French bank SNP over SP, short-call AT1s, mortgage covered bonds; be patient on domestic REITs; hedge with the sovereign, not with corporates; don’t catch the OAT knife before two of the five catalysts are in place.

What’s inside

• This week: 34bp in five sessions

• Part I — The road to the Élysée

• Part II — The race

• Part III — The third half: the Assembly

• — paid subscribers from here —

• Part IV — The fiscal arithmetic nobody escapes

• Part V — The programmes, through a market lens

• Part VI — Ratings: how low can France go?

• Part VII — OATs: from 140bp to 172bp, and what turns it

• Part VIII — The ECB backstop, and its conditions

• Part IX — Credit: the calm way to own France

• Part X — French banks: a proxy, not a problem

• Part XI — Covered bonds: the AAA holds

• Part XII — Sectors: REITs, utilities and power, defence, AI

• Part XIII — Equities: a lot in the price?

• Part XIV — Five scenarios, every asset

• Part XV — What I’d do

• Part XVI — The dashboard

📌 This week: 34bp in five sessions

I finished the first draft of this piece with OAT-Bund at 138bp. Five sessions later it was 172bp. Here’s where things stood on Friday morning.

Three things in that table matter more than the headline.

• France has decoupled from Italy. BTP-Bund widened too, but by half as much: 17bp on the week against 34bp. The OAT now yields about 49bp more than the BTP. A year ago that sentence would have read as a typo.

• Bunds rallied while OATs sold off. The 10-year Bund is 14bp lower on the month. This is no longer a global rates story with a French beta on top. It’s France.

• Credit noticed, on a different scale. Euro IG spreads widened 12bp this week and high yield 25bp, against 34bp on OAT-Bund. Over the month, the IG index still returned +0.5%, because it prices off swaps and Bunds, not OATs. High yield (−2.1%) and Tier 1 (−1.3%) took the hit. That’s the thesis of Part IX in a few numbers.

What changed? Less than the price suggests. The Assembly started unpicking the 2027 budget: seven government articles deleted, the vote on the revenue side pushed back to 3 November, and an estimate from the economy minister that the amendments so far cost €10bn. Student-led protests that began in September over schools and the cost of living turned into a day of nationwide unrest on 17 October: 550,000 marchers by the government’s count, violence in several cities, well over a thousand arrests. Macron addressed the nation the next evening. The first poll taken after the riots put Le Pen at 36%, up two points in a week.

None of that is a 34bp story on its own, and the rates strategists I read agree: they put the move down to fragile sentiment, positioning, another round of washouts in the French front end and thin liquidity, rather than any single catalyst. That’s reassuring and not. Reassuring, because nothing fundamental broke this week. Not, because a market that moves 34bp on no news can move another 34bp on some.

Then, after Friday’s close, the reprieve. Moody’s kept France at Aa3 with a negative outlook, one notch above S&P and Fitch, when much of the market had braced for a cut to A1. Read the statement, though, and it’s a warning rather than a vote of confidence: slower growth and a higher deficit have pushed the debt burden “somewhat higher than previously expected”, yields “have risen more rapidly than we expected”, and the protests “could constrain fiscal consolidation efforts”. Its baseline is that a deficit-reducing 2027 budget will eventually be adopted, “possibly by decree”. The same day, the Assembly threw out the revenue side of the social security budget, 115 votes to 1.

The next ten days are still dense: the ECB meets on 28–29 October and the Assembly votes on the revenue side of the state budget on 3 November. Paid subscribers get the full catalyst checklist in Part VII.

Part I — The road to the Élysée 📅

Start with the dates, because the market will trade every one of them.

Before Christmas, a crowded calendar. Moody’s left France at Aa3, outlook negative, on 23 October; the ECB meets on 28–29 October; La France Insoumise presents the new version of its programme on 6–7 November; S&P reviews on 27 November. Underneath all of it sits the 2027 budget. The social security budget has already fallen at first reading: on 23 October the Assembly rejected its revenue side, which sends the government’s original text to the Senate from 9 November. The Assembly votes on the revenue side of the state budget on 3 November and on the whole bill on 17 November. From 17 December, when the 70-day constitutional limit runs out, the government could in principle enact it by ordinance, something no government of the Fifth Republic has done for a budget; Article 49.3 (adoption without a vote, unless a censure motion passes) is the more likely route. If nothing is in force by 1 January, France can run on a minimal “voted services” regime (spending limited to what keeps public services running), and the public finance inspectorate doesn’t rule out that lasting nine to twelve months.

12 March: the field narrows. To stand, a candidate needs 500 signatures from elected officials (mayors, MPs, senators) spread across at least 30 departments or overseas territories, validated by the Constitutional Council. Close to 40 people have declared. In 2022 twelve made the ballot; the record is sixteen, in 2002. The threshold is hard for anyone without a local network, so the list will shrink, and the number of names on it matters more than it sounds (more on that below).

5 April to 2 May: the official campaign. Strict equal airtime, regulated billposting, results on the evening of each vote. A detail worth knowing: the eve of the second round falls on 1 May, a day of union marches and political activity in France, but the electoral code imposes silence the day before a vote.

18 April and 2 May: the two rounds. The constitution requires the vote 20 to 35 days before the incumbent’s term ends. Emmanuel Macron’s mandate expires on 14 May, when the new president takes office.

June: the vote that might matter more. The new president is very likely to dissolve the National Assembly in the first days of the mandate to try to win a majority. Several candidates have said as much. The legislative election then has to take place within 20 to 40 days: a first round on 6, 13 or 20 June. Macron could in theory dissolve early to synchronise the two votes; he has ruled out another dissolution before the scheduled elections, and that isn’t the base case.

So the realistic window in which French fiscal policy gets decided runs from now to late June 2027, with the budget fight at the start and the legislative election at the end. The presidential vote is the middle of the story, not the end.

Part II — The race 🏁

First round: Le Pen, then a three-way fight

Marine Le Pen (RN) reaches the second round in every scenario pollsters have tested. Her first-round score ranges from 31% to 38% depending on the poll and the configuration, well above the 23% she took in 2022, and the latest barometer has her climbing: 32% in late September, 34% in mid-October, 36% after the 17 October riots. The July appeal ruling left her eligible to stand; a final appeal is pending, with a ruling expected before the first round.

Behind her it’s crowded, and it got more crowded on 11 October. Raphaël Glucksmann (Place Publique) won the social-democratic primary in the first round with 66% of the vote and is now also the candidate of the Socialist Party. He has ruled out any deal with La France Insoumise. Before the primary he polled 9.5–11%; since, 11–16.5%. Jean-Luc Mélenchon (LFI) is at 15–18% in most polls. The centre is still the swing variable: Édouard Philippe (Horizons) polls around 12–13% if Gabriel Attal (Renaissance) also runs, and 15–18% in most polls as the single centrist candidate. Bruno Retailleau (Les Républicains) polls 5–10% and is struggling to clear 10% nationally despite a strong local network. David Lisnard on the liberal right now polls 4–7% when tested. François Hollande, who stayed out of the primary, is being tested as a fallback for the left (7–10%).

Put the single-centrist scenarios side by side and the picture is a three-way tie for second place, with Glucksmann the only one of the three clearly moving. Mélenchon hasn’t fallen much, but he no longer has the left to himself.

Three things decide who joins Le Pen in the run-off:

• How many names are on the ballot. A crowded field lowers the qualifying threshold and makes a run-off between the two ends of the spectrum more plausible. That’s why a fragmented centre is a market risk in itself, and why I expect some consolidation before March. In 2022 Mélenchon missed the second round by roughly 400,000 votes, half of what the Communist candidate took.

• Turnout. Low turnout tends to favour the extremes, whose voters are easier to mobilise.

• Tactical voting. “Vote utile” accelerates as polling day nears: voters drift towards whoever looks best placed to qualify and to win.

Second round: what the polls say, and what they can’t

If the vote were tomorrow, Le Pen would win every run-off pollsters have tested.

Against Mélenchon the margin is wide: 68–32 in the latest poll. That’s driven by his very high rejection rate and by vote transfers from the right (in an earlier survey, 68% of Retailleau voters would back Le Pen, 3% Mélenchon). Most centrist and Glucksmann voters would abstain or vote blank; among those who don’t, centrist voters lean Le Pen and Glucksmann voters lean Mélenchon.

Against everyone else it’s much closer, and remarkably uniform: 56–44 against Philippe, 57–43 against Glucksmann, 58–42 against Attal, 59–41 against Hollande. The transfers run the challenger’s way: in the earlier survey, 61% of Glucksmann voters, 47% of Retailleau’s and even 35% of LFI voters would back Philippe (most of the rest would abstain). A formal “front républicain” call, of the kind that shaped the 2024 legislative elections, would amplify those flows. And one statistic stands out for anyone pricing fiscal risk: 26% of voters now say debt and deficit reduction is the main driver of their vote.

Rejection rates tell the same story from the other side. Every leading candidate is rejected by a majority of voters. Le Pen is rejected least and has the largest block of committed support, which is what makes her the favourite. Mélenchon is rejected most.

Why I don’t trust any of this yet 😅️

Six months out, French presidential polls have a poor record of naming the winner !

In 2017, six months before the vote, Macron polled 15.7% and François Fillon was the projected winner. Macron scored 24% in the first round; polls only started putting him in the run-off in January, four months out, after the Fillon scandal broke. In 1995 Édouard Balladur was the favourite and Jacques Chirac won. In 2002 Lionel Jospin was expected in the run-off and didn’t make it. In 2012 polls were broadly right but hesitated over the Socialist candidate: eight months before the primary, François Hollande had 11%. In 2007 and 2022 they got the front-runner right.

The mechanism is consistent: tactical voting systematically overstates fading candidates a year out and understates rising ones. In 2022 Valérie Pécresse polled 12% a year before the vote and took 4.8%; Anne Hidalgo polled 5–10% and took 1.75%, because voters concentrated on the candidate best placed to beat Le Pen.

💭 My View: For markets, the first-round polls matter less than they look. What I watch is second place. A Le Pen–Mélenchon run-off is the configuration that pushes the France premium to its peak; a run-off between Le Pen and anyone else is a much more ordinary election from a bond-market perspective, whoever wins it.

Glucksmann’s rise matters more than people think. It splits the left vote Mélenchon needs to qualify, and a Le Pen–Glucksmann second round would carry none of the debt-cancellation or Eurosystem questions that LFI’s programme raises. Consolidation behind one candidate, centrist or social-democrat, before the 12 March deadline is the political catalyst I’d pay for.

Part III — The third half: the Assembly 🏛️

France’s legislature has two chambers. The Senate can’t be dissolved; the National Assembly can, and when the two disagree persistently the government can give the Assembly the final word. In June 2024 Macron used the dissolution power. What followed was two years of a fractured Assembly, no absolute majority for anyone, and a legislative paralysis that is still with us.

A new dissolution after the presidential vote is widely expected, and several contenders have announced it. That gives four basic scenarios: Le Pen elected with or without a majority, or a centrist elected with or without one.

Here’s the part I think the market underrates. Whoever wins, a majority isn’t guaranteed. The most likely outcome is another three-way split, reinforced by the two-round system and by tactical withdrawals of the kind that produced the 2024 Assembly. In that configuration, any president governs through alliances. That does two things at once: it limits the risk of radical economic policy, and it limits the capacity to consolidate the public finances.

The other possibility shouldn’t be dismissed: a honeymoon effect in which the new president sweeps the legislative vote because rival parties collapse after failing to make the run-off. It happened in 2017, when the president’s camp won 350 of 577 seats and Les Républicains lost 84.

💭 My View: If you remember one thing from the free part: the presidential election decides the direction, the legislative election decides whether anything happens. A president without a majority is the most likely outcome, and it’s also the one in which the deficit keeps drifting. Markets can live with a lot. They struggle with a fiscal path that nobody has the votes to change.

Thanks for reading this far. Everything below is the market work: 35 charts, the numbers behind the fiscal path, the programmes costed, what the agencies are likely to do, where the France premium sits across OATs, Credit, Banks, Covered bonds, Sectors and Equities, Five scenarios mapped to every asset, and how I’d position.

Read more

That’s how much of next year’s global earnings growth consensus now expects to come from one sector, Information Technology. In thirty years, excluding the years when earnings fell, only the 2022 energy shock produced anything more concentrated. The top sector normally does about forty percent. In the US alone the number is closer to four-fifths.

I’m not an equity investor, so my first question wasn’t whether that’s a bubble. It was narrower: what does a market this dependent on one theme do to my book when the theme wobbles?

This week we got a live test !

🔌 What actually cracked

Start with the bit that made the screens go red. CoreWeave’s euro bonds due 2032 came to market in June. By Wednesday their spread was over 1,000bp, a yield around 15%, after dropping nearly ten points at one stage. Four months.

In addition to debt tied to CoreWeave, bonds funding data centers developed by Tract Capital Management, Prime Data Centers, and Cipher Digital all fell this week to their lowest levels since issuance.

The trigger list was a cocktail: the CEO of a large chipmaker warning about excessive speculation in AI, fresh worries about Chinese competition in semiconductors, and a hot US inflation print on top. The semiconductor index fell as much as 11% intraweek before clawing some back.

The part I find more interesting is the divergence. Tech equities hit another all-time high relative to global equities this week. Over the same days, CDS on US chipmakers and on the hyperscalers themselves hit new cycle wides. Oracle’s five-year CDS jumped to a record.

Euro credit didn’t escape entirely. The euro hyperscaler bonds widened to their widest level of the year and are now visibly peeling away from the rest of senior non-financials. Further down the curve, roughly a third of euro high yield trading at distressed levels is now in tech, media and telecoms.

Why now? The cost-of-capital maths moved against the sector as the long end sold off, some US data-centre projects are running into local political resistance, and the price of a token keeps falling. Usage is rising exponentially; revenue per unit isn’t. When you’ve borrowed at 8.5% to build the capacity, that ratio matters.

💭 My View

Credit tends to notice before equities. Not always right, but usually first, because bondholders only get the downside and they price it like it.

What I’d take from this week isn’t “AI is over”. It’s that the newest, most levered part of the build-out has started paying a real cost of capital. That’s a healthy thing for a market to start doing. It’s just not comfortable to watch.

🇪🇺 Why a euro book can mostly shrug at the earnings story

Here’s the thing about two-thirds. It’s mostly an American number.

Look at consensus for the third quarter now being reported. In the US, tech earnings are expected up around 65% on the year. In Europe: 6%. The European headline is being carried by energy, materials and industrials, the old economy, which is also roughly what sits in a euro corporate bond index.

And the backdrop for those old-economy names is decent. Guidance coming into the season is positive at a rate near the highs. Global PMIs are at multi-year highs. European operating margins are running well above pre-Covid. Equity strategists going into the season were reading the credit widening as a rates-and-politics story rather than any fundamental deterioration, and nothing this week contradicted that.

So for euro IG fundamentals, the AI concentration isn’t the risk. The risk reaches us through two other channels. One of them is, oddly, good news.

⚖️ Channel one: supply, and why a little fear helps

For about a year, the single most common complaint I’ve heard from IG investors has been hyperscaler supply. Every few weeks another jumbo deal, another long-dated tranche, another set of concessions to absorb.

Most of that borrowing has been in dollars. Of roughly $252bn-equivalent the big hyperscalers have issued this year, about a tenth has come in euros.

The full-year forecast for euro hyperscaler issuance was €50bn. Issued so far: about €23.5bn. There are maybe six weeks of primary left before the holidays. A week like this one doesn’t make a treasurer rush to print a ten-year in euros (although most AI financial directors don’t seem to pay too much attention to their spread levels & borrowing costs these days haha)

That’s the good kind of AI panic. It doesn’t change anyone’s fundamentals; it just delays a supply wave the market has been bracing for all year.

The bad kind is different. Data-centre construction and the wealth effect from AI-driven equity gains have been doing a lot of the work holding global growth up. If financing conditions tighten enough to stop the capex, not slow it, that hits growth, and Europe would not decouple. When the US sneezes, and so on.

💭 My View

I want this wobble to last exactly long enough to scare a few euro deals into next year. Not a week longer.

Last week I was underweight euro hyperscalers into Q4 supply. Half of that thesis has now played out (they’re at the wides of the year) and the other half, the supply itself, may not show up. I’m taking profit on half.

🧭 Channel two: where a bad widening travels

If tech credit does have a bad widening, it won’t stay in tech.

The sectors the market has priced as AI “enablers” (capital goods, some utilities, the people selling power, cooling and switchgear into the build-out) have seen forward earnings estimates jump this year. Their spreads tightened while hyperscaler spreads went the other way. They’ve been trading as the safe way to own the theme.

That’s exactly what makes them vulnerable. They carry an AI premium in their earnings expectations without carrying it in their spreads.

💭 My View

If I’m going to be underweight anything AI-related in euro IG, it’s the long end of the enablers, not tech itself. The stress is in the newest, most levered corner. The AA balance sheets will be fine. The 15-year capital goods bond priced as if the capex cycle never ends is the one I don’t want to own if it does.

📊 Meanwhile, euro credit did something sensible

While all this was going on, euro IG tightened on the week and returned about half a percent. Euro equities lost money. High yield widened, and the way it widened is the whole lesson of the week.

Investment grade tighter, BBs a little wider, single-Bs a lot wider, CCCs blown out. The AI names and the cable and telecom complex did the damage in the lower ratings. Up in quality, in a single chart.

Three other things moved in the right direction.

💰 The money came back. After weeks of outflows, euro IG funds took in money again, high yield fund flows went flat after four straight weeks of redemptions, and strategic bond funds swung back to inflows. One week isn’t a trend. But the flows are doing what they always do: following returns.

📏 Valuations stayed cheap. Measured against their own past year, both euro IG and euro high yield spreads screen more than two standard deviations cheap. Euro IG is the cheapest of euro, dollar and sterling on that measure. Historically, crossing that line has been a better entry signal than an exit one.

📅 The quiet buyer has a sales pitch again. Fixed-maturity funds (the fonds datés that have been one of the biggest sources of euro credit demand since 2022) have been seeing net outflows since June, the first sustained run since the product took off. Partly because older vintages lost money as yields rose. But the outflows have slowed to almost nothing, and three-to-five-year euro IG now yields over 4%, a full point more than in January. That’s a very easy fund to sell.

Add the supply picture (this is on course to be the weakest October for euro IG issuance since 2023, while the deals that do come are still well covered) and the technicals look better than they did a week ago.

💭 My View

Up-in-quality isn’t how you leave credit. It’s how you stay in it through weeks like this one: most of the carry, a fraction of the drama.

🥖 France, briefly

Last week was all France, so this is short. The one chart worth adding:

Since the sell-off started in early September, French credit has lost about a third of what OATs have, and a fifth of what the CAC has. The French names in the iTraxx Main CDS index still trade inside the index, while French sovereign CDS has been setting records. Systemic French risk is going through government bonds first, not through corporates.

The catch for anyone running a euro benchmark: France is close to 27% of euro credit indices, the largest country weight in most of them. That weight is the main reason euro IG lagged dollars and sterling in September, and why it may keep lagging for a while. That’s a reason to expect a lag, not a reason to be out.

Moody’s reviews France on Friday. A downgrade to A1 is broadly expected and, as far as I can tell, broadly priced. French covered bonds should keep their AAA through a one-notch move. And with little real consolidation in next year’s budget and the presidential election in April, I’d expect any relief rallies in OATs to be shorter and smaller than the last ones.

🗓️ The week ahead

• Wednesday: UK CPI.

• Thursday: euro area consumer confidence, French business climate.

• Friday: flash PMIs, then Moody’s on France.

• Earnings: Kering, Carrefour, Sodexo, Edenred, Michelin, among others.

• Then: the Fed on 27–28 October. Markets price the next hike by December rather than this month, so a hold isn’t a shock for the front end. The long end is another matter.

🧭 Where that leaves me

💭 My View

Still long euro IG, still buying dips. Two standard deviations cheap, flows turning, supply light. That combination doesn’t come around often.

In high yield, up in quality. BBs over single-Bs, and no hero trades in CCCs while the AI and cable names are being repriced.

Three-to-five-year IG. It’s the bucket the fixed-maturity funds buy, and a 4% handle is their best launch argument in years.

Underweight the long end of AI-enabler IG. That’s where a bad tech widening goes next.

Hyperscalers: half off the underweight. Wides of the year, and the supply that justified the trade may not come.

France: credit over OATs. Still the calm way to own it.

What would turn me: the hyperscalers cutting capex guidance when they report. That’s the bad kind of AI panic, and if it comes I’d rather hear it from them than from a headline.

Two-thirds of the world’s earnings growth from one sector is a strange thing to live next to. But for euro credit this week, the fear did more good than harm. I’ll take a little more of it.

📬 If this was useful, the Notes during the week take each piece further: the enablers, the fixed-maturity funds, France’s layers. Subscribe, and forward it to the colleague who still thinks AI risk is an equity problem. — @ZeGoodTrader

Levels as at 15–16 October 2026 unless stated. Earnings figures are consensus estimates. Fund flow, issuance and valuation figures are approximate. Nothing here is investment advice, and I don’t know what your mandate is.

Not September. The year. Every coupon clipped since January, gone. Over the same stretch the euro equity market is up more than seven percent, and the Crossover index, which is what most people reach for when they want to hedge high yield, is still in the green.

I spent most of this week trying to decide whether that gap is information or noise. My answer, in one line: it’s a sovereign problem that one part of the credit market has decided to treat as a credit problem. And the plumbing that would turn the first into the second (the plumbing that did exactly that in 2011) is mostly gone.

🥐 The fastest French sell-off in twenty years

Start with the sovereign, because everything else hangs off it.

Measured from the low before each episode, the OAT-Bund spread has widened faster this time than in 2008, faster than in 2011, faster than in the Covid shock. On Tuesday it gapped 15bp tighter on a shadow budget from Le Pen’s party promising a deficit under 3% by 2030. By Thursday it had given all of it back, which is roughly what the market thought of the arithmetic. Italy now moves with France almost tick for tick: the 20-day correlation between the two spreads is 99%.

The numbers underneath aren’t new, but they’re worth having in one place:

• Advanced economies have cut government debt by 11 points of GDP since the post-Covid peak. France: unchanged.

• France hasn’t run a primary budget surplus since 1974. 🤦

• Public debt is roughly level with the US. Private non-financial debt isn’t: 214% of GDP against 140%.

• The median French retiree takes home almost as much as the median worker.

None of which changes on a two-day view. It’s why I keep calling France a denominator problem rather than a deficit problem, and why I don’t expect OAT-Bund back at summer levels without either a budget the market can add up or a central bank that says more than it currently does. Lagarde reminded everyone this week that the ECB has tools against “disorderly” fragmentation. Subject to eligibility criteria. That second sentence is doing a lot of work.

🔌 Why this isn’t 2011: the wiring is cut

At the worst point of the euro crisis, euro investment grade traded around 330bp. It’s around 110 today. The sovereign stress is comparable in places; the credit stress isn’t close. The difference is transmission, and in 2011 it ran through three channels.

🏦 Banks owned their governments. Domestic sovereign bonds were a big slice of bank balance sheets, so a sovereign sell-off became a bank capital problem. Euro area banks held about 4.6% of their assets in local government debt in 2014. It’s under 3.2% now.

🏭 Companies borrowed from banks. In 2011 euro area firms (Italian firms especially) funded themselves mostly with loans. Once banks were paying more for money than their own clients, people started asking whether credit would simply stop flowing. Today a much bigger share of that debt sits in the bond market.

📈 So bank spreads blew out over corporate spreads. Periphery senior bank paper traded several hundred basis points wide of non-financials. That gap was the alarm bell.

This time? The gap between bank bail-in and senior non-financials in euro IG has moved 3bp since the start of September. Three. French banks have done worse than the rest of euro financials, by about 7bp, which is what you’d expect and still nothing like a funding scare.

The cleanest tell is the shape of credit curves. At the end of 2011 almost 70% of euro IG issuers had inverted 2s5s curves: the market charging more for the next two years than for the next five. Today you can barely find one.

One caveat, and it’s a real one. In 2011 the problem was the periphery, about 27% of eurozone GDP. Today it’s the core: France alone is over 18%. A smaller crisis in a bigger country is not automatically a smaller crisis.

💭 My View

If things get worse, this is the section to come back to. The test for “this is becoming a credit problem” isn’t OAT-Bund at 160. It’s the bank–corporate gap starting to widen, and issuer curves starting to invert.

Neither is happening. If either starts, I’m wrong, and I’d rather find out from those two charts than from a headline.

💹 Cash high yield is trading like an OAT

This is where it gets practical.

When OATs blew out at the end of September, euro HY cash took the brunt. The cash index spread widened about 70bp to its peak. Crossover widened 18bp. Same asset class, broadly the same companies, very different moves, and the cash move was bigger than higher Bund yields alone can explain.

Then the bounce. Crossover has retraced around 85% of its widening. The cash index, around 35%. The OAT-Bund spread? Around 40%.

Three bars, the whole story. Cash high yield is trading as a sovereign proxy, not as corporate credit. It’s being priced off France.

Why would cash behave like that? My read: cash is where European real money sits. Those funds get redeemed when returns turn negative, and they sell what they can rather than what they want to. Crossover is where faster money hedges, and it moved first in both directions.

💭 My View

For a cash buyer, this is the trade of the month. Euro HY cash is pricing a sovereign problem that, on everything in the previous section, isn’t a credit problem. If OAT-Bund calms down, cash should be first to catch up, because it’s the furthest behind.

If you want the recovery without naked exposure, the index is now the cheap leg to hedge with: it’s already done most of its retracing. And if risk-off comes back, Crossover’s usual higher beta should reassert itself from here, which helps the hedge. Long cash, a bit of index protection on top. Not complicated.

🎯 Three places the sell-off sorted things wrong

Sell-offs are blunt. They sell what’s liquid and what’s labelled. Three things stand out this time.

1. Rates-sensitives got 2022 treatment for a not-2022 problem

Utilities, banks and real estate, taken together, have underperformed euro IG by about 4.5 standard deviations. That’s as bad as the peak of the last hiking cycle, when Russia invaded Ukraine. They’ve also been hit harder in credit than in equities.

The 2022 comparison is exactly the problem. That was a demand shock and the ECB had to hike hard. This one is a supply shock, energy, and with France wobbling the ECB’s pushback on hike pricing gets louder by the week. The Banque de France governor’s line was that “the geopolitical shock is transmitting into a financial shock.” Central bankers don’t say that when they’re planning to add to it.

Now put that next to the bank number from earlier. If bail-in barely moved, the banks didn’t do most of those 4.5 sigmas. Utilities and real estate did. That’s where I’d look first.

2. Single-Bs got away with it. BBs didn’t

In a normal risk-off, single-Bs widen more than BBs. Higher beta, weaker balance sheets. This time BBs widened more, and the single-B/BB spread ratio collapsed from 2.22x to 1.84x in six weeks. BBs now sit at 92% of their 2026 range, the wide end of everything I track in euro credit. Single-Bs are at 66%.

Some of that is fair. Single-Bs are more exposed to growth than to rates, and growth is fine. But the history is lopsided. Across the 18 euro high yield drawdowns of more than 2% since 2004, single-Bs typically fell further and then bounced less. In every one of the 18 recoveries, BBs returned at least as much.

This time single-Bs barely fell. So they come into the recovery with less cushion than usual and the same historical handicap. Nice table from JPM right here!

3. Hyperscalers hid in plain sight

Over the last two weeks euro hyperscaler spreads widened about 4bp, against 9bp for the rest of senior non-financials. Three reasons: this was a European-issuer scare more than a euro-currency one, so US names looked defensive; there hasn’t been a euro hyperscaler deal since Alphabet in May; and they were already cheap after the summer supply wave.

The second reason expires soon. Q3 blackouts end, there are reports Meta is looking at euros, and the street expects more than $100bn of further hyperscaler supply before year-end. I’d treat these bonds as inverse beta from here: they outperform if the sovereign mess gets worse, they underperform if it fixes itself. I lean to the second, so I lean out.

📊 The flows are still bleeding. The order books aren’t

Euro IG funds just had their seventh straight week of outflows, the longest run since May 2022. Strategic bond funds: six weeks, the longest since October 2022.

Look inside it, though. Short-duration funds lost about €540m. Total euro IG outflows were €113m. Do the subtraction: everything that isn’t short-duration took in roughly €430m. People are selling the short end and buying duration.

And primary turned. This week’s euro IG deals were covered 4.5 times on average, against 3.4x for the year so far: the first week above 4x on meaningful supply since early July. Concessions came back to single digits on aggregate measures, and the deals traded tighter after pricing.

I’ve been writing about book attrition for over a month: books that looked huge at launch and shrank hard by pricing. This week it mostly stopped. Air Liquide’s 5-year and 10-year finished around 7x covered with almost no attrition, right after the company announced its first big buyback, €4bn, under pressure from an activist. Nobody seemed bothered.

💭 My View

Flows lag, books lead. Fund flows follow returns; they tell you what already happened. Order books are people choosing to commit money at a specific price, today.

The books turned first. I’d put more weight on that than on a seventh red bar.

🚦 What turns this from a trade into a call

Everything above is relative value. For an outright view you need rates volatility to peak, and the cleanest signal for that is US data rolling over: the point where hikes start to bite and the market can see the end of the cycle.

It isn’t happening yet. US hard data, global forward earnings estimates and hyperscaler capex forecasts are all still going up. The S&P, the Nasdaq-100 and global tech all made new highs this week. Thirty-year Treasuries hit a cycle high of 5.73% despite a softer payrolls print the Friday before. Rates markets price roughly 71bp of Fed hikes through June, 62bp from the ECB and 91bp from the Bank of England.

Good news is still bad news, in other words. US CPI on Wednesday is the next test.

🧭 Where that leaves me

💭 My View

Long euro HY cash, light Crossover protection on top. Cash is priced off France. I don’t think it should be.

BBs over single-Bs. The ratio has done in six weeks what usually takes a cycle, in the wrong direction.

Start adding rates-sensitive IG: utilities and real estate before banks. Banks never got cheap.

Underweight hyperscalers into Q4 supply.

Still underweight French non-cyclicals trading inside their own government curve. Roughly a third of French corporates now trade inside the OAT, which puts France in the same neighbourhood as Mexico and Brazil. Not a sentence I expected to write. Rich isn’t safe.

Watching two charts: the bank–corporate gap and the share of inverted curves. If either moves, all of the above gets revisited.

France is a real problem. It just isn’t a credit problem yet. The market has priced one corner of credit as if it were, and that’s the corner I want to own.

📬 If this was useful, the Notes during the week take each piece further. Subscribe, and forward it to the one colleague who’s still comparing this to 2011. — @ZeGoodTrader

Levels as at 8–9 October 2026. Returns are index total returns; spread changes are index OAS unless stated; retracement figures are approximate and as of 7 October. Nothing here is investment advice, and I don’t know what your mandate is.

That's a syndicate banker, quoted this week, and I have some sympathy. Ten-year Treasuries at 5.16%, thirties at 5.55%, both near twenty-five year highs. Bunds, Gilts, OATs and JGBs all at cycle highs together. Every new high yield deal that priced this week traded badly once it was freed up. The largest non-investment-grade bond in history came to market and dropped five points (thanks for the table with Paramount at the bottom 😅)

So: nothing is safe. Fine. But let’s see when looking at duration, seniority, fundamentals, subordination and what are the exceptions… 👀

The mechanism, finally in a chart

I have been arguing for a year that credit spreads don't track the level of rates, they track the volatility of rates — because volatility is what triggers redemptions, and redemptions are what force selling. Until this week I was making that argument without a picture (except the JPM that looks decent vs 5y Bund discussed in my prior report below)

But now BofA published a better one. Euro rates volatility on one axis, weekly credit fund flows on the other, and the relationship is close to mechanical: the higher the vol, the weaker the flow.

Which reframes the whole week. The number that mattered was not the 150 basis points on OAT-Bund, dramatic as that is. It was the MOVE index, up 16.6% on the week and 47% on the year.

And it gives you a falsifiable test. If this is a rates accident, the damage should sort by duration. If it's a credit event, it should sort by rating/subordination

Leveraged loans are, on any fundamental measure you choose, the riskiest of those three. Weakest documentation, highest leverage, most refinancing dependence. They were the only one that made money in September, because they carry no duration.

Now do it by sector inside high yield.

Nothing in that ranking tracks credit quality. It tracks years.

And one more, which is the cleanest of the three.

Test two: the fundamentals didn't move

Earnings compounding roughly three times faster than debt, with coverage comfortably above four times. In Europe the equivalent picture is less spectacular but pointing the same way: second-quarter EBITDA growth back to 4% after a flat first quarter, margins holding at 21.1%.

Whatever happened in September, it did not happen to the companies.

Test three: this is not 2022

Everyone reached for the 2022 analogy this week. One chart separates them, and almost nobody showed it… (I mean BofA did but who read BAML research really? 😅)

In 2022, essentially no part of the credit market offered a positive real yield. You were being paid a negative real return to take corporate risk, which is exactly why the unwind was violent and indiscriminate — there was no level at which the asset class became obviously worth owning, so it just kept going (and money went out the window !! 😭)

Today a buyer stepping in gets a genuinely positive real yield across every rating band from AA to single-B. That is a floor. It is the single most underrated difference between this episode and the one everybody is pattern-matching to.

💭 My View — Three tests, three answers pointing the same way. This was a valuation reset, not a credit rupture, and I want to be clear that I'm not hedging that.

What would change my mind is narrow and specific: rates volatility staying elevated into third-quarter earnings, or those earnings showing consumer-led deterioration rather than cost pressure. Not a wider spread. Not another bad week in primary. Those two things.

The thing actually holding this up has nothing to do with credit

If you want to understand why spreads have absorbed a 25-year high in long yields with only moderate damage, stop looking at credit and look at earnings expectations. 💪

Global forward earnings expectations have been rising straight through a rate hiking cycle. That essentially does not happen. In late 2021, as the last tightening cycle got going, EPS expectations rolled over — and spreads followed them out of the door. This time the inflection point was upward, in early 2025, at the same moment as the DeepSeek shock.

So what is underwriting this market is a belief that AI delivers a step change in profitability for every company, not just the technology ones. Which means the chain that would actually break credit runs like this:

Long-end yields stay uncapped → hyperscaler cost of capital exceeds return on capital → capex gets cut → earnings expectations roll over → then spreads widen properly.

Watch that sequence rather than the ECB. It is the only path I can see from here to a genuine credit cycle, and none of the first three links have moved yet.

Exception one: France stopped being a story and became a problem

Two weeks ago I wrote that 110bp was the widest since 2012 and that it was the one genuine credit story in an otherwise rates-driven market. I was right about the direction and badly wrong about the speed. I thought this was a structural grind. It did forty basis points in nine sessions.

Two things changed its character. The budget didn't land — the market looked at the proposals, looked at the opposition's response, and concluded nothing had been resolved. And the contagion arrived: BTP-Bund is back above 100bp, so this is no longer a French problem that everyone else gets to watch.

The underlying arithmetic is the part that doesn't get said clearly enough, and it isn't really about the deficit.

A country is not punished for the size of its deficit. It is punished for its deficit divided by its nominal growth. US is running a nominal GDP boom — the Atlanta Fed has the current quarter above 9% — which is why it can carry a 5.5% thirty-year yield and a large deficit simultaneously. France is barely back to its pre-Covid trend.

France has a numerator problem and a denominator problem. Only one of them is on the agenda in the budget debate.

💭 My View — The corporate consequence is the bit I'd actually act on, and it is counterintuitive.

Two more French long ends crossed under their own sovereign last week: Carrefour and ADP. The reflex is to read that as a compliment to the credit. It isn't. Carrefour still trades about 30bp wide of its equal-rated UK counterpart — so nothing improved about Carrefour. The benchmark deteriorated and dragged the bond through a line it never earned.

A corporate yielding through its sovereign has no room left to tighten, is priced off a benchmark still moving the wrong way, and sits in the jurisdiction causing the problem. That is a rich asset, not a safe one. ADP makes it sharper still: its long end crossed under OATs in the same week that a proposed infrastructure tax rise to 12.2% appeared, worth a high-single-digit earnings headwind. The bond got richer and the credit got worse, simultaneously.

I'd rather be paid in French banks and cyclicals, where the sovereign isn't competing for the same buyer and where at least I'm compensated for the correlation.

Exception two: the documents are loosening while the market is tightening

This is the one that genuinely bothers me, and it got almost no attention this week because the headline numbers were so loud.

Paramount Skydance raised roughly $52bn to fund the Warner Bros Discovery acquisition — $30bn of IG across eight tranches, $12.4bn of cross-border second liens, $9.46bn of loans. The second-lien piece is the largest non-IG cross-border bond issue ever recorded. Peak orders on the IG component reached about $109bn, among the largest books in history.

Two details inside that, neither of which made the headlines.

They pulled the proposed eight-year euro tranche. European demand was more tepid, and the issuer didn't want the duration either. What printed in euros was a five-year. And the dollars at the long end dropped more than five points out of the gate; the euros a couple. Portfolio managers were unusually blunt about why — overallocation, and insufficient premium for a media mega-merger whose synergies nobody wants to underwrite.

Now set that against what else cleared in the same five days. PureGym printed a sterling and euro dividend recapitalisation with looser covenants than the bonds it refinances — baskets upsized, calculation flexibilities loosened, value-leakage protections deteriorated, on an independent read. InPost stripped out a co-operation-agreement blocker, which had been attempted in the loan market before but never in European high yield.

And all of this with high yield funds recording their fourth consecutive week of outflows.

💭 My View: Record size, record documentation flexibility, weak secondary, four straight weeks of outflows. You can normally have two of those at once. Having all four in the same week is the only genuinely late-cycle signal in this week's data.

The honest counter-argument, which I think is strong: broader new issue performance held up outside Paramount, which suggests the indigestion was deal-specific rather than systemic. Credit access on the stress indices actually improved on the week. So this is not a market losing the ability to fund itself.

But document erosion is a lagging indicator that turns into a leading one. Covenants written in 2026 are what you live with in 2029. The pattern — borrowers who have performed well refinancing into more permissive terms, with pushback only where there is a fresh credit concern — is exactly how the last two cycles built up their eventual recovery-rate problem.

I'm not trading on it this quarter. I am writing it down.

One correction to something I've been saying

I have spent most of this year arguing for carry and treating “up in quality” as the obvious defensive expression. The Q2 European fundamentals have a line in them that undercuts that.

BBB leverage was stable last quarter. Single-A leverage rose almost 0.3x.

For two years the story has been BBB companies repairing balance sheets, and that part is still true. What has changed is that the top of the ladder is now moving the other way. The quality gap is closing from above rather than from below.

Which is a much better explanation than duration alone for why AAA and AA have been the worst-performing buckets in euro credit this year. You are not being paid for quality that is quietly eroding underneath you. I'd stop treating the top of the stack as automatically defensive.

So where does that leave me

Constructive on the asset class, much more selective inside it, and watching exactly two things.

The flow data supports the constructive read more than the tape does. Money rotated out of spread and into government bonds and equities rather than being redeemed outright — and rotations reverse far faster than redemptions. Within high grade, long-dated funds actually took money in while short and intermediate funds bled, which is not a defensive posture. That is somebody buying duration into a sell-off because they think yields are closer to the top than the bottom.

Positioning:

• Long carry, and loans over bonds at the margin while duration rather than credit is what constrains demand. Loans still carry a yield premium over single-B bonds even after the reset.

• Liquid, non-distressed names that have underperformed. There’s a three and a half points of decile dispersion in a month. It’s a whole argument for doing the work.

• Underweight the French non-cyclicals that have crossed under their sovereign. Rich is not safe.

• Stop treating AAA and AA as the defensive answer. The leverage is going the wrong way at the top.

And the two questions, which point in opposite directions.
- Does rates volatility keep falling — it already came off hard on Friday, with hikes priced through June cut by roughly a third across the Fed, ECB and Bank of England curves? If yes, the last fortnight was an opportunity.
- Does OAT-Bund stabilise? If no, the first question stops mattering, and French banks leading the move wider is the transmission channel I'd watch for it.

Nothing is safe is a good line. It just isn't what the data says. What the data says is that almost nothing broke, two things quietly got worse, and the gap between the best and worst half of the market got very wide indeed.

Levels as at 2 October 2026. September performance figures are monthly index total returns; spread changes are index OAS unless stated. The 17 September OAT-Bund level is derived from a reported weekly change rather than a reported close. Nothing here is investment advice, and I don't know what your mandate is.

European IG credit spreads responded by widening 3 basis points.

That is the entire credit market’s answer to the biggest rates move in two decades. And yet euro IG total return is −1.38% on the year, with most of the damage arriving in the last month. Something is clearly wrong. It just isn’t credit.

Look at the order of that chart. It isn’t sorted by credit risk — if it were, the two investment grade markets would be at the top and high yield at the bottom. It’s sorted by duration. Floating-rate loans won. High yield, with its shorter maturities and fatter coupons, is still positive. The two long-duration high-grade markets are the losers, and the longer one loses more.

Nobody defaulted. Nothing broke. Your bonds just got repriced against a much higher risk-free rate, and that is a completely different problem with a completely different solution.

Three ways to prove it

One. Ask the synthetic market.

There is a derivatives index that references broadly the same European high yield universe as the cash market. Same companies, same default risk, same everything — except it carries no interest-rate duration at all.

Thirty-nine basis points apart, in opposite directions, on identical credit. That is about as clean a decomposition as this market ever hands you: on the synthetic read, the credit component of 2026 is negative. Every basis point of cash widening is duration and flows.

Two. Walk down the rating ladder.

In a genuine credit scare, quality protects you. Here it did the opposite.

Every rung down the ladder improved your return, all the way from AAA to single-B, because every rung down carries less duration and more coupon. AAA was the worst place to be in European credit this year. Read that sentence again if you run a conservative mandate.

And then CCC breaks the pattern, hard. Which is exactly right, and reassuring rather than alarming: CCC is the one band where actual default risk is large enough to overwhelm the rates effect. The ladder is telling you precisely where credit risk starts to matter, and it isn’t anywhere in investment grade.

Three. Regress spreads on the right variable.

One weekly I read put two scatter plots side by side: euro IG spreads against Bund yields, then the same spreads against Bund volatility. The first is mush — no relationship you could trade. The second is tight, and has been for a decade.

Spreads don’t care where rates are. They care how violently rates are moving, because violent moves trigger redemptions and redemptions are what actually widen credit. Which makes the size of this move the question that matters:

💭 My View — I need to sharpen something I wrote here a fortnight ago.

I said credit doesn’t fear the hike, it fears the cut. I still believe the destination half of that. But the framing was too clean. Credit doesn’t really have a view on direction at all. It has a view on the second derivative. Yields grinding higher over six months is fine. A hundred basis points in three weeks is not, whichever way they go.

So the useful question this week isn’t “are rates going up?” It’s “has rates volatility peaked?” Different question, different answer, different trade.

Who is actually selling

The mechanism playing out underneath all this is textbook to the point of being dull, and worth being precise about because two flow datasets appear to contradict each other.

Retail credit funds are redeeming — a couple of billion euros out of euro IG since the start of the month, overwhelmingly from mutual-style vehicles rather than ETFs. Meanwhile the broader dataset, the one that picks up institutional share classes, shows euro IG taking money in.

That isn’t a disagreement. It’s a handover. Retail money chases trailing total return, sees a 1.4% loss on the year, and leaves. Institutional money looks at a 4.46% index yield and buys.

Insurers are doing neither. They have anecdotally stepped aside entirely — not selling, just waiting for the dust to settle — which makes them the largest deferred bid in this market. History says they show up eventually, because insurers respond to higher yields by raising return targets rather than by permanently shrinking allocations. And the 2022 rates shock is where the entire fixed-maturity fund industry was born: higher sticker yields plus flatter curves is exactly the soil that product grows in.

Now the part that is genuinely a credit story

And it isn’t corporate.

Ten-year OAT-Bund hit 110bp this week — the widest since the eurozone crisis in 2012. There was no single catalyst, just a stack of them: global yields pressuring debt-sustainability arithmetic, a 2026 deficit revised up to 5.4% from a 5.0% target, funds getting stopped out of longs, and an October budget where the Prime Minister has said the 2027 deficit goes to 6.5% without cuts while hunting for €54bn of consolidation he may struggle to legislate.

The clean way to see what that means for credit is to drop the French sovereign into the euro investment grade rating ladder and see where it lands.

It lands between the index and BBB. A sovereign that is rated three to four notches above most of the companies domiciled in it is now priced as though it sits among them.

Which leads to the number that actually stopped me this week.

And the distribution is not what you would guess.

Health care and food. Transport infrastructure. Your defensive bucket — the paper you bought precisely so that you would not have to think about it. That’s the part now yielding less than the French government, while banks, autos and real estate are barely touched.

💭 My View — There’s a capital mechanism underneath this that turns it from a curiosity into something structural, and it doesn’t get said out loud enough.

Under the Solvency II standard formula, an insurer pays a zero spread capital charge on EU sovereign paper. Zero. On a corporate bond of the same maturity it pays a real one.

So the choice facing a French insurer right now is not 110 against 99. It is 110 that consumes no spread capital against 99 that does. On a return-on-capital basis that is not close, and it is not a credit judgement — it’s an accounting one. Which is worse, because accounting doesn’t change its mind when spreads start looking attractive again.

This is the first time in this cycle I’ve seen a structural argument for a persistently weaker high-grade bid rather than a cyclical one. I don’t think it derails the market. I do think it caps how tight the defensive non-cyclicals can get from here, and I’d rather be paid in banks and cyclicals where the sovereign isn’t competing.

For the bear case in reverse: Italian sovereign paper traded wide to almost the entire investment grade market

for most of 2010-15, and again through the 2018-20 populist coalition period. It happened, the market absorbed it, life went on. So this is survivable. French banks would not enjoy the acute version of it, and that’s the tail I’d hedge rather than the one I’d ignore.

Primary paid up. It didn’t help.

September euro IG issuance reached €73bn with days still to run — the biggest September since 2019. Sterling had its largest month on record in the dataset. High yield printed €9.8bn in a single week, the biggest week of the year.

Nobody is being shut out of anything, and that matters enormously.

It is the strongest fundamental argument for tighter spreads that currently exists: in April 2022 high yield primary froze solid, and the refinancing fear that followed was half the spread widening. This time the window is wide open and issuers are walking straight through it.

But look at what they had to pay, and what they got for it. Issuers handed out an average of nine basis points of new issue concession that week. Then this happened.

Nine basis points of free money, handed over, taken — and the paper cheapened anyway.

💭 My View — I started tracking book attrition instead of book size about a month ago because the headline cover ratio kept lying to me. It has now called the break performance four weeks running, and this is the most emphatic reading yet.

What one-out-of-thirty-three describes is an order book made of price-sensitive money rather than real money. Accounts go in at initial talk to see where the deal lands, then drop out when the spread is tightened. That is a market with a demand problem dressed up as a market with a supply problem.

Practical version: in this tape I want the deals where the book held, almost regardless of the concession. A small premium on a stable book beats a fat premium on a book that halved.

And then Volkswagen, to remind you that frameworks don’t help

A profit warning dropped at Friday’s close last week — Porsche impairments, implying the profitable segment stays difficult for a while yet, with no near-term recovery.

Six of the ten worst-performing euro investment grade bonds this week were Volkswagen lines. The perpetuals went wider too, autos was the worst sector in the index, and the whole OEM complex got dragged along, with parts and leasing outperforming almost by default. One agency has said it will take a view once it has clarity on cost timing and FY27 guidance. Downgrade risk has clearly gone up.

I can write two thousand words about rates volatility and Solvency II capital charges, and the single biggest hit in anyone’s euro IG portfolio this week came from a German carmaker’s press release on a Friday evening. Keep both things in your head.

Two things I’m filing away

Energy as a structural shock, not a spike. Commodity forecasts were revised up again this week — $95 Brent for the second half, around $80 average through 2027, with continued Middle East skirmishes as the baseline rather than the risk case. The credit expression of that isn’t a view on oil names, it’s an energy-intensity screen on what you already own. Retail, services, health care and telcos screen as least energy-intensive. Reverse Yankees screen better than their European peers on the same measure. And so do higher ratings — which means up-in-quality and hedging an energy shock turn out to be the same trade. Cheap filter to run.

The next AI panic is probably a buying opportunity. A new model launch hit bank stocks this week on the theory that agentic AI erodes the customer inertia behind cheap deposits. Semiconductors rallied. Fourth time we’ve run this tape in under two years. The precedent nobody quotes: in the first quarter of this year, when software obsolescence fear peaked, the “more AI-disruptable” spread basket underperformed hard — and not just in tech, with double-digit two-day equity drops across real estate, asset managers and materials. By June the entire gap had retraced. European software leveraged loans went from 89 to 96.

Europe’s credit index barely contains any technology.

AI obsolescence here is a slow fundamental risk that plays out over years, not a spread event that plays out over weeks. So when the indiscriminate selling arrives, and it will, the move is to buy it.

So where does that leave me

Constructive, with one reservation, and the reservation isn’t corporate.

Rising risk-free yields don’t frighten me. The volatility does, and it will fade. Fundamentals are emphatically not the problem: euro non-financial corporate debt is 65% of GDP, down on the year and eleven points off the 2010 peak, and net of €3.6tn of cash deposits it is 43% — the lowest in twenty-five years. Household debt service is low, savings are high. Second-quarter earnings growth turned positive again after a flat first quarter and margins held.

What I want before adding size is stability in OAT-Bund, not a peak in Bund yields. Those are different signals and only one of them is a credit signal.

Positioning, honestly:

• Long carry, and specifically single-B over BB. That chart isn’t a fluke, it’s duration, and the duration trade isn’t finished.

• The belly of the curve still — and I’ll admit it cost me this week. The front end is the only part of the curve that hasn’t repriced at all, which makes it expensive protection rather than cheap safety. Everyone crowds there in a rates scare, and that crowding is exactly why it stops paying.

• Underweight the defensive non-cyclicals, purely on the sovereign competition. Forty-two percent of health care and forty-one of food inside matched sovereigns is not a level I want to own into a French budget in October.

• Book attrition over book size in primary. Four for four.

Three basis points of widening and a 1.4% loss. If you spend this week arguing about the three, you’ll miss what actually happened.

AT1— the deeply subordinated instruments banks issue for regulatory capital, the ones that can skip coupons, may never be repaid, and sit first in line to absorb losses if the bank fails — have been 75% less volatile than high grade corporate bonds over the last ten days. lol

Not 75% more. Less.

For context on how strange that is: in 2023, when $17 billion of these things were written to zero in a single weekend, AT1 was roughly TEEEEN times more volatile than high grade. During the worst of the Iran war earlier this year, about twice as volatile. Today it runs at a quarter of the volatility of the boring stuff.

REALITY? Nothing about the instrument changed. What changed is what the market is actually frightened of!!

The capital structure inverted because the risk moved

Long-dated government bonds are the ones absorbing everything right now — inflation anxiety, fiscal anxiety, a wall of corporate supply. Investment grade sits right behind them, because investment grade is mostly a duration product wearing a credit costume.

AT1 has almost no rates sensitivity and almost no macro sensitivity. Its price is a function of a fat coupon and a call date. So when the volatility in the market is duration volatility rather than default volatility, the thing furthest down the capital structure ends up being the thing that moves least.

The proof point came on Thursday. US Treasuries handed back the entire rally that followed the Treasury Secretary’s intervention to stem rising yields. Global AT1 prices moved by less than two hundredths of a cent.

A fund manager I read this week described them, half-jokingly, as a risk-free asset. He was making a provocation, not a recommendation. But look at the last twelve months and the provocation is uncomfortably close to the data.

💭 My View — This is real, and almost everyone is drawing the wrong conclusion from it. AT1 is not low-volatility because it became safe. It is low-volatility because it is short spread duration, carry-driven paper in a market where all the volatility currently lives in the rates leg. That is a statement about the catalyst, not about the instrument.

Subordination is being rewarded. Duration is being punished. That is the exact inverse of the 2022 playbook, and it will invert again the moment a shock arrives with a banking transmission channel. I’m long the carry. I refuse to call it defensive, and I’d argue with anyone who books it as low-beta in a risk report. The beta is dormant, not dead.

Three things making me nervous here

The global AT1 index dropped below 200bp last week. First time ever. Perpetual bonds. Coupon can be skipped. First-loss capital. Under 2% spread.

Here’s what’s underneath that number.

The buyer base is changing — and not in a good way. Fixed-maturity funds holding perpetual AT1s have almost DOUBLED since November. Let that sink in. These funds are sold to retail on a promise of a maturity date. They’re buying bonds with no maturity date. I’ve been tracking fixed-maturity funds for months now as a structural problem — they’re just buying baskets to lock in a yield number, not doing actual credit work. Now they’re doing it in bank capital.

The bid is ALL carry. A survey out this week showed 80% of AT1 investors can hit their return target from coupon alone. That’s the asset class right there. Almost nobody is analyzing the bank. They’re analyzing the coupon.

Reset spreads aren’t being priced. A major European bank just printed a dollar AT1 at its tightest-ever reset spread. US banks have been doing the same thing with preferreds, setting post-crisis tights. But a reset spread isn’t just some technical detail. It’s the CALL MECHANISM. Tighter reset = less incentive to refinance = higher chance the perpetual never gets called.

Extension risk packaged as yield. And the worst possible holder of a non-called perpetual? A fund with a fixed maturity.

Look, European bank balance sheets HAVE improved. Bank equities outperformed the Mag 7 this year. If the next crisis has nothing to do with banks, maybe the old playbook doesn’t work. Fair point. But that’s not a reason to add exposure at 199bp.

Now run the same test in the other direction

If the bottom of the bank capital stack is trading like a safe asset, what is trading like junk?

Investment grade paper, as it turns out.

SpaceX brought a $25 billion benchmark across the curve. The 2056 tranche priced at 175 basis points over comparable Treasuries, tightened in from 200 in initial talk — the ordinary choreography of a deal that works. Within a month the spread was 231.

A $100 million allocation was worth about $90.7 million in under a month. The long-end Treasury selloff did some of that damage, but spread widening did more than two thirds of it. Over the nine sessions after the bonds entered the major indices at the end of June, it was the single worst-performing dollar triple-B benchmark bond — out of more than five and a half thousand bonds in that index, fourteen hundred of them a billion dollars or larger.

Oracle is running it close.

Plot the whole triple-B universe and the market is now quoting the two in line with each other. Overlay the average double-B curve across maturities and the risk being assigned to both of them is, quite plainly, junk risk. Two investment-grade rated issuers, priced by the market as high yield.

I wrote a fortnight ago that the useful way to cut AI-adjacent credit was hyperscalers funding from operating cash flow versus leveraged AI infrastructure borrowers funding from the bond market. I’d refine that now. The dividing line isn’t the sector at all. It’s whether the capex is funded by cash flow or by creditors. Both of these companies were modelled by analysts as tapping public debt markets as their primary external funding source for years to come. That profile is precisely what is now being charged double-B for.

💭 My View — Set this beside the AT1 section and you have the whole week in one line. The most subordinated instrument in European finance is trading like a rates-free safe asset, and two investment-grade issuers are trading like high yield. The labels have stopped describing the risk in either direction.

That’s not a market failure, it’s a market doing its job faster than the rating agencies can. But it does mean anyone running a mandate written in rating buckets is now holding something quite different from what the mandate says. And the euro read-across is coming: the dollar-euro funding differential still points these issuers toward European primary. When they arrive, we import the repricing.

So who is on the other side of all this?

Nobody defensive. That’s the honest answer, and it explains everything above.

Consider the backdrop for a moment. Hormuz is still closed. US growth leans on an AI cycle increasingly funded off balance sheet. The Fed may yet hike. Yields are climbing across the world, Japan has a bond problem, private credit is visibly stressed and leverage is engorged more or less everywhere.

Now the August reading from the most widely followed monthly survey of global fund managers:

• Cash down to 3.5% — the sixth lowest level since 1998

• Global equity allocation at net 56% overweight, the highest since November 2021

• Expectations for double-digit earnings growth at their highest since August 2021

• A record 56% now expect “no landing” for the global economy. Hard landing: 4%

• 72% expect no Fed hike before the November midterms

It is, by the survey’s own reckoning, the third most bullish reading on investor sentiment since 2022. The consensus is essentially: no landing, no hike, no AI capex cut, no bears.

Two details in there matter far more than the headline euphoria.

First, the one crack in the consensus is the exact one I’ve been worrying about. Yes, 72% say no Fed hike. But the “yes” camp went from roughly 14% in July to 22% in August — up eight points in a single month. The headline still reads “no hike”. The direction of travel reads the opposite. That is precisely the assumption I said at the top of this piece I’d stopped relying on.

And here is what should bother a euro credit investor more than any of it. The survey question is about the Fed, before the November midterms. The house forecasts I find most credible have one 25bp hike from each central bank — the Fed in December, after the midterms, entirely consistent with the survey majority. But the ECB in September.

That is three weeks away. The entire market is arguing about a Fed question that resolves in November while the European one lands before the autumn calendar has even opened. Wrong central bank, wrong deadline.

Second, credit is already pricing something the survey says nobody expects. “No AI capex cut” is part of the consensus. Meanwhile the bond market has just repriced two investment-grade-rated AI-linked issuers to double-B spreads, and euro paper from an AI infrastructure borrower issued only this summer trades in the high 80s at above a 10% yield — it fell another two points this week while high yield as a whole tightened.

Somebody has to be wrong. Those two things cannot both hold.

💭 My View — On the AI capex contradiction I side with credit, and for a structural reason rather than a directional hunch. The bond market is discriminating name by name — hyperscalers funding capex from operating cash flow versus borrowers funding it from creditors — while the survey asks a single binary question. Credit is answering a more precise question, so it is producing a more precise answer. I’d expect the survey consensus to converge toward the credit market’s view this autumn, not the other way round.

On positioning: 3.5% cash is what a market with no defensive capacity looks like. It explains the negative concessions on hybrids. It explains a sub-200bp CoCo index. It explains why order books could halve without hurting performance — there is simply nowhere else for the money to sit.

That is enormously supportive, right up until the moment it isn’t. Because the same condition that produces the bid also removes the marginal buyer who normally turns up when something gaps ten basis points wider. Rotate within risk. Don’t reload.

The label problem has a middle, too

Between those two extremes sits a third one that got less attention this week.

Thirty-year German government debt — the closest thing the euro area has to a risk-free asset — is now trading just twenty basis points tight to the euro investment grade corporate index. The gap between lending to the German state for three decades and lending to a diversified pool of European companies has narrowed to almost nothing.

Long-end yields are hitting multi-decade highs across several currencies. The usual suspects are there: rising government debt burdens, persistent deficits in Japan, the US, the UK and France. But note the newer driver, because it closes a loop with the section above — heavy issuance of long-dated corporate debt tied to AI capex, particularly in dollars, alongside a rebound in energy prices lifting inflation expectations.

The AI funding wave isn’t only repricing the companies doing it. It’s one of the forces pushing the entire long end higher, which then reprices everything else. That’s a feedback loop, and it is young.

Where I’d resist the panic: the move is most acute at the thirty-year point, which is not where European corporates actually borrow. And on the historical record, thirty-year German debt has traded wide of the euro IG index several times over the past two decades, occasionally by fifty or sixty basis points. Uncomfortable optics, not a regime break.

The reason it hasn’t hurt: duration. Euro corporate yields are up ten basis points on the month and that produced only minor losses, because the product is short. Which is the same composition argument I made earlier from the ratings angle — the index is shorter and better rated than the one everyone keeps comparing it to. This week it earned its keep.

Meanwhile, primary came back from the dead

The week before last, the entire European new issue market printed one deal. This week it printed nineteen issuers across twenty-three lines in four sessions — roughly €11bn in euros, another £1.6bn in sterling, heavily led by financials. The busiest week since the opening fortnight of July. High yield, notably, printed nothing at all.

And the flow data underneath it says something the headline number hides. European investment grade funds took in €376m on the week. Of which the short duration bucket alone accounted for €419m.

Read that again. The short-duration sleeve took in more than the entire asset class — meaning everything outside it was net negative. Money is still arriving in European credit. It is flatly refusing to take duration risk while doing so.

That is the cleanest confirmation of everything above. Investors are buying the spread and explicitly declining the rates leg. Same instinct that has first-loss bank capital outperforming and the thirty-year under pressure.

What made the drought interesting was that it was never purely seasonal. Refinancing premiums had blown out to the point where issuers had no economic reason to come. That arithmetic hasn’t fully reversed in four sessions, but the tape tells you the deferred calendar is real, queued, and waiting for September.

Here’s the part I got wrong, and I’d rather say so plainly.

All year I’ve treated attrition — order books shrinking between the first indication and final pricing — as a leading indicator of poor performance and wider concessions to come. This week the books were brutalised. A three-year corporate went from 5.4x covered to 3.0x, down 44%. A perpetual hybrid dropped 39%. A senior bank deal lost a third of its book. Six separate lines shed 30% or more.

And the average deal still tightened a basis point on the break. Only four of twenty went wider.

💭 My View — The attrition-equals-weakness relationship breaks down when the starting point is a supply drought, and the mechanism is obvious once you say it out loud. After a week with one deal, syndicates get buried in placeholder orders from accounts with cash and nothing to buy. Those orders were never real, so their disappearance signals nothing.

What actually predicted performance this week was direction of travel, not size. The single best-performing new issue of the week — 13bp tighter on the break — was the only deal in the entire tape whose book grew between first take and pricing. That’s the metric I’m watching from here.

And the hybrids priced through fair value

Two of the three corporate hybrids that came this week were priced through fair value — negative thirteen basis point concessions. Investors paid the issuer for the privilege.

The results sorted themselves cleanly. Where subordinated paper was made to pay a concession, it performed: one hybrid tranche came 10bp cheap to fair value and rallied 13bp. Where it was priced 13bp through fair value into a book that had already lost 39% of its size, it went wider and finished the week as one of the two worst performers on the tape.

Same issuer, two tranches, different call dates, opposite outcomes. Investors are being paid for extension risk and they know exactly what they’re doing when they take it.

Negative concessions on subordinated paper, into halved books, in the third week of August, is the clearest single sign of reach-for-yield I’ve seen this month.

So where is the actual deterioration?

Not in investment grade. European credit quality is splitting three ways, and the gap is widening:

• Investment grade: ratings drift +3.4% over six months — with net upgrades in every single month for two years

• High yield: -1.7% — deteriorating, slowly

• Leveraged loans: -6.0% — four years into a downtrend

Share

The first line matters more than people give it credit for. Part of why European investment grade “looks tight versus 2021” is that it is not the same index. The BBB share has fallen from 53% to 46%. Single-A has risen from 38% to 45%. Duration has come down from 5.4 years to 4.6. You are being asked to accept a tighter spread on a measurably better, shorter asset — and the upgrade machine that produced that shift is still running.

At the other end, CCC-rated loan issuers are running a net downgrade rate of -18% over six months, and the pool of debt rated triple-C-equivalent or worse has grown by €10bn in a year to €23bn. Though even there, the honest read is deceleration rather than acceleration: the worst-rated single-B cohort went through a genuinely ugly patch late last year, with drift averaging -13%, and that has since slowed to -3%.

Note where the bad numbers live. In the part of the market that trades least, and gets marked least often.

Which brings me to the $175 million invoice

I’ve been running a private credit red-flag list for over a year — payment-in-kind interest, dividend recaps, opaque valuations, ratings arbitrage, agency credibility, mislabelled exposures. This week produced a different kind of data point, and I think it’s the most informative one yet.

Jane Street raised $14.6 billion privately and used roughly $11 billion of it to retire its public debt. The new debt costs about 311 basis points against roughly 152 on the paper it replaced. That’s about $175 million a year of additional interest, plus around $200 million of one-time make-wholes to get out early.

This was not rescue financing. The firm had full access to the public market. It chose, voluntarily, to pay nine figures a year in order to reduce the number of creditors who get to see its financials.

So the privacy premium finally has a quote: roughly 159 basis points over the public alternative.

The punchline arrived days later, when a reported $15 billion July trading loss reached the press anyway, through an employee note. You can buy a smaller lender list. You cannot buy a smaller list of people who know things.

(That framing isn’t mine — it comes from an excellent piece by JunkBondInvestor this week, which is worth your time.)

💭 My View — I’ve deliberately been the measured voice against the “private credit is the next subprime” line, and I’m staying there. But this trade reframes the whole debate. The product being bought and sold in private credit is not principally yield, or covenants, or structuring. It is the absence of observers.

Borrowers buy it as confidentiality. Investors buy the smooth, uncorrelated-looking mark that confidentiality produces — what Asness calls volatility laundering. Both sides are paying for the same feature under two different names, and neither side has an incentive to say so out loud.

Which means the risk I’d actually underwrite here is not a default wave. It’s a repricing of the smoothness premium — slower, duller, and far more likely.

Subscribe now

The industry is building the thing that kills its own product

Watch what the biggest managers spent this summer doing, and hold it against what Jane Street just paid for.

One giant began marking its investment grade direct lending book daily from 1 July, with the full credit book on daily prices by October — explicitly so the assets can sit inside retirement accounts and daily-NAV wrappers. The same firm is the anchor client for a new persistent-identifier system that lets a private loan be tracked through amendments and transfers the way a CUSIP tracks a bond. Credit secondaries ran at $20.4 billion in the first half, double last year.

The motivation is rational and slightly desperate. Non-traded business development companies raised $2.0 billion last quarter, down 82% year on year, against roughly $5.8 billion of redemptions — nearly three times gross fundraising. The wealth channel just spent two quarters proving it will run. Getting into 401ks and insurance ledgers is the answer to that problem, and those systems cannot hold an asset with no number attached.

But liquidity is made of information, and every transaction creates a reference point that didn’t exist before. One manager moved roughly half its loan book into a secondary vehicle in a deal expected to cut net asset value by 10.4%. Another reportedly halved a planned continuation fund rather than accept the bids — which tells you exactly where the bids sat relative to carrying value.

I don’t think this ends with a real, continuously observable market. Trading needs strangers who can price the risk, and the asset class is organised specifically around not letting strangers price anything. What emerges is a permissioned market: enough prints to embarrass the marks, never enough participation to make the exit dependable.

💭 My View — Here’s the practical consequence, and it’s the most underrated relative value trade in credit right now. The direct lending premium over syndicated loans has compressed to the tightest of this cycle. That premium is the entire compensation a limited partner receives for illiquidity, no ratings, and quarterly marks — and it is being competed away at precisely the moment the marks are becoming contestable.

Meanwhile in public high yield you get the spread, the rating, a daily price, and the ability to actually sell the thing on a Tuesday afternoon. That has been an unfashionable argument for three years. It gets less unfashionable every quarter.

The index that keeps ejecting its own worst credits

One last piece, and it ties the whole thing together.

Investment grade spreads are the tightest since 1998. Virtually every rating band sits at multi-decade lows. There is exactly one exception, and it’s the bottom one.

Triple-C is the only band that hasn’t participated. What makes that measurable rather than vague is that the ratios between rating bands are normally extraordinarily stable — double-B trades around 1.8 times the spread of triple-B, triple-B around 1.6 times single-A, and those relationships have held within twenty basis points about two thirds of the time across thirty years.

Triple-C usually runs around 2.1 times single-B. It is currently around 3.4 times — the widest on record.

Loans have been worse than bonds. The triple-C loan index is down roughly 3.3% this year, worse even than leveraged loans to software companies. The hundred-odd triple-C loans totalling around $100 billion of face value have fallen from an average price near 83 a year ago to about 71.5, pushing yields past 27% from just under 20%.

One caveat before anyone repeats that 3.4x figure, because I nearly made this mistake myself: it is a dollar statistic. Europe does not look the same, and the numbers are worth running.

Euro triple-C sits at 1,191 basis points against euro single-B at 449. That’s a ratio of roughly 2.65x — stretched against the 2.1x long-run norm, but a long way from the dollar record. And the recent direction is the opposite one: euro triple-C tightened 31bp on the week and 110bp on the month, against a broad euro high yield index that was flat over the month. European junk has been rallying, not blowing out.

Where Europe does confirm the thesis is the total return line, and there it’s stark. Euro high yield has returned +1.87% this year. Double-Bs +1.84%, single-Bs +2.15%. Triple-C: -0.03%. Precisely nothing, despite a 14.8% yield to worst and a hard rally in the last month.

That is what it looks like when the carry is entirely consumed by price losses and defaults.

Before anyone reads that as a distress signal for the market as a whole, two things.

First, the “triple-C spread” isn’t a spread. It’s an average of a bimodal mess. Bonds paying more than 1,500 basis points over Treasuries contribute almost sixty percent of the entire index spread, and only about a quarter of bonds by weight trade wider than the index average. The illustration is almost comic: the largest dollar triple-C issuer is Altice’s American cable business, $15.5 billion of bonds at an average spread north of 3,500 basis points. The second largest is Altice France, $6.9 billion at 309. Same ultimate owner. Roughly 3,200 basis points apart, sitting in the same index.

Second, and this is the one that matters: the public triple-C index has become an adverse-selection residual. Its par value peaked around $240 billion in 2009 and now sits near $140 billion, while the rest of the high yield market more than doubled. Triple-C is down to about 8% of high yield.

Those issuers didn’t get better. They went private. Through 2023 and 2024 the average loan refinancing out of the syndicated market into private credit was rated B3, while loans coming back the other way were B2 — which mechanically improved the reported quality of the public loan index. And once a deal sits inside private credit it is effectively unrated and can’t be bucketed at all, so “triple-C loans in private credit” is a category that cannot be measured. Outside listed BDCs, those funds report nothing about how their holdings are performing.

💭 My View — This reframes the ratings numbers I quoted earlier. Investment grade drift at +3.4% and loan drift at -6.0% are both real, and both are partly a composition effect. The weakest credits have been walking out of rated public markets for a decade. When a B3 borrower refinances into an unrated private vehicle, the index it left looks healthier and the index it joined does not exist.

We are measuring the health of a pool that quietly ejects its sickest members. That doesn’t make the investment grade improvement fake — the upgrades and the shorter duration are genuine. It does mean the aggregate statistics flatter the true state of corporate credit.

So I don’t read a record triple-C ratio as broad distress. The sample is too small, too concentrated and too selected to carry that weight. I read it as the clearest evidence available that the credit cycle hasn’t disappeared, it has relocated — and that the leftover public triple-C tail is now the best free early-warning system anyone has on private credit, because it is the only part of that same underlying economy that still gets marked every single day.

Watch the triple-C to single-B ratio, not the triple-C spread level. The ratio is the signal. The level is a handful of blown-up names.

Where that leaves me

Long carry in investment grade and high yield, up in quality in high yield, and honest about the fact that the one assumption I’ve leaned on all year — that rates volatility stays subdued — is the one now under pressure. The AT1 data isn’t a curiosity. It is the evidence that volatility has migrated from the credit leg to the duration leg.

In subordinated paper I’d stop adding and get picky. In September’s reopening I’d rather be the one setting the concession than the one accepting a negative concession because the book looked big three days before pricing.

And I’d mark one date. Everyone is arguing about whether the Fed moves before November. The ECB decision is in three weeks. If you wanted the hedge, August was when you put it on.

And I’d keep watching the part of the market that doesn’t print. That is where the numbers are getting worse, and it is also where somebody just paid $175 million a year to make sure you can’t see them.

The thread running through every section of this note is the same one. A rating is a label, and the market has spent this month demonstrating that it will happily ignore the label in whichever direction it likes. First-loss bank capital priced as a safe asset. Investment grade priced as junk. A triple-C index too hollowed out to mean much of anything. And beneath all of it, a growing pool of credit that carries no label at all, because that turns out to be the thing people are willing to pay the most for.

If this was useful, share it with someone who still thinks subordinated bank debt is the risky part of a portfolio. 📬

Follow me on X @ZeGoodTrader for daily takes on Credit, Equities & Financial Markets.

All views expressed here are entirely personal and do not represent the views of any employer or institution. Nothing in this publication is investment advice, a solicitation, or a recommendation to buy or sell any security. Do your own work.

🚨 Worst monthly performance on record for tech momentum trades — down 39% in the US, down 23% in Europe. The Korean market has given back essentially its entire second-quarter tech surge, down 39% since late June. The Nasdaq dipped into bear market territory. Oracle’s five-year CDS widened to 215bp, wider than where it traded at the depths of the 2008 financial crisis. And a euro-denominated AI-infrastructure bond issued only last month is already yielding north of 10% — distressed, before its first coupon.

Now the European credit market’s response to all of that: IG finished the month 1bp tighter. Not wider. Tighter. HY tightened -8bp last week alone. Crossover is 63bp tighter YTD

Something has to explain that gap. I think two things do, and only one of them is being discussed.

1. July was a rotation, not a retreat

The tell is in what outperformed while momentum was being taken apart: financial stocks, healthcare, UK equities, and equal-weight indices. The equal-weight S&P has now been beating the Nasdaq since August of last year — a full twelve months of quiet broadening that nobody labelled a rotation until July made it violent.

In credit the dislocation is almost comically small. Euro IG tech spreads went +4bp wider on the month against a broad market -1bp tighter. Imagine that 4bp is the entire credit consequence of the largest tech positioning purge in the history of the series, not exactly impressive.

Meanwhile the consumer sector saw almost none of that positioning cleanout.

💭 My View — The reflex is to buy the TMT underperformance. I’d resist. Tech credit still looks tight relative to where the equity has travelled, and the supply pipeline behind it is enormous. The cheaper defensive right now is consumer credit, precisely because nothing has been flushed out of it. You’re buying a sector where the positioning risk hasn’t been created yet, rather than one where it’s been half-cleared.

2. Oracle’s CDS is wider than it was in 2008

Obviously it’s been very discussed, a company generating a large multiple of the cash flow it had in 2008, with a fundamentally stronger balance sheet, trading at wider default protection than it did during a global banking collapse.

This is not a solvency signal. It’s a plumbing signal. There is no clean way to short the AI capex cycle in cash markets — you can’t sell short a hyperscaler’s balance sheet, and the equity keeps confounding you. So the anxiety migrates to CDS, where it’s cheap to express and liquid to exit. It’s the same phenomenon as tech CDS volumes exploding from nothing to the most actively traded non-financial contracts in the market.

Once you read the level as positioning rather than fundamentals, it stops being scary and starts being useful. Wide CDS on a cash-rich issuer is a crowded hedge, and crowded hedges are the fuel for the next squeeze.

3. The 30-day distressed bond

CoreWeave’s €2bn euro bond was issued LAST MONTH at an 8.5% yield.

It is already trading around 11–12% 🔴

AI-linked. Distressed-style pricing before its first coupon payment.

Investors paid up for the story. Books were massive with 7bio of investment... (chart) and the market is now repricing the cash flows… Every cycle has one deal priced to the narrative instead of the fundamentals, this one took 30 days haha ⏱️.

Add it to the list I’ve been keeping. It sits alongside PIK usage creeping higher, dividend recaps at records, valuation opacity in private books, and software exposure that turns out to be mislabelled. Different wrappers, one common cause: capital raised against a story with a duration shorter than the bond’s.

💭 My View — I don’t think this is systemic. I do think it is diagnostic. When a brand-new bond breaks to distressed inside a month, the marginal buyer of that risk was not a credit investor — it was someone buying exposure to a theme. That buyer doesn’t average down. Watch the next AI-adjacent primary deal’s book quality far more closely than the spread it prints at.

4. The €100 billion coupon machine

Here’s a number that actually explains the resilience of the Credit markets (another one!)

Over the next twelve months the euro IG market is projected to pay out €100 billion in coupon payments. First time in its history. In April 2022 that same forward projection was €40 billion. Non-financials alone now account for just under €60bn of it. Euro HY adds a record €21bn on top.

That is the most underrated technical in European credit, and it’s underrated because it doesn’t show up as a “flow.” Coupon income is a structurally reinvested, price-insensitive bid. It doesn’t read the headlines. It doesn’t care whether spreads are at the 5th or the 50th percentile. It lands in accounts every month and most of it goes straight back into the asset class. And because it’s a function of the coupon stock rather than the price, it grows every time yields rise — it is the one technical that is strengthened, not weakened, by a selloff.

Now put it next to the bear case. The consensus worry for the next three years is that the AI buildout drowns European credit in supply. Fine. But the market’s absorption capacity has grown 150% in four years, and it is still compounding.

💭 My View — This is why I keep pushing back on the “spreads are too tight to absorb the AI supply wall” argument. It’s not wrong about the supply. It’s wrong about assuming the demand side is static. The right question isn’t “how much paper is coming” — it’s “which is compounding faster, the supply or the coupon bid?” That’s the whole trade for the next two years. It doesn’t make credit cheap. It makes the indigestion thesis a lot weaker than it sounds.

5. The index you’re quoting isn’t the index you own

Every valuation argument I’ve read this month anchors on the same comparison: 89bp today versus the 94bp low of 2021 (that’s when using the JPM EUR IG Corporate Index) but if you use the All Cash IG EUR OAS, you actually have a strong upside for further potential tightening (chart)

The problem is that the index changed underneath the number.

• BBB share of the euro IG benchmark: 53.4% at the 2021 peak → 45.9% today

• Single-A share: 37.9% → 44.7%

• Index duration: 5.4 years → 4.6 years

• Average maturity of new issuance: nearly 8 years → about 7

Hold the composition constant and 2021 would print roughly -6bp tighter than the headline suggests — much closer to where we are now. The benchmark has quietly become better rated and shorter dated while everyone argued about the spread level.

In reality one who could say (I also say it haha): you are not being paid less for the same risk. You are taking materially less risk for slightly less spread. That is a different sentence, and it’s the one that matters when you’re sizing a book. Less spread duration also means less to lose in a widening.

💭 My View — This does not make credit cheap. Adjusted or not, spreads are expensive against their own history and I’m not pretending otherwise. What it kills is the lazy version of the bear case — the “tightest since X, therefore sell” line. If you want to be short European credit, be short it on the supply calendar or on the energy shock, or even on Europe low Gas reserves. Don’t be short on a percentile that’s comparing two different indices.

6. AI is not yet paying for the energy shock

The optimistic corporate story of 2026 goes like this: yes, energy costs are up, gas inventories are down, but AI-driven efficiency will offset it at the margin line. Let’s check.

Median EBITDA margins across European credit issuers, split three ways:

• Tech vs non-tech: tech margins have expanded clearly over the past year. Non-tech has flatlined.

• By domicile: US issuers borrowing in euros have seen margins rise steadily. Core European issuers have gone sideways.

• By size: even the largest firms — the ones most likely to have deployed AI at scale — show no clear expansion.

And the euro area composite PMIs show input prices jumping while output prices barely move. That gap has a name, and it’s margin compression. Central bankers have said as much publicly: corporates are absorbing the cost shock rather than passing it through.

So the AI productivity dividend, to date, is landing almost exclusively on the issuers who need it least, while the energy bill lands on everyone.

Share

💭 My View — If your credit thesis rests on “AI margins will offset the energy shock,” the data says be patient. The counterweight is that ratings migration remains strongly positive — net €30bn of paper upgraded this month alone, part of the multi-year drift that turned the index more single-A than BBB. But upgrades are a lagging series and margins are a leading one. If the margin flatline persists into Q4, the upgrade cycle is where it shows up first.

7. Primary didn’t slow down. It stopped.

Last week: €1bn of euro IG— one single deal. Zero euro HY. Not “quiet.” Zero, kinda normal for this time of the year but still… What are HY investors buying in this context?

Some of that is simply August. But July had already run hot, with roughly €28bn of IG supply against a five-year average closer to €19bn — issuers front-running the shutdown. And the queue behind the silence is enormous: something on the order of €27bn of hyperscaler paper is expected in the September-to-November window alone.

On the other hand inflows have continued and I’m expecting even more money into Credit IG EUR considering the low rates volatility and the positive recent spread performance (chart).

8. And the tail is still stressed

One number keeps me honest about the index-level calm. BBs trade at 204bp. CCCs trade at 1,231bp, on a 15% yield. CCCs tightened 39bp last week — and are still 94bp wider year to date, in a year where the broad index went nowhere.

In this area this is all about the name picking. Chemicals is my favourite sector atm but we’ll go into this some other time.

Where I actually am

• Constructive, not enthusiastic. IG at 89bp on a better-rated, shorter-dated index with a compounding coupon bid. This is a carry market, not a compression market. Own it for the income, don’t underwrite tightening. But remains involved, Credit is today 2.9x less volatile than rates in average post pandemic and continue to offer its decent excess return. Why not? 😃

• Consumer over TMT in IG. 4bp of underperformance is not compensation for what’s coming in supply.

• Respect the CCC signal. Not by selling the index. By being ruthless about what’s in the bottom decile of the book.

• Hormuz is not fixed. The draft that markets rallied on reads restrictive, and there’s a parallel escalation broadening the conflict. Energy risk has been repriced to roughly zero. That’s a hedge worth owning cheaply, not a risk worth ignoring.

July told us what happens when positioning breaks and fundamentals don’t. Earnings grew 17% year on year last quarter — the strongest print since 2022 — through a war and a closed strait. That’s the fact that keeps getting buried under the AI noise.

The next test isn’t in August. It’s the first week of September, when the coupon machine meets the supply wall.

If this was useful, the best thing you can do is subscribe, share it with one person who trades credit, or upgrade to paid if you want the full weeklies & much more.

👉 Follow me @ZeGoodTrader for more on Credit, Equities & Financial Markets

These are my personal views only. Nothing here is investment advice, a recommendation, or a solicitation to buy or sell any security. Do your own work.

Let’s recap the first half like a trader, not a historian: a shooting war in the Gulf, the Strait of Hormuz effectively shut for four months, Brent up 33% at the peak, the ECB’s first hike since 2023, record gross supply. And euro IG spreads closed H1... one basis point wider. High yield: +16bp. The market stared down an energy crisis and shrugged.

The shrug was earned. Two straight years of net rating upgrades. Q1 EBITDA growth of +6% y/y — the best print since 4Q22 — and margins at an all-time high above 21%. Everyone came into the year positioned defensively, and yield-sensitive buyers showed up on every dip. Even the March wobble (110bp on IG, 375bp on HY) was shallower than the 2025 tariff episode.

CHART 1 — Euro IG & HY spreads, 2y history: iBoxx € Corp / € HY OAS

Where the Street landed

The mid-year outlooks are in, and they rhyme. JPM’s year-end target for euro IG: 90bp. Spot: 90bp. Their HY target: 300bp, versus ~310–318bp spot. BofA is even sunnier, calling for fresh post-GFC tights over the summer on their index measures. Strip out the branding and the message across the sell-side is identical: spreads go nowhere, carry is the return, keep clipping.

Here’s my issue. When every strategist’s December number is July’s spot price, you’re not being paid for anyone’s forecast. You’re being paid to lend at the tights and hope nothing happens. That can absolutely work — it usually does in carry regimes. But “nothing to do” markets are precisely where positioning builds quietly and exits shrink.

Three quiet cracks under the consensus

Read more

Meanwhile, my market shrugged off a genuinely ugly macro week — Brent +$4.5, 10y Bunds +13bp, Eurostoxx -2.2% on renewed Middle East escalation — and EUR IG tightened 3bp to 88bp, now trading through the consensus year-end targets set in January.

[CHART 1: Oil price vs 5y UST / Gilt / Bund yields — the great divorce]

1. Yield is the only arbiter left 📐

The mechanism is the one I’ve been hammering all year: in this regime, credit trades as a yield product, not a risk product. All-in EUR IG yields at 3.77% keep pulling in insurance and pension money on every backup, and last week spreads simply absorbed part of the rates move. Add peak summer seasonality and you get tightening into a war headline.

💭 My View — Here’s the uncomfortable corollary: the biggest H2 risk for spreads isn’t higher yields — it’s LOWER ones. A proper growth-scare rally in govies would drain the yield bid that has been the entire support structure. Everyone is hedged for a bond selloff; almost nobody for a bond rally. I’ve started building cheap exposure to that upside-down scenario.

2. The $200bn hyperscaler wall is causing indigestion 🏗️

One jumbo $ deal this week took global hyperscaler bond supply to almost $200bn year-to-date — and $20bn+ single prints are now routine. The demand side is blinking:

Oversubscription on these deals has fallen from ~4.7x in February to around 2x recently, and hyperscaler spreads now systematically underperform on jumbo-issuance days as investors make room for the next round.

But here’s the nuance the headlines miss: hyperscaler paper in euros, Swiss francs and Canadian dollars has been far less affected by the supply waves. Domestic bids in those currencies are providing genuine shelter while the $ primary machine rolls on.

[CHART 2: Hyperscaler oversubscription levels declining; spread reaction by currency after jumbo deals]

💭 My View — For a EUR-based book this is a gift: same credits, less supply pressure, and concessions that keep being paid to cross the Atlantic. I keep preferring EUR-denominated hyperscaler paper over $ tranches — and I treat jumbo-issuance days as entry points, not warnings.

3. A decade that humiliated “risk-free” 🏛️

Ten-year cumulative total returns, as of this week:

Japanese govies -19%. German govies -13%. French govies -11.5%.
EUR BBBs +14.5%. EUR BBs +43%. EUR HY +46%.

The 2020s are inflationary times — conflict, tariffs, geopolitics — and the balance-sheet discipline has been entirely on the corporate side: nearly four in five economies reduced corporate debt/GDP over the past two years, while barely half managed it for sovereign debt. Angry electorates demand spending; CFOs don’t have that luxury.

[CHART 3: 10y cumulative total returns — EUR BBs vs Bunds vs JGBs]

4. The rotation inside credit nobody talks about 🔄

Everyone watches equity rotations; the credit ones are more violent right now. € AT1 spreads tightened 3 standard deviations this month while retail-sector spreads widened 1.5 SD. Divergence like that is exceptionally rare in a technically-driven market. The trigger: subordinated bank paper was flagged as under-owned in June surveys — and under-positioning in a bull tape gets squeezed, brutally.

[CHART 4: Rolling z-scores of € AT1 vs € IG retail spreads]

💭 My View — At 3 SD I’m not chasing AT1s; corporate hybrids and T2 give me subordination beta without the crowding. The more interesting side is IG retail: the widening is positioning-driven, not fundamental, and it’s starting to screen attractive for buy-and-hold yield accounts.

5. Technicals: the bid re-accelerated 💶

EUR IG took in €880m last week — a second consecutive week, split across mutual-style funds (€502m), ETFs (€254m) and short-duration (€124m). Sterling IG and strategic funds were positive too. Primary stayed quiet (€13bn, blackout + US holiday) but healthy: a grid operator printed a €3.5bn four-tranche jumbo with books around 3x and near-zero concessions. Quiet, not broken — and gross supply is tracking full-year forecasts almost to the euro.

[CHART 5: Weekly IG fund flows; European credit supply]

6. The next AI battleground for credit: the UK 🇬🇧

Outside the US and China, the UK is emerging as the strongest AI contender — among the largest destinations for private AI investment over the past decade. UK software job postings are falling again; customer-service postings have collapsed while civil-engineering postings hold up. If February taught us anything, it’s the playbook: when AI disruption fear hits a credit market, own the “hard asset” sectors — telcos, utilities, real estate — against labour-heavy services. Sterling credit is next in line, and I’d argue the theme migrates to EUR services and staffing before long.

[CHART 6: UK job postings — customer service vs civil engineering]

The week ahead 🗓️

US CPI on Tuesday is the event of the week given the hike-risk repricing; then US PPI Wednesday, UK GDP and US retail sales Thursday. A hot CPI print into a market priced for perfection at 88bp — that’s the setup to watch.

Thanks for reading! If this was useful, share it with a colleague and follow me @ZeGoodTrader for daily takes on Credit, Equities & Financial Markets.

Views are strictly personal and do not constitute investment advice.

Beneath that calm index, though, the market is quietly splitting in two. Let me walk you through what I’m actually seeing — and where I’d put my money.

Where spreads stand — and where they go

Spreads have already clawed back about 65bp from the late-March wides. Put this year’s wobble next to the 2022 energy crisis or last spring’s ‘Liberation Day’ air-pocket and it barely registers on the chart. The consensus call is for spreads to drift back toward ~300bp into year-end, worth roughly a 5% total return on the year — with carry, not compression, doing the heavy lifting. The bolder voices see a summer squeeze toward 230–240bp before heavier post-summer supply nudges things back out to 250–260bp.

Chart: EUR HY spreads & yields

💭 My View: Direction right, magnitude limited. I’m constructive on a carry basis, but I’m not chasing compression here — the index simply doesn’t have much left to give. I’d rather earn the coupon and stay selective than bet on new tights.

No pain… so where’s the gain?

Spreads have been remarkably sticky, but the rates leg has done the damage: high yield has returned roughly +1.5% this year, a whisker above cash. The saving grace — and the reason the structural bid holds — is the shape of the asset class. Index duration sits at a record low near 2.8, while the average coupon has climbed to ~5.2% (from ~3.5% in late 2021). Translated: it now takes a ~210bp jump in yields to wipe out a full year’s carry. That’s a thick cushion, and it’s what keeps yield buyers glued in even at tight spreads.

💭 My View: This is why HY keeps behaving like a short-duration carry product, not a risk product. As long as the story stays ‘inflation & rates’ rather than ‘recession & defaults’, that buffer keeps the whole thing sticky.

Why the technicals still hold

The retail base has fully recovered its early-year withdrawals, with roughly €9.5bn of cumulative inflows since the start of 2025 — about 10% of assets. The pattern is simple: the money keeps coming so long as the worry is inflation and rates, not recession and defaults. Positioning backs that up — fund beta has recovered from a defensive ~87% to a neutral ~98%. Lower oil and a calmer rates path are the price of admission for that inflow run to continue.

[ Chart: EUR HY fund flows since Jan-25 ]

The primary market: busy, but not scary

Combined bond and loan supply hit about €170bn through May — running just ahead of last year — but net supply (~€32bn) isn’t overwhelming, and the mix tells the story. Issuance has skewed heavily to BBs (~66% of non-financial volume, versus the usual 50–55%), and corporate hybrids now make up a quarter of this year’s supply. Two things jump out: issuers are locking in fixed over floating, and the missing ingredient is event-driven supply — M&A and dividend recaps have both underwhelmed.

[ Chart: HY bond & leveraged loan origination ]

💭 My View: The supply backdrop is friendly — lots of refinancing, little net new risk. The day buyout and M&A financing genuinely re-accelerates is the day to start watching net supply. We’re not there yet.

The real story: single-Bs vs double-Bs

This is the part the headline index hides. Underneath the calm, the gap between single-B and double-B bonds has decompressed all the way back to its 2022 levels, even as BBs stay welded to BBBs. BB spreads sit around the 7th percentile of the last twenty years; their gap to BBBs is near the 6th. So the upside for the best junk is really just a rates bet. Single-Bs, by contrast, pay you about 260bp more than BBs — enough cushion to absorb idiosyncratic losses if defaults behave. Across desks, long-single-B-versus-BB screens as the single most attractive compression trade in all of credit.

[ Chart: B–BB & BB–BBB ratings curves ]

💭 My View: This is my preferred way to be long high yield — down in rating, not down in quality-of-business. Single-Bs give you carry and a cushion; BBs give you duration risk dressed up as safety. I’d be overweight single-Bs and I would not chase the BB–BBB compression here.

Ratings & defaults — the CCC canary

Ratings have been drifting lower since mid-2024, but the move is bifurcated: BBs have swung back to a positive net upgrade rate over the last three months, while single-Bs and especially CCCs keep deteriorating. Default forecasts are still benign — around 3.75% for bonds, with the market-based distressed ratio steady near 5%. The warning light is at the very bottom: the global CCC-to-single-B spread ratio just hit an all-time high.

[ Chart: HY up/downgrade rates by rating + CCC/B ratio ]

💭 My View: Index-level defaults are a non-event for now. But that record CCC/B ratio is the canary — distress is concentrating in the lowest tier, and that’s usually where a cycle starts, not at the index.

Loans vs bonds: has the loan trade run its course?

Since February, loans have outrun bonds, helped by a recovery in software loan prices as investors got more surgical about AI-disruption risk. But the loan market is also where the stress is showing: about 7% of the loan index now trades below 80 cents, largely software-driven, and loan defaults are projected to climb above bonds into 2027. With over half the loan index now bid above par — capped out — versus more room for price upside in bonds, I think the marginal call leans back to bonds. The one caveat: the CLO machine keeps performing loans extremely well bid.

[ Chart: loan vs bond price & software distress ]

💭 My View: The easy loan money is behind us. I’d rotate marginally back toward bonds for the convexity — while respecting the CLO bid. Where I’m most cautious: software-heavy loan structures. That’s exactly where CLO and private-credit exposure overlaps most.

The most counterintuitive trade of the year

Here’s one that surprised almost everyone: war was bullish for chemical credit. A blocked Strait of Hormuz throttled Chinese exports and broke the ‘dumping’ narrative that had crushed the sector for years — and HY chemicals tightened ~150bp, their best first half since the financial crisis. Defensives and short-duration cyclicals (oil & gas, healthcare, gaming, utilities) led the recovery, while media and technology screen as the clear laggards on AI and advertising fears. Software remains the obvious idiosyncratic risk pocket after a sharp February wobble that fully retraced.

Chart: HY sector 1m / YTD spread moves

Rising Stars: a double-edged tailwind

The upgrade pipeline is filling up — roughly €25bn of potential Rising Stars are flagged for the full year, versus just ~€4bn so far, with names like Renault, Teva and Carnival in the frame. It’s a tailwind for BB technicals and a reason the top of high yield stays well bid.

💭 My View: Rising Stars cut both ways. Great for the individual credit — but they pull the best paper out of high yield and leave the average quality lower. One more reason I lean to single-Bs for spread rather than crowding into already-tight BBs.

A valuation reality check

Strip out the widest 10% of bonds and the ‘clean’ index sits around 255bp — close to an all-time tight. Subtract expected default losses and the ‘excess spread’ you’re left with is roughly 210bp, below long-run averages. The takeaway: there’s genuine carry to be earned, but precious little cushion if the benign default base case turns out to be wrong.

Chart: HY excess spread / ex-10% index

💭 My View — the bottom line: I stay invested for the carry, down in rating (single-Bs) rather than down in quality, with a marginal tilt back to bonds over loans. I respect the technicals — flows, low duration, fat coupons — but I’m under no illusion that this is cheap. It’s a clip-the-coupon, stay-selective, watch-the-CCC-tail kind of market.

Thanks for reading. If this was useful, there are three ways to support the work — all of which genuinely help:

📧 Subscribe (free) — get the next deep-dive straight to your inbox.

💙 Upgrade to paid — if these notes save you time or sharpen your thinking, a paid subscription keeps them coming and funds the hours behind them.

🔁 Share it — send this to one person who trades, allocates, or just likes thinking about credit. That single share beats any algorithm.

Nothing here is investment advice. These are my personal views, not those of any employer, and you should do your own work before acting on any of it.

EUR IG spreads traded in a 4.5bp range over the past month. That’s a decade low for realised volatility. Outside summer holidays or the ECB-CSPP financial repression periods of 2017 and 2021, we have never seen the market this still.

And the 2017/2021 analog is not flattering. Back then, spreads stayed pinned because the ECB was hoovering up €5–10bn of corporate bonds every month and investors didn’t dare sell what they couldn’t replace. Today, the ECB is going the other way: QT just printed a record €7.9bn monthly drawdown in corporate holdings. So vol is dead despite the biggest official-sector seller of credit running at full speed. That’s odd…🧐

At the same time, primary markets are showing the first real cracks of the year. New issue premiums have backed up to 11bp (March wides), book coverage ratios are down to 2.7x — close to two-year lows. May printed €103bn of EUR IG supply, the largest single month on record. Indigestion is starting to show, even if secondary refuses to react.

👀️ My view: the macro repricing is now done in nominal rates and the geopolitical premium is still embedded in oil. But credit has stopped reacting to either. That kind of stillness has a name in this business, and it doesn’t end well.

Below, the credit roundup: the 16% paradox, why this isn’t 2022, the easing era ending, where margin pressure actually hits, and where I’d put money to work despite all of the above.

📊 Credit spreads are just 16% of yields. The other 84% is govies.

Here is the single most important chart in credit right now. Strategists’ work mapping the breakdown of corporate yields between the spread component and the government component shows the spread share is at 16% for global IG. That’s the lowest reading since July 2007.

Global BBBs are at the lowest spread-share since the GFC. Even Euro HY — the part of the market where spread is supposed to dominate — is now below 50% spread for the first time in years.

CHART: Global IG — credit spreads as % of total corporate bond yield, since Dec-2000. Lowest reading since July 2007. | Source: ICE Data Indices

Why does this matter? Because when 84% of your “credit” return comes from the risk-free curve, you’re not getting paid for credit risk anymore. You’re getting paid for duration. And the relationships that follow are uncomfortable:

• If the govie curve sells off harder (rates vol re-prices higher), the 16% cushion is paper-thin.

• The buyer base also changes character — retail, pension and insurance money that came in for the all-in yield doesn’t behave the same way as a dedicated credit fund when things wobble.

• And the last time we were here — July 2007 — we all know what happened next.

This doesn’t mean spreads need to widen tomorrow. The structural bid is real: 7–10y BBBs yield around 4%+ versus deposit rates that have come down. That gap is what’s pulling retail into IG funds even as institutions sit on their hands. It is the “glue” keeping the market in place. But it is glue, not concrete. 😅️

⚔️ 23 months vs 3 months — why this war is not 2022

Three months into the US-Iran conflict, here’s where Euro credit sits versus the same point in the Russia-Ukraine timeline:

• Euro IG: -6bp

• Bank bonds: -7bp

• Energy debt: -10bp

• Corporate hybrids: -15bp

• AT1: -17bp (doing extremely well YTD +2.24% - chart)

• Single-Bs: -18bp

Tighter. Across the board. Reminder: it took 23 months for Euro IG to retrace its 2022 widening. We’re three months in this time and credit is already net positive.

CHART: Euro IG spread performance around US-Iran vs Russia-Ukraine, both indexed to 100 at start of conflict. Visually striking divergence — 2022 went to ~125 (wider), today is at ~94 (tighter). | Source: ICE Data Indices

Why the asymmetry? Three reasons:

1. The curve. In 2022, ECB hikes flattened the front end and made corporate bonds less attractive than cash. In 2026, despite the rate repricing, curves have steepened. 7–10y BBBs offer a much more compelling yield versus bank deposit rates than they did three years ago. Retail follows yield. Yield is here 😎️

2. The starting point. 2022 began with policy rates at zero and a fully accommodative ECB. The rate adjustment needed to fight that inflation shock was enormous — 0 to 4%+ in eighteen months. Today we start with policy rates already at 2%, no post-pandemic overheating, and a much smaller adjustment required to tackle an oil-driven inflation pulse.

3. The buyer base has changed. All-in yields at 3.8%+ are a structural magnet for pension and insurance flows. Since 2023, every rate-vol spike has been met with more demand from yield-sensitive buyers, not less.

This is the bull case. Take it seriously.

The bear case is also real, and arrives in the next two sections.

🏦 14 hikes vs 13 cuts — the easing era is officially over.

Since the US-Iran conflict started, 14 central banks have hiked rates and 13 have cut. South Africa was the latest to flip hawkish. The synchronous global easing trade that defined the start of 2026 is dead.

Some numbers from this week’s macro repricing that should make every credit investor uncomfortable:

• 10y Bund yields broke entrenched multi-year resistance

• 30y gilts hit fresh multi-decade highs on UK political risk

• Long-end JGB total returns since the start of the conflict: -9.6%

• The 2061 Gilt now trades at a price of 22

• Fed STIRs went from pricing 6bp of cuts a week ago to fully discounting a hike at the March FOMC

• 75bp of ECB hikes are now priced through year-end. 68bp for the BoE.

The whole “rates vol stays low forever” consensus that anchored credit valuations into year-end has been dismantled in a matter of weeks.

CHART: Real policy rates (central bank rate minus headline inflation) — ECB, Fed, BoE, since 2003. All three still firmly negative despite the nominal repricing. | Source: Bloomberg, Haver

But — and this is the part I keep coming back to — real policy rates are still firmly negative. After the next 25bp ECB hike, real ECB policy rates will still be below -1%. That’s a textbook stimulatory backdrop. And historically, market accidents happen when real yields rise, not when they fall from already-negative levels.

So we’re in a strange place: nominal rates repricing aggressively, but real rates remain accommodative. Credit prefers the second number to the first. For now.

💭 My View: The credit pain trigger is not nominal rates re-pricing. It’s the moment real rates go positive. That requires the ECB to either hike harder than the curve currently prices, or inflation to fall faster than expected. The first is unlikely. The second is the actual scenario credit should fear most — disinflation forces nominal rates lower and real rates higher simultaneously. Watch the disinflation path more than the hiking path.

🔥 The chart European credit investors should be staring at

Composite PMI input prices are now rising materially faster than output prices in Europe. The textbook leading indicator of corporate margin compression.

What’s striking is that this move is faster than in the 2022/2023 inflation shock, and far more pronounced in Europe than in the US. The same chart in the US shows margin pressure but at a much more contained level.

[CHART: Composite PMI output prices LESS input prices — Euro Area vs US, since 2003. European compression looks staggering vs US developments. | Source: Haver]

This is the credit version of the macro thesis: inflation is a margin story before it is a default story. The companies that suffer first are those without pricing power — Transport, Chemicals, Consumer Discretionary, Leisure. The companies that absorb it are those that can pass through input costs into selling prices.

Looking at EBITDA margin volatility since 2000 across European IG sectors, the standout “pricing power champion” is Capital Goods. Lowest margin volatility through every cycle, including 2022. Utilities, branded Consumer Goods and Healthcare follow.

The losers list is unsurprising but worth keeping in front of you: Leisure & Hospitality, Transport (fuel costs eat margins, limited pricing power on shrinking volumes), and Chemicals (where China dumping is a structural overhang, not just cyclical).

If you believe in compression-down-in-quality during inflation eras — and the 2022/2023 data strongly supports this — then this is your map. Real yield opportunities are concentrated in Real Estate, Insurance, Leisure (yes, leisure HY has higher headline yields), and Utilities. At the IG retail end of the spectrum, only ~60% of bonds even offer a positive real yield right now.

🆕 Cracks are showing where you’d expect them first — primary

While secondary spreads sit dead-still in their 4.5bp range, primary is starting to tell a different story.

• New issue premiums: back to 11bp on a 20-deal rolling average. That’s the March wides. Up from ~1bp earlier in the year.

• Book coverage: 2.7x, close to two-year lows. The structural bid is becoming selective.

  

• Supply: €103bn in May — the largest single month on record. €365bn YTD, also a new H1 record.

• Fund flows: €4.3bn YTD inflow into EUR IG funds. That sounds positive, but it sits at the middle of the post-pandemic range — and was supposed to be the bumper year given all-in yields. This week saw an outflow of €840mm, mostly from short-duration funds that act as cash replacements.

[CHART: EUR IG new issue premium (20-deal rolling average), Jan-2021 to today — back to March wides at 11bp. | Source: market data]

The dynamic here matters. Heavy supply meeting weaker visible demand and aggressive ECB QT typically widens spreads. That correlation has held for years. This month it broke. The technicals index sits at a three-year low of -1.5%, which historically would correspond to meaningfully wider spreads — but secondary has just refused to react.

Pick your interpretation:

• Bull: institutional flows (harder to observe) are absorbing the supply quietly. Position covering is doing the rest.

• Bear: primary is the leading indicator. Secondary will catch up.

History sides with the bears here. Primary NIC widening is almost always followed by secondary widening, with a lag. The lag has been longer this time, but the relationship hasn’t broken — it’s just stretched.

🏗️ The hidden hyperscaler concentration nobody is pricing

The talked-about version of this story: hyperscaler debt is “only” ~3% of the global IG market. Not concerning. Telcos hit 16% of Euro IG in 2003, real estate hit 7% in 2022. So the panic is overdone.

The untold version: hyperscalers are now ~11% of the Euro IG 15+yr non-financial market. Because they love the long end. Reverse Yankees + 30y issuance + 100y deals all stack into one corner of the curve.

[CHART: Hyperscaler debt outstanding as % of 15y+ global IG non-financials, since 2015 — ~8% globally, ~11% of Euro IG long end. | Source: ICE Data Indices]

The actionable angle: the hyperscaler supply binge of H1 2026 is largely done. Cash generation in the sector is structurally high. So new supply pressure should ease into H2 — and given the long-end skew, this should mechanically support the very long end of Euro IG.

That’s a curve flattening trade, with a credit overlay. If you believe rates vol calms into the back half of the year (a big “if”), the 15+yr Euro IG segment is the asymmetric long.

💭 My View

Tactically cautious. Structurally constructive.

Vol cannot stay this low. A 4.5bp monthly range with 11bp NICs, record supply, ECB QT at full speed, and a global central bank cycle that has flipped from cutting to hiking is not a stable equilibrium. Spread targets I’d watch in a “Diplomatic Limbo” scenario where the Hormuz issue drags into the summer: EUR IG 160bp, EUR HY 450bp. That’s the asymmetric downside if anything wobbles.

But the structural underpinnings remain intact. All-in yields are an institutional magnet. Real policy rates are still negative. The retail bid for steeper curves is doing the heavy lifting. Earnings are holding. Inflation, however uncomfortable, is partly a margin story not a default story — and the margin story is sector-specific, not market-wide.

Trades I’d actually do here: own the pricing-power champions (Capital Goods, Utilities, branded Consumer Goods, regulated Healthcare). Carry hedges on iTraxx Main via August payer options — cheap insurance, asymmetric payoff if the inflation/oil narrative re-escalates. Long the very long end of Euro IG into H2 if you believe hyperscaler supply slows. Avoid Transport HY, Chemicals HY, and IG Consumer Discretionary — the margin compression list.

The line I keep returning to: when 84% of your credit return is the govie curve, it’s not credit risk you’re underwriting — it’s a duration product with a brand name.

That’s fine as long as govies behave. The minute they don’t, the 16% cushion is gone fast.

Watch real rates, not nominal rates. Watch primary NICs, not secondary spreads. Watch input PMI vs output PMI in Europe, not the headline number. The lead signals are all flashing yellow. Secondary just hasn’t decided to look up yet.

Have a great week.

Thanks for reading. If this was useful, share it with someone who pretends to read credit research but actually just reads my Substack. Subscribe if you haven’t — paid tier helps keep this independent!

Markets and EUR credit are back near their tightest levels of the year. iTraxx Main at 53bp — basically the all-clear scenario although with a bit of widening in cash bonds to 66bp (chart 3). The ceasefire narrative has squeezed spreads hard, iTraxx volumes have collapsed from their panic peaks, and primary markets just delivered one of their biggest weeks of the year despite the shortened calendar.

But beneath the euphoria, a few things don’t add up. EUR IG is trading at the 7th percentile of spreads since 2015. The spread-to-rates-vol ratio sits at the 2nd percentile. The Strait of Hormuz is still formally closed. PMIs just printed a 3-year low at 47.5. And May is historically one of the worst months for credit seasonality.

💭 My View

Near-term, I’m cautious. Risk-reward at 66bp IG (chart) doesn’t compensate for the tail risks that are still live. My preference is to hold hedges and wait for either (a) a confirmed Hormuz deal with real implementation, or (b) a technical reset wider. Medium-term, the structural backdrop — ample liquidity, strong fund flows, solid earnings, lower rates vol — remains genuinely supportive. This is not 2022. But it’s also not as cheap as markets are acting.

🌡️ Has stagflation arrived in Europe?

The Eurozone Composite PMI fell to 47.5pts in May — a 3-year low. GDP growth is running at close to zero. The details were universally soft: employment dropped to 48.4pts, new orders slipped to 47.1pts, input prices spiked to 70.5pts. France saw its lowest print since 2020 at 43.5pts. The UK Composite came in at 48.5pts.

Combined with rising inflation (oil-driven), this is a classic stagflationary setup — weak growth, higher prices. Central banks are now priced to hike: BoE +90bp by year-end, ECB +80bp, the Fed flipped to pricing cuts-to-hikes. UK and German 10y yields hit their highest levels in well over a decade.

[CHART: European Composite PMIs (Eurozone & UK), Jan ‘10–May ‘26 | Source: S&P Global / market data]

The saving grace vs 2022: policy rates are already at neutral (ECB at 2.0%), wage growth is decelerating (the US Quits rate has been falling for 18 months), and long-run inflation expectations remain anchored. The rate adjustment needed is far smaller than 2022. Credit is outperforming that analog precisely because the structural starting point is different — but that doesn’t make current valuations attractive.

💭 My View

I wouldn’t expect a full CB tightening cycle - what’s priced in massive! table below — more like emergency hikes at the margin if energy stays elevated. The key risk is yield curve flattening: bund 2s10s has reversed sharply from its steepening trend, and that steepening was the engine behind the 36-week IG inflow streak. If the curve flattens further, the technical bid for IG becomes structurally weaker. Watch the 2s10s — not the spread level — as the early warning signal.

🛢️ Iran: The “Mission Accomplished” thesis is getting harder to defend

Three months in, conviction on a rapid resolution is fading. The Strait of Hormuz remains essentially closed. Iran continues targeting regional infrastructure. The US is deploying additional forces. Trump’s rhetoric shifted from “very complete” to “all over the place.” A 15-point US peace plan exists — Iran and Israel are both ignoring it.

The asymmetry here is underappreciated. OECD crude inventories are approaching operational minimum levels, with commodity strategists forecasting year-end Brent at $95/bbl in a deal scenario, $115/bbl if Hormuz stays closed through June, rising ~$15/bbl for every additional month of closure. Ras Laffan LNG complex (17% of capacity) was hit — that’s a 3–5 year repair timeline, meaning structural gas price elevation even after a ceasefire.

[CHART: EUR IG Spreads vs Dec ‘26 Brent Future Price — Deal vs No Deal scenarios | Source: market data]

Market positioning has shifted from “every day matters” to “what’s an extra month?” — but the actual economic impact of delays is moving in the opposite direction: each additional week of production loss matters more as inventories become scarce. This creates the risk of a non-linear spread reaction if we stay stuck in diplomatic limbo.

💭 My View

My base case remains that the knee-jerk credit move will ultimately fade — historical precedent, military asymmetry, OPEC+ buffer, Trump’s political incentives all point that way. But the timeline has stretched materially from my initial assumption, and the Ras Laffan damage introduces a structural energy premium that survives any ceasefire. I still believe investors should add into meaningful weakness, but at 91bp IG today? There’s not enough cushion for the scenarios that are still credibly on the table.

📊 Technicals: The iTraxx skew signal you should be watching

One of the cleaner unwind signals of the week: iTraxx skews collapsed to year-to-date lows. Crossover skew went from 46bp at the March peak to just 7.4bp. Main skew compressed from 3bp to near parity at 0.1bp.

This matters because index skew turns positive in systemic bear markets (when macro protection is bought via indexes) and compresses in bull markets (as indexes rally faster than single names). The normalization tells you: most of the macro-driven protection put on during the Iran shock has been unwound. Index buyers who layered in hedges at wide levels are lifting.

[CHART: iTraxx Main & Crossover skew, Jan–May 2026 | Source: Bloomberg / market data]

Broader technicals remain genuinely supportive: IG funds registered a 36-week inflow streak (technically almost intact despite recent wobbles in momentum). HY funds bounced back strongly — €1bn of net inflows in one week (1% of total AUM). iTraxx outstanding positioning is at the 95th percentile of the past 5 years: investors were extremely hedged, and the unwind of those hedges is part of what’s driving the squeeze.

The flip side: EUR IG outflows registered last week (€229mm, 0.1% of AUM), book coverage ratios stayed at their lowest rolling average of the year (3.1x vs YTD average of 3.7x), and new issue premiums have been drifting wider — the rolling 20-bond moving average hit 13bp, a YTD high.

💭 My View

The skew collapse is the most bullish signal in the technical picture — it means the last of the fear-driven hedges is gone. But the NIC drift and declining book covers are telling you something different: the marginal buyer is getting selective. Both things can be true simultaneously. The inflow streak matters, but the quality of those inflows is deteriorating. I’d rather see 5x cover at flat NIC than 3x at +13bp.

🏗️ Primary: €50bn+ week and the supply wall is not done

EUR primary just delivered one of its biggest weeks of the year: €33bn of IG plus €3.2bn of HY, with the Thursday session alone among the busiest since April 9. YTD gross issuance is now €348bn EUR IG (+1% vs 2025) and €57bn EUR HY (+47% vs 2025).

The standout stories this week:

• SAP €3.5bn four-part deal — first EUR issuance in 6 years. The software giant returning to lock in rates amid AI disruption threats is the kind of signal that should make you think. They know what the market doesn’t about the revenue durability of their business.

• European banks locking in capital for a decade — AT1 “non-call 10s” now represent 25% of all major-currency AT1 issuance in 2026 vs near-zero historically. Santander ($1.5bn NC10), NatWest (£500mm callable 2036), Deutsche, BNP, UBS all issued long-call AT1s this year. Banks are front-running a regime shift.

• Eurozone sovereigns borrowing in foreign currencies at the fastest pace in years — already $4.6bn YTD vs $2.4bn for the same period in 2025. Belgium, Finland, Austria, Slovakia issuing in USD, CHF, AUD, even RMB. The ECB as buyer of last resort is gone, and smaller issuers are scrambling.

[CHART: EUR IG Monthly Issuance vs 2021–2025 average | Source: market data]

May is historically the largest single month for EUR IG issuance — the rush before summer. Combined with the record Hybrids pace (€21bn net YTD, tracking toward an all-time record) and continued Reverse Yankee momentum (€50bn 1Q), the supply picture looks like a structural force pushing NICs higher.

💭 My View

The AT1 NC10 story is the most structurally interesting development in European bank credit right now. At ~300bp for 10 years of Tier 1 capital, these are attractive for yield-hungry buyers — but for credit people, you’re owning double the duration in a tight-spread environment with thin secondary liquidity. The bull case (ABN Amro: “if equities rally the AT1 will need to move in tandem”) is real but contingent.

The bear case — rates spike, spreads widen, you’re stuck holding illiquid high-duration paper — is equally plausible at current levels.

📈 EUR IG index quality is improving — and hyperscalers are the reason

A structural shift worth flagging: the proportion of BBB-rated bonds in the EUR IG index just fell to 46.7% — the lowest level since 2016. It peaked at 53.4% in July 2021. This has been driven by two forces: sustained rating upgrades (€61bn net upgrades in April alone, the strongest month since November) and, more recently, the flood of AAA/AA-rated US hyperscaler issuance — already 0.9% of the benchmark from 0.2% at the start of the year.

With hyperscalers expected to borrow >€70bn in EUR this year (nearly 10% of total projected supply), the index quality improvement is set to continue. This has a direct implication for passive and maturity fund buyers: if you thought you were buying a “spread product,” you’re increasingly buying a rates product. A flood of AAA/AA paper from tech firms with minimal spreads changes the beta of the index.

[CHART: EUR IG Market Value Proportion by Rating (AAA/AA/A/BBB), 2003–2026 | Source: market data]

🚩 Private credit: European story is different from US — for now

As I’ve been highlighting for a year, red flags in private credit continue to accumulate in the US: PIK usage, record dividend recaps, opaque valuations, wrong labels, rating arbitrage, and now Blackstone’s BCRED experiencing record redemptions ($1.7bn). But the European picture in Q1 tells a more nuanced story.

European private credit origination stayed broadly stable at €17bn in Q1 26, roughly in line with Q3 25. Importantly, Software sector volumes remained resilient (19% of origination), despite the AI disruption narrative dominating headlines — though given deal lead times, Q2 data will be more telling.

The relative value case for European direct lending is real: Q1 unitranche margins of ~550bp offered 2.1% more yield than B3-rated syndicated loans, and 85bp more than US unitranches — at 0.7x lower leverage. Capital raising for drawdown funds is also stabilising (€13bn in Q1, up from €8–9bn in prior quarters).

[CHART: Quarterly European Private Credit Volume (Total & Unitranche), Q1 23–Q1 26 | Source: market data / KBRA]

💭 My View

The European data is genuinely better than the US narrative suggests — lower leverage, better protections, and a yield pickup to both public markets and US peers. But I’d be careful about using Q1 data to declare all-clear: the AI disruption to Software borrowers takes time to show up in origination data given deal lead times. Watch Q2 numbers carefully. And the systemic contagion risk (via insurance assets, bank exposure, institutional LP stress) is not European-specific — it’s a global question. I’m not panicking on European private credit, but I’m not complacent either.

📉 Cross-asset: Low vol equity — the trade that works either way

An interesting equity signal worth tracking for its credit implications: low volatility stocks (Staples, Healthcare, Utilities, Insurance, some Industrials) have underperformed badly in recent months, tracking the move up in bond yields. In Europe, the low vol basket is down 5% from its highs. In the US, the underperformance vs S&P is 5% YTD.

The case for these names now is asymmetric: if yields spike further (US 10y → 5%), the broader market struggles and low vol stocks outperform as a defensive play. If yields stabilise or fall — the base case for most strategists on a 6–12 month horizon — low vol stocks catch a bid. The entry point after 6 months of underperformance is attractive on either scenario.

The credit angle: these sectors (Utilities, Consumer Staples, Industrials) are also among the most resilient in an inflationary environment from a credit perspective — they have the pricing power and EBITDA margin stability that protects credit quality when input costs rise.

[CHART: European Low Vol factor vs Stoxx600, YTD 2026 | Source: market data]

💭 My View — Bottom Line

EUR credit has rallied hard and fast. The technical picture — unwound hedges, strong inflows, yield buyers — explains why spreads are here. But valuations at the 7th percentile since 2015, spread-to-vol ratio at the 2nd percentile, deteriorating book covers, widening NICs, and a geopolitical situation that hasn’t actually resolved... these aren’t noise. They’re signals.

I’m not calling a blowout. The structural support from yield buyers, ample liquidity, and solid earnings remains real. But at those levels of spreads today, the market is pricing near-certainty of a perfect outcome. The bar for disappointment is very low. I’m keeping tactical hedges in place and waiting for either a confirmed deal (add into strength) or a technical reset (add into weakness). The medium-term remains constructive — the path to get there could be bumpy.

Thanks for reading. If this was useful, hit the like button, share with a colleague, and follow me on X @ZeGoodTrader for more on Credit, Equities & Financial Markets 📊

This newsletter is for informational purposes only and does not constitute investment advice. All views are personal and do not represent those of any employer or institution. Past performance is not indicative of future results.

Chart: MTD Performance across asset classes

And the punchline: USD STIRs now fully discount a Fed HIKE at the March FOMC (vs 6bp priced one week ago), 75bp of ECB hikes are priced through year-end, 68bp for the BoE. The “rates vol stays low forever” consensus we entered 2026 with has been quietly dismantled.

My view hasn’t changed: tactically cautious near-term, structurally constructive medium-term. But the near-term risks I flagged are now firing one by one — and the inflation print this week is the most uncomfortable one of all 🔥

Below is my full credit roundup covering: the inflation shock, rates vol as the only signal that matters, gas storage running 4 weeks behind 2022, primary attrition, the barbelling fund flows story, why this still isn’t 2022, and where I’d actually put money to work.

🔥 Inflation is back — and credit hasn’t priced it

The single most important chart of the week is US core PPI ripping to +5.2% y/y, the highest since 2022. Core CPI also surprised to the upside at +2.8% y/y. And on the other side of the world, China — which has been exporting disinflation globally for years — appears to have exited deflation with PPI flipping to +2.8% y/y.

💭 My View: Two big disinflationary forces of the past few years — Chinese overcapacity and well-anchored US inflation expectations — are simultaneously losing potency, RIGHT as an energy shock is working through. Credit hasn’t priced this. EUR IG is sitting at 91bp, basically pre-war levels.

[CHART: US Core CPI & Core PPI, % y/y (long history showing core PPI at highest since 2022) Source: Bloomberg]

[CHART: China CPI & PPI, % y/y (PPI flipping positive after years of deflation) Source: Bloomberg]

📊 Rates vol is the ONLY signal worth watching

I’ve made this point all year and it keeps proving correct. EUR credit spreads are strongly correlated to rates volatility — but barely related at all to the LEVEL of yields. Look at the chart: every spike in normalised realised Bund volatility maps directly onto a spread widening episode, while the level of 5y Bund yields drifts independently 📉

The mechanism: when mark-to-market losses pile up too quickly, momentum-chasing retail investors withdraw from fixed income funds, and institutional investors wait on the sidelines for better entry points. Speed of repricing matters more than the absolute level.

The 2022 analog is the playbook. A 2.75% rise in 10y Bund yields caused a 14.2% total return loss on EUR IG, leading to €16bn of outflows at the worst, and ultimately pushed spreads to a peak of 250bp — even though the Eurozone never actually slipped into recession 🤯

CHART: EUR IG Spreads vs 5y Bund Volatility — clean correlation since 2010
Source: Bloomberg

[CHART: EUR IG Cumulative Total Returns in 2022 (-14.2% peak drawdown) alongside EUR IG Cumulative Fund Flows in 2022 (-€16bn worst)
Source: Bloomberg]

💭 My View: Where are we today? EUR IG YTD total return just turned to ~0% (likely negative after this morning’s move). Historically, the risk of an outflow cycle rises sharply once year-on-year returns turn negative. We’re not there yet — but we’re a single bad inflation print away.

This Substack is reader-supported. To receive new posts and support my work, consider becoming a free or paid subscriber.

📏 European credit is priced for perfection

Let me give you the most uncomfortable stat I’ve seen this week:

• The ratio of EUR IG spreads to rates volatility is at the 2nd percentile of its history.

• The same ratio for Sterling IG is at all-time lows.

Translation: spreads have barely moved while rates volatility has surged. In 2022 this ratio looked completely different — spreads were wide AND vol was high, reflecting proper risk compensation. Today, spreads are priced for perfection while the macro environment is anything but ⚠️

On absolute levels: EUR IG sits at 91bp — only +2bp wider YTD despite Iran entering its 3rd month, an energy crisis, and global inflation re-accelerating. EUR HY at 310bp, +8bp YTD. Both indices trade close to their 1y tights.

CHART: One Year OAS Ranges By Sector Index — most sectors near tights
Source: MNI / Bloomberg

⛽ The Iran clock keeps ticking — and gas storage is the silent risk

Trump this week said the ceasefire is “on massive life support“, with a large gap remaining between the two sides on the Strait of Hormuz status, war reparations, and the nuclear file. Late in the week his frustration visibly increased — and he claimed he “doesn’t need to re-open the Strait of Hormuz” 🤨

We don’t need actual escalation to be worried — we just need more time to pass without a resumption of oil flows. Brent scenario forecasts now point to:

• ~$100/bbl if Hormuz opens by June 1st

• ~$116/bbl if it opens by July 1st

• ~$130/bbl if it opens by August 1st

• ~$140/bbl if not until September 🛢️

And the second-order risk that’s getting underpriced: European gas storage is running at 35.9% full, lower than the same day in 2022. The refilling rate over the past two weeks has been just 0.21%/day — alarmingly slow heading into the season where storage is meant to rebuild for winter.

TTF gas has been well-behaved at €49/MWh so far. But in 2022, natural gas was THE main source of European inflation, with gas averaging €131/MWh over the year and peaking above €300/MWh 😱

CHART: Europe Gas Storage, % full (current vs 2021-2025 historical lines, red dots highlighting 13-May of each year — 2026 visibly below 2022)
Source: Bloomberg

💭 My View: The market is anchored on €49/MWh TTF and assumes refilling will catch up. Both assumptions are looking shaky. If gas storage doesn’t materially rebuild by end-June, we walk into next winter with the lowest buffer since 2022 — and that’s the scenario where headline EU inflation has real upside.

Subscribe now

🏗 Primary markets: deals printing, but attrition is biting

EUR IG primary slowed to €13.1bn this week across 20 lines from 14 tickers. 62% tilt to corporates. EUR HY remained busier with €4.4bn printed across single-B and BB names.

The headline metrics look fine: average cover at 3.5x, shrinkage at 24%, both broadly in line with YTD averages. Aggregate NICs even ticked 1.5bp tighter WoW.

BUT zoom out and the picture is less comforting. NICs in May are averaging +9bp vs +7bp in April and just +1bp YTD. Book coverage has dropped to 3.6x in May vs 4.1x in April. The marginal buyer is clearly getting pickier 📉

Notable prints this week:

• 🔵 BMW — 3y/6y/10y triple-tranche, 5/2/5bp NIC, tightened 2-4bp in secondary

• 🔵 ING — 12NC7 Tier 2 €1bn at +143bp, 5bp NIC

• 🔵 Unicredit — 10NC5 sub €1.25bn, 10bp NIC (one of the wider prints)

• 🟢 Rolls-Royce — first time back since 2020 (when it was HY!) — 5y/10y priced 8bp THROUGH our fair value, monster books

• 🟢 Arla Foods — inaugural deal, books 6-7x covered, +10bp tightening on the break

• 🔴 Arcadis — 5y came 13bp wide of FV, +2bp wider in secondary (most attrition this week)

• 🟡 Amazon printed a massive CHF 2.8bn six-tranche Swiss franc deal — going as long as 25-year tenors. Reverse Yankee theme stays alive 🤖

💭 My View: The deals that price well are crossover credits (Rolls-Royce returning to IG, Arla inaugural). The deals that struggle are the ones being priced too aggressively into a thinning bid. Watch for May rush volumes post-blackout — historically May is the largest issuance month of the year, hitting a market where the secondary bid is already wobbling.

💶 Fund flows: the great barbell

The most interesting flow dynamic of the past 3 months: investors are barbelling. Massive inflows into government bonds and EM debt. Outflows from money market funds. Credit flows lagging meaningfully.

Cumulative % AUM flows over the past 3 months 👇

• 🟢 Government bonds: +3.0%

• 🟢 EM debt: +2.6%

• 🟡 IG: -0.1%

• 🟡 Money market: -0.2%

• 🔴 HY: -2.9%

The driver is clear: as “risk-free” yields rise into the 3%+ range and curves steepen, money market funds lose their appeal and pension/insurance buyers reach into government duration. Credit is caught in the middle.

BUT this week showed signs of a turning tide:

• EUR IG funds: +€711mm (3rd consecutive weekly inflow, mid-term funds leading for 5 weeks)

• EUR HY funds: +€1bn (1% of AUM!) — biggest weekly inflow in several months, with mutual-style funds leading and HY ETFs benefiting from broader risk-on 🚀

  

• Government bond funds: +€2.4bn (3rd straight week of strong inflows)

[CHART: Cumulative weekly flows in % of AUM — Gov’t +3.0% and EM debt +2.6% leading, HY at -2.9%
Source: BofA / EPFR]

💭 My View: The HY inflow this week is a meaningful change of regime AFTER 7 consecutive weeks of outflows. Combined with EUR IG ETFs running their 5th consecutive weekly inflow, retail is starting to chase yield again. Yields at 3.79% on EUR IG and 5.83% on EUR HY are simply too attractive for a structurally yield-starved buyer base to ignore.

⚖️ Is this 2022 again? No — and that matters

Every credit conversation I’ve had this week has started with: “feels like 2022 setting up again”. I disagree, and here’s why the playbook matters:

US 5y5y inflation swap has also remained remarkably contained vs 2022, breaking its usual tight relationship with oil prices — which says the market still has faith in central bank credibility over the medium term ✅

💭 My View: 2026 is NOT 2022. This makes the sell-off more contained — not the valuation more attractive. The downside from here is more limited than 2022, but so is the upside from current spreads. Risk-reward is genuinely unattractive at 91bp IG / 310bp HY for adding here.

🗺 Where I’d actually put money to work

If you HAVE to be invested (and most of you do), here’s where the risk-reward still makes sense:

🟢 BUY:

• EUR single-Bs — only segment still wider than long-term average. Tactical highest-conviction. +1.3% FX-hedged pickup vs US single-Bs. Single-B credits are back in primary (TDC, EnQuest, Lottomatica all priced well) — strong sign of market repair

• Real estate IG & HY — 80%+ of bonds still wider vs pre-conflict. BBB-heavy = double factor of any ECB pivot back to dovish. LEG Immo, Kojamo, Vonovia, TAG Immo all looking constructive on results

• 5-7yr EUR IG belly — 73rd percentile cheapest bucket. Avoided in March par flows (clear underweight). RV opportunity if curve stops flattening

• Telcos — surprise outperformer post-Liberation Day playbook. Cheap enough that defensive label doesn’t suppress the bid

🔴 AVOID / UW:

• Chemicals HY — China dumping is structural, not cyclical. 18% of EUR HY industrial bonds in distressed territory

• Consumer goods IG — 85% of bonds already tighter than pre-war. No margin of safety, real inflation pass-through risk

• Transport HY — oil still $20 above pre-conflict. Not resolved by ceasefire optimism

• 3-5y IG bucket — crowded consensus. Already owned

🟡 HEDGES TO HAVE ON:

• iTraxx Main payer options into June expiry — JPM tracking 75bp payer is a clean way to hedge an inflation-driven outflow scenario

• Underweight GBP IG vs EUR IG — UK political risk premium is back and 10y gilt/bund just crossed +200bp

🎯 The bottom line

Near-term tactically cautious. Medium-term structurally constructive.

At 91bp EUR IG / 310bp EUR HY, markets are pricing a benign outcome with high conviction. The 2nd-percentile spread-to-rates-vol ratio tells you risk compensation is the lowest it’s been in years. The May 15 OECD crude inventory deadline has passed without a Hormuz reopening. Gas storage refilling is running well behind 2022. Global inflation is re-accelerating. The Fed is now priced to HIKE.

Could spreads still grind tighter from here if Hormuz reopens and inflation rolls over? Yes. But the asymmetry is poor.

This is a “have hedges on, sit in defensive carry, and wait for a better entry point” market. The structural backdrop — yield buyers, ample liquidity, solid Q1 earnings, lower (so far) rates volatility realised — supports a constructive medium-term view. But the near-term setup screams “be patient” 🎯

Thanks for reading. If this was useful, hit the like button, share with a colleague, and follow me on X @ZeGoodTrader for more on Credit, Equities & Financial Markets 📊

Two months into a Middle East conflict that has not formally ended, with the Strait of Hormuz still constrained and three regional Fed presidents now openly dissenting against further easing, the S&P 500 is up 8% YTD, the Mag7 just printed 19% revenue growth and 52% earnings growth, and Korean equities are quietly running at +91% YTD — triple what avg commodities have done. 🚀

Chart: A few movers YTD

Markets are not pricing chaos. They are pricing a very specific story — and that story has a name: AI is real, AI is now, and everything else is noise.

Here is what is actually happening underneath, why I think the bullish case has just gotten meaningfully stronger — and the three things that could blow it up. 👇

If you find this kind of analysis useful, the best way to support the work is to subscribe — free or paid — or simply forward this piece to one person who would benefit from it. Independent research only exists because readers like you keep it alive. 🙏

1. The Earnings Print That Changes the Conversation 💥

Forget the macro noise for a second. With 82% of S&P 500 companies now reported, 72% of the Stoxx 600, and Japan ramping up, the Q1 scoreboard is locked in — and the gap between regions is the widest I have seen in years.

Q1 2026 Earnings Scorecard:

Source: blended consensus tracking, market data as of early May 2026

Read that table again. The US is delivering +23% EPS growth on +10% sales growth, with 83% of companies beating estimates and a positive surprise factor of 18 percentage points. Europe? +3% EPS, -1% sales. Japan slots in between with a respectable +17% EPS, but only 24% of companies have reported so far.

This is not a small gap. This is a regime. And it has been getting wider every quarter for two years. 📈

📊 S&P 500 minus Stoxx 600 quarterly EPS growth differential vs. US/Eurozone Composite PMI relative — the earnings gap has been widening for 6 consecutive quarters and now sits at +20pp

The blended Q1 EPS revision is telling 📌

Here is a stat that does not get enough airtime. The S&P 500 blended Q1 EPS estimate, which is what analysts had penciled in for the index aggregate, has moved from $70.1 at the January lows to $74.7 today — a sharp upgrade during the reporting season itself. Combined with the fact that 44% of US companies have raised their full-year guidance — above the 15-year historical average of 42% — and you have a corporate sector that is meaningfully more confident than the market commentary suggests.

📊 S&P 500 blended Q1 2026 EPS — sharp upgrade from $70.1 (Jan low) to $74.7 (May), with the bulk of the move concentrated in the past 4 weeks

The sector contribution math 🧮

Of the 23.3 percentage points of S&P 500 EPS growth this quarter, here is where it came from:

• Communication Services: +7.9pp (Meta, Google — AI ad revenue)

• Financials: +5.3pp (banks benefiting from rate volatility & trading)

• Information Technology: +4.7pp (semis, software, hyperscaler infrastructure)

• Discretionary: +3.6pp (Amazon AWS dominates this number)

• Everything else combined: +2.4pp

• Healthcare: -0.6pp 🔻 (the only negative drag)

That is the math. Three sectors that all touch AI — Comm Services, IT, and Discretionary — delivered 16.2pp of the 23.3pp, or 70% of total earnings growth. Healthcare was the only negative contributor.

And the headline number that matters most is buried in the capex guidance. The four big hyperscalers — Amazon, Google, Microsoft, Meta — collectively guided to $695–725 billion of capex in 2026. 💰 That is not a typo. That is roughly the GDP of Switzerland. Spent. On infrastructure. In a single calendar year. By four companies.

💭 My View — The FOBO Trade Is Now Self-Funding

There is an acronym worth knowing: FOBO — Fear Of Being Obsolete. Every CEO sitting on a board today understands that if they underspend on AI infrastructure relative to peers, they risk being the next Kodak. That is the dynamic actually driving capex, and it is the reason this cycle does not look like 2000.

The dot-com bust happened because companies were spending money they did not have on revenue that did not exist. Today’s hyperscalers are spending operating cash flow they already generated, on demand they can already see in their cloud growth numbers. AWS, Azure and Google Cloud are not guessing — they are rationing capacity.

That does not mean valuations cannot correct. It does mean the structural story is fundamentally different — and that any pullback in Tech should be treated as a buying opportunity rather than the start of a regime change.

2. The European Problem Nobody Wants to Talk About 🇪🇺

If you only read the headlines, Europe is having a fine time. Stoxx 600 EPS up 3%, 59% of companies beating EPS, better than expected, a positive surprise of 2 percentage points. Fine. 🤷

Look one layer deeper and the picture is uglier. Sales growth is -1%. Discretionary EPS is down 20%. Healthcare is printing negative EPS growth. Industrials are -4%. Six out of eleven Stoxx 600 sectors are showing negative or flat sales growth.

This is not an earnings recession. It is something more uncomfortable: an earnings stagnation in the middle of a global capex boom. European companies are not participating in the AI build-out as winners — they are participating as bystanders, and in some cases as casualties. ⚠️

The profit warnings tell the real story 🚨

Since early March, I have counted over 40 profit warnings from listed companies in the US and Europe — and the pattern is striking. Read the commentary attached to each one and you find the same handful of culprits showing up again and again:

• Middle East / Iran conflict 🛢️ — cited as the primary driver in at least 18 warnings (Wizz Air, EasyJet, Booking Holdings, TUI, Pernod Ricard, Lamb Weston, NOV, Technip Energies, O-I Glass, and many more)

• Rising fuel costs ⛽ — Alaska Air, United, Southwest, SkyWest, Norwegian Cruise, Carnival — basically the entire airline and cruise sector

• Consumer weakness 🛒 — Nike (China), Domino’s, Wayfair, Clorox, Crest Nicholson — broad-based softness in discretionary

• Specific shocks 💥 — Future PLC -24% (Google search ecosystem changes hitting ecommerce traffic), Alstom -29% (cash flow guidance withdrawn), Crest Nicholson -34% (UK housing)

The single most-cited cause is the Middle East conflict, which means: if the conflict resolves, a meaningful chunk of these warnings reverse. If it does not, expect more.

📊 Q1 2026 profit warning tracker — share of warnings by stated cause: Middle East conflict, fuel costs, consumer weakness, tariffs/trade, company-specific. Heat map of 1-day relative stock reaction

The Cyclical-Defensive split is the real signal 🔍

Inside the US numbers, there is a fascinating divergence. Cyclicals are delivering +30% EPS growth vs. Defensives at +21%. Cyclical sales growth is +13% vs. +10% for Defensives. In Europe, the pattern flips — Cyclicals -4% EPS, Defensives flat. In Japan, Cyclicals +21% vs. Defensives +19%.

Translation: in the US and Japan, the cycle is working. Companies geared to the real economy are growing faster than the safe-haven plays. In Europe, defensives are outperforming cyclicals only because cyclicals are getting hit harder — not because the economy is strong. It is the wrong kind of defensive outperformance.

3. The Korean Surprise Nobody Is Talking About 🇰🇷

Here is a stat that should reframe how you think about 2026 so far: Korean equities are up 91% year-to-date. That is more than triple the return on commodities. It is the best-performing major equity market on the planet. 🥇

📊 YTD 2026 performance leaderboard: KOSPI, MSCI Korea (in $), MSCI Taiwan, MSCI EM, MSCI Eurozone, S&P 500, Nasdaq, Russell 2000, Brent crude, Gold, Bitcoin

Why is this happening? Because Korea is the cleanest pure-play on the AI infrastructure build-out outside of the US. Memory chips. HBM. Foundry capacity. The Korean exporter model is structurally levered to exactly the capex cycle that the hyperscalers just guided to.

Taiwan is similar. China is benefiting from the same theme via a different channel — domestic AI capex and the fact that whatever happens in US-China trade negotiations, the country is going to need its own AI stack. Trump is in Beijing on May 14-15 to meet Xi. 🤝 Whatever the press conference says, the underlying dynamic is that both sides need a deal — the US needs rare earths, China needs chips. A fragile detente looks like the base case.

EM aggregate consensus 2026 EPS growth is now sitting at +51%, up from +18% expected in January. 🔥 That is not a typo either — it is the biggest upward revision of any major region. The story is not if EM Asia is participating, but how aggressively to lean into it.

This is the part of the trade that European investors keep underweighting. EM Asia with high beta to AI is not a side bet. It might be the cleanest expression of the entire macro narrative.

4. The Central Bank Plot Twist 🏦

Here is the part of the story that should make you nervous if you are sitting in risk assets — and the reason I am still keeping hedges on despite everything I just said.

Two months ago, the consensus was that the Fed would deliver around 60 basis points of cuts by year-end. As of last week, that consensus has been fully unwound. The market is now pricing the Fed on hold for the rest of 2026 — and at the last FOMC meeting, three regional Fed presidents dissented against the statement, signaling they think inflation risk is now greater than labor market risk. 🚨

Meanwhile in Europe, both the ECB and BoE delivered a hawkish hold last week and explicitly signaled they are ready to hike in June if the data warrants it. UK pricing now has 90bp of tightening by year-end. Norway already hiked in May. Australia hiked for a third time. The global central bank scorecard has flipped from cutting to hiking in the space of weeks. 🔄

📊 EUR and GBP monetary policy expectations vs. Brent oil price — 1Yx1Y ESTR and SONIA show R² > 92% with front-month Brent since the Iran conflict began

And here is the kicker — the correlation between oil prices and rate expectations in EUR and GBP is over 92%. This means the entire hawkish repricing is being driven by an energy shock that is, in principle, reversible. If Hormuz fully reopens and oil drops back to $70, the rate hike pricing will fade. If it does not, central banks may actually have to deliver.

💭 My View — This Is What Asymmetry Looks Like

I have seen this movie before. The 2022 playbook would tell you that when central banks pivot hawkish on an energy shock, you sell credit, sell duration, sell anything risk-sensitive. But 2026 is not 2022.

We are starting from rate levels that already reflect normal monetary policy, not zero. The required adjustment is smaller. Long-term inflation expectations — the US 5y5y inflation swap — have barely moved despite oil’s volatility. The market is telling you it believes the shock is temporary.

The right trade here is not to fight the AI rally. It is to hold the AI exposure and add tactical hedges through rates volatility — because if I am wrong about temporary, the transmission mechanism will be a violent flattening of the curve, and that is the trade that protects you on the way down.

5. The Three Things That Worry Me 😰

a) The May 15 deadline nobody is pricing 📅

OECD crude inventories are projected to hit operational minimums around May 15. Even if Hormuz reopened tomorrow, oil takes 20-45 days to ship from the Persian Gulf to Europe. The inventory stress is already locked in. IEA emergency reserves cover roughly 33 days of full-disruption losses, and after that, prices move non-linearly.

Markets are pricing a base case where partial reopening happens in the next few weeks. Fine. But the asymmetry is brutal — if reopening does not happen, the move higher in oil will not be 10% or 15%. It will be a step-change, and central banks will have no choice but to respond. ⚡

b) The concentration problem 🎯

S&P 500 concentration, measured by the Gini index, is close to historical highs. The volatility dispersion between the top-75 weighted stocks and the index itself is at multi-year highs. This is the legacy of AI dominance — it is real, it is fundamental, but it also means that any reversal in the AI narrative would mechanically amplify into a broader index move.

The same concentration that delivered the 23% earnings growth this quarter is the concentration that creates fragility on the way down. Meta’s 8.6% post-earnings drop despite reporting 33% revenue growth is the canary worth watching. 🐤 Markets are no longer rewarding good results — they are punishing anything short of perfect. The median 1-day relative performance of S&P 500 EPS misses this quarter was the worst in two years (-3.2%), while beats only delivered +1.6%. That asymmetry is a positioning warning.

📊 US stock price reaction to Q1 quarterly EPS beats vs misses — median 1-day relative performance shows asymmetry: misses punished -3.2% on average, beats only rewarded +1.6%, the worst beat/miss spread since 1Q 2022

c) Seasonality is no longer your friend 🌡️

May through August is historically the weakest four-month window for risk assets. May is usually the heaviest single month of EUR IG corporate supply — the rush before the summer slowdown. June through August sees institutional clients de-risk before vacation. August has the worst credit market liquidity of the year. September brings the largest single-month supply wall.

Yes, 2025 made the seasonality argument look stupid. But 2025 was an easing cycle, not a hiking conversation. With central banks now having opened the right tail of policy outcomes, seasonality has its teeth back.

6. How to Position 🎯

Putting it all together, the framework I am working with looks like this:

• Stay long the AI winners. ✅ US Tech, Communication Services, and EM Asia with high AI beta — Korea, Taiwan, China. The fundamental story is intact and Q1 earnings just made it stronger.

• Underweight European cyclicals. ❌ The earnings data is unambiguous — European companies are losing the AI capex cycle as participants and getting hit hardest by the energy shock. Discretionary -20%, Industrials -4%, sales -1%. Wait for cheaper entry points or rotate into US/EM equivalents.

• Add Healthcare as the recession hedge. 🩺 It is the only sector screening cheap on inflation factor models, has record-low positioning, and benefits from a midterm gridlock outcome. Low beta to Fed expectations means it is insulated in a stagflation scenario.

• Hedge with rates volatility, not credit shorts. 🛡️ If I am wrong about the temporary nature of the energy shock, the transmission mechanism is a curve-flattening shock. Conditional bear flatteners and limited duration exposure are the right structure.

• Tactical caution near term, structural bull medium term. ⏳ The May 15 deadline, supply pickup, seasonality, and weakening fund flow data are real near-term risks. But ample liquidity, fair valuations, solid earnings, and the AI capex cycle support a constructive view once the geopolitical dust settles.

📊 EUR IG and US IG spreads vs. rates volatility — credit spreads are tight relative to vol (EUR at 2nd percentile, Sterling at all-time lows), suggesting the right hedge is rates vol, not credit beta

The Bottom Line 🎬

The bull case here is not complacent. It is not pretending the conflict is over. It is not assuming central banks will save the day. It is built on something more concrete — the largest corporate capex cycle in history, funded by operating cash flow, deployed by companies that have already pre-sold the capacity.

That story is not invalidated by an Iran deadline, by a Fed dissent, or by seasonality. It is just made bumpier. And bumpy is exactly when the people with the longest time horizons get to compound at the best entry points. 💎

If your portfolio looks like the world Europe is reporting from — slow growth, defensive tilt, sales contraction — you are not wrong about the macro. You are just on the wrong side of the regional split. Q1 earnings just gave you a clean dataset on where the growth actually is.

Thanks for reading. 🙏

If this was useful, the best thing you can do is forward it to one person who needs to hear it. If you want my weekly take on credit, equities, and the cross-asset stories that move the tape, subscribe below — and follow me on X for the daily commentary that does not make it into the long-form pieces.

As always, the views above are my own. They are not investment advice. They are one person’s read of the tape, written for thoughtful investors who want to think for themselves.

The response was strong. But the most useful pushback came from people actually inside these structures: “you’re conflating the retail vehicle problem with the underlying credit problem.”

They have a point. And because being wrong in public is more useful than being right in private, this week I engage seriously with the bull case 🐂️ — armed with the most granular data currently available on the asset class.

25 charts, all built from primary data. Let’s go. 👇

1. Retail Is Only 23% of the Market

The redemption gate narrative is real. But it’s concentrated in one specific vehicle type — non-traded BDCs and interval funds. And those vehicles are a surprisingly small slice of the total.

The other 77% sits in institutional drawdown funds and listed BDCs — structures with no quarterly redemption window and no gating mechanism. LPs committed for 8–10 years. They can’t run. And they haven’t.

Chart 1: Private credit market composition — retail vehicles ($400bn) vs institutional ($1,353bn). Source: JPM, BDC Collateral, Preqin.

💭 My take: The gate crisis is real, but it’s a retail vehicle crisis playing out within a much larger asset class that is not gating. Conflating the two is analytically sloppy. That said, the retail segment is what drives the headlines — and headlines drive sentiment. The 77% institutional base is stable. The 23% retail base is not. Both are simultaneously true.

2. The PIK Discrepancy Everyone Needs to Understand

I’ve been citing bad PIK at 6.4% of all private loans (Lincoln International). I stand by that. But there’s a second dataset that reads differently — and understanding the divergence matters.

Looking at the 30 largest BDCs, total PIK on the balance sheet actually fell in Q4-25 — from $50.3bn to $46.9bn (16.17% to 15.36% of the debt portfolio). Bad PIK — the specific measure of PIK added mid-loan to relieve cash flow stress — was flat at 1.99%.

Why the gap with Lincoln’s 6.4%? Two reasons: (1) Lincoln covers a broader universe including smaller, weaker lenders. (2) The Q4 BDC decline was driven almost entirely by Medallia going cash pay under Blackstone — but Medallia is marked in the high 60s despite being technically cash pay. It’s heading to non-accrual.

Chart 2: Total PIK (%) and Bad PIK (%) on BDC balance sheets Q4-24 to Q4-25. Bad PIK flat at 1.99%; weighted average mark on PIK: 97.80%. Source: JPM, BDC Collateral, Company Reports.

💭 My take: Bad PIK at the top-30-BDC level is 1.99%; the Lincoln 6.4% reflects the long tail of smaller, less seasoned lenders. Both are real for different parts of the market. The Medallia dynamic is the tell — cosmetically positive, fundamentally negative. Watch the next two quarters: PIK should rise as activity slows and short-term rates stay elevated. The weighted average mark on PIK (97.80%) will decline as software loan valuations fall in Q1-26.

3. Software: The Bull Case Is Specific, Not General

The core data point the debate keeps missing:

Chart 3: Software market weight — HY 4.7%, Leveraged Loans 13.8%, Private Credit 20.8%. The asymmetry that explains why private credit has a different conversation than HY. Source: JPM, S&P/IHS Markit, KBRA DLD.

Read more

→ EUR IG at 91bp — 1bp from the “All Clear” scenario target of 90bp. Market pricing near-certainty of full resolution.

→ May 15: OECD crude hits operational minimums. Even with Hormuz fully open today, tankers won’t arrive until mid-May. The inventory crisis is already locked in.

→ Primary firing: 4.5x cover, 5bp NICs, €19bn IG + €6.3bn HY last week — but YTD supply on pace for a record with demand at its weakest in three years.

→ EUR HY distress is a chemicals/China-dumping problem, not a software story. The market has it backwards.

→ Best remaining value: EUR real estate (80%+ bonds still wider than pre-war), EUR single-Bs vs. US (+1.3% FX-hedged), belly of the IG curve (5–7yr cheapest bucket).

📊 The ceasefire rally in numbers

The US-Iran ceasefire triggered one of the most violent credit squeezes in recent memory. EUR IG rallied 9bp from the conflict peak, EUR HY compressed 23bp. The curve bull-steepened — 2s10s back to 1.54x. Average daily iTraxx volume collapsed from €49bn at the panic peak to €27bn. The market has essentially priced the conflict as resolved.

The bull-steepen is real but incomplete. 2s10s at 1.54x remains below January’s 1.75x peak and year-end’s 1.64x. The reason: markets are still pricing more than two ECB hikes this year. Unlike every other major central bank where rate expectations eased on the ceasefire, ECB hike pricing has not budged — keeping the front end under pressure. The curve cannot fully steepen until that changes.

EUR IG credit curve: 2s10s spread ratio (Jan–Apr 2026)

👉 Three scenarios — and where markets are pricing today

The war-gaming framework sets three explicit spread targets based on geopolitical resolution. At 91bp IG and 329bp HY, the market is pricing Scenario 1 with near-certainty — despite 48 days of conflict with no signed deal and the naval blockade still formally in place.

IG at the 7th percentile since 2015. The spread-to-rates-vol ratio at the 2nd percentile. The asymmetry from here: a perfect deal tightens spreads maybe 5bp. Diplomatic limbo adds 70bp. That is not a spread trade — it is an options trade. The iTraxx Main 75bp June payer screens as the rational hedge at these levels.

My view: “91bp IG is Scenario 1 territory. But with IG at the 7th percentile since 2015 and the spread-to-rates-vol ratio at the 2nd percentile, I’d be putting on iTraxx Main 75bp June payer protection here. The asymmetry is brutal: perfect outcome = -5bp. Diplomatic limbo = +70bp.”

The deadline nobody’s watching: May 15

Markets priced Friday as “conflict over.” EUR IG closed at 91bp. Brent crude, however, is still $20 above pre-conflict levels. One of these is pricing hope. The other is pricing physics.

OECD crude inventories hit operational minimums on May 15. Oil takes 20–45 days to ship from the Persian Gulf. Even if Hormuz fully reopens today, tankers departing now won’t arrive in Europe before mid-May. The inventory stress is already locked in regardless of what happens diplomatically in the next 48 hours.

The asymmetric trade: “Perfect outcome: EUR IG tightens to 86bp (-5bp). Diplomatic limbo past May 15: 130bp+. Five basis points of upside, forty of downside. The June iTraxx Main 75bp payer is not paranoia — it is arithmetic.”

🚩 Primary markets: the illusion of strength

The primary numbers look excellent in isolation. NICs compressed from 10–20bp to 5bp over four consecutive weeks. Book coverage reached 4.5x — up for the third straight week. Last week: €12.1bn across 18 lines, +3bp secondary performance post-price. Notable deals: Goldman Sachs (€1.4bn + £500mm), SoftBank (€1.8bn HY triple-tranche), Banca Transilvania (€1bn), Santander (€1bn). YTD 2026 EUR IG issuance is on pace for a record €243bn.

Read below the surface and the picture is less comfortable. Current EUR IG fund inflows are at their weakest since 2023 — last week’s net flows were just +€1.0bn, a fraction of the multi-billion weekly streaks seen in 2024. Record supply plus weakest demand in three years equals 4.5x coverage only because NIC compression is how issuers buy their books.

The moment the blockade extends rather than ends, this window can snap shut — exactly as primary conditions collapsed within weeks of Russia’s February 2022 invasion after appearing “fine” in January. Issuers: price now. Investors: get selective.

🔍 Where value remains after the rally

Share

Highest conviction — EUR Real Estate: 80%+ of real estate bonds (IG and HY combined) are still wider than end-February levels — the most widespread residual cheapness in the EUR credit universe. The ECB has turned more dovish since early April, structurally reducing the refinancing cliff risk overhanging the sector. Combined with the BBB/A cheapening (real estate is heavily BBB-weighted), this is the double-factor value case.

🔍 The EUR IG curve: where to be long and short

The ceasefire rally has reshuffled relative value across the EUR IG curve. The par flow data from March reveals exactly where institutional investors were hiding — and where they weren’t.

🧪 EUR HY: chemicals, not software

The dominant US credit narrative — AI overhang, software LBO distress, tech valuations — is largely irrelevant for European HY. European leveraged finance has a fundamentally different sector composition, and the distress data makes this unambiguous.

Subscribe now

“18% of EUR HY industrial bonds distressed. China dumping — structural, not cyclical. EUR HY bond distress (7.1%) now below loan distress (7.4%) — first time since 2000.”

The root cause: China dumping — Chinese manufacturers flooding EU markets with cheap product at below-cost prices, destroying European chemical producers’ pricing power. This is a trade and geopolitical issue, not a cyclical earnings deterioration. Until the EU-China goods trade deficit narrows or anti-dumping enforcement accelerates, chemicals and industrials deserve a structural underweight.

🌏 Cross-Atlantic value: EUR single-Bs

EUR single-Bs remain the only segment of the entire EUR credit complex still materially wider than its long-run average. The FX-hedged yield pickup vs. US single-Bs of 1.3% is the largest since March 2023. HY/IG spread ratio at the 85th percentile and Bs/BBs at the 99th percentile confirm the best beta pair in EUR credit is long HY vs. IG.

Labour-lite: the 2022 playbook is repeating

During the 2022 inflation shock, wages rose 6% trough-to-peak in the Euro Area following energy inflation — hitting corporate cost bases at the worst possible moment. Labour-intensive sectors underperformed. Labour-lite sectors (industrials, healthcare) outperformed.

The structural headwind this cycle: the EU-US labour productivity gap is near decade-wide levels. US productivity has risen significantly since 2022. European productivity has barely moved — reflecting slower AI adoption in an SME-dominated economy. If energy inflation persists and wages follow with their usual lag, European firms face the same cost squeeze without the US digital productivity offset.

Overweight (labour-lite, headcount declining): industrials, healthcare, media, real estate.

Underweight (high headcount, no pricing power): leisure, transport.

💰 Winners and losers in an inflationary environment

In an inflationary environment, one metric separates the winners from the losers in credit: EBITDA margin stability. Firms that can pass through input cost increases via higher prices — the pricing power champions — will protect their credit quality. Firms that can’t will see margins erode, leverage rise, and spreads widen. The screening of the entire IG universe for EBITDA margin volatility from 2000 to Q4 2025 is unambiguous.

The 2022 data is instructive: capital goods EBITDA margins were remarkably stable throughout the entire energy and inflation shock. Transportation margins oscillated dramatically as fuel costs surged and volumes contracted simultaneously. This is not just a backward-looking observation. PMI data is already confirming that input costs are rising faster than output prices in several European sectors — firms with low margin variability are the safest credit longs in this environment.

Key trade expression: Long capital goods IG vs. short transport HY captures the full pricing power divergence. Utilities and consumer goods (branded) also screen well as defensive longs. Technology credit is more volatile than consensus assumes and warrants selective positioning rather than broad overweight.

🚀 Five reasons for markets to keep on performing

The risks are real and documented above. But credit markets are not priced for collapse — they are priced for perfection. There is a difference, and the bull case deserves equal airtime as I do believe that one should add when we’ll get a bit of weakness.

The balanced view: The May 15 inventory deadline is a genuine near-term risk that justifies tactical hedges (iTraxx Main 75bp June payer). But the medium-term structural backdrop — ample liquidity, fair valuations, solid earnings, potential ECB pivot — supports a constructive credit stance once the geopolitical dust settles. The mistake would be to chase the hedge after the event rather than maintaining it now at 91bp.

Share

🗓️ What to watch going forward

Weekly European credit intelligence by @ZeGoodBanker · April 20, 2026

For informational purposes only. Does not constitute investment advice. Views are personal and do not represent those of any employer or affiliated entity.

PIK loans. Wrong labels. Rating arbitrage. Opaque valuations. Record dividend recaps. I’ve covered all of it. Each time, the response from the industry was some version of: “nothing to see here, the underlying credit quality is fine.”

This week, the Q1-26 filing data landed. And the numbers are extraordinary.

This post covers everything: what just happened, why the maturity wall is the real story, the hidden mechanism most people are still missing, the red flag scorecard, whether this is actually 2008, and what it all means for Credit investors.

Let’s go. 👇

📰 What Just Happened — The Q1 Filing Data

The 5% quarterly redemption cap is the tripwire the entire private credit industry watches. Cross it, and a fund is forced to gate withdrawals and lock investors in. It’s the line between “semi-liquid” and “trapped.”

Here’s where things stand after Q1-26:

🔴 Blue Owl — investors asked to pull 40.7% of shares from its technology fund (OTIC). And 21.9% from its flagship OCIC. Both gated at 5%. Blue Owl scrambled to announce $1.3bn in refinancings to demonstrate liquidity.

🔴 Barings ($4.9bn fund) — investors requested 11.3% back. The fund paid out less than half — 44.3% of requests honored, holding back ~$180m of investor capital.

🔴 Goldman Sachs BDC ($15.7bn) — requests came in at exactly 4.999%. One decimal place below the tripwire. The fund sent a letter to shareholders saying it was “the only non-traded BDC in the peer group whose repurchase requests came in below the standard 5% cap.” The fact that they felt the need to send that letter tells you everything.

🔴 Blackstone BCRED — raised its quarterly redemption cap to 7.9% to meet elevated demand. Shares fell to a two-year low on earnings day, despite management defending portfolio quality.

🔴 Moody’s revised its BDC sector outlook to Negative (April 7) — the first change after holding Stable for over two years. AI disruption was explicitly named as a “disruptive force” putting the sector “on defense.”

And then the development that matters most, buried in the noise:

🔴 JPMorgan began independently marking down the collateral value of software loans used as security in back-leverage facilities for private credit managers. This is not a realised loss. But it reduces borrowing capacity precisely when managers face redemptions — and can trigger margin calls. Banks are starting to reprice what private credit managers have been marking at cost.

Dimon said it months ago: “when you see one cockroach, there are probably more.” 🪳

💭 My take: The redemption numbers are the symptom. The JPMorgan collateral markdown is the early signal of something more structural — the lending infrastructure that quietly supported private credit leverage is being tightened. Watch this space.

⚡ Three Forces Converging

Before we get into the mechanics, here’s why everything is happening at once:

1. The SaaSpocalypse. Software loans represent ~20% of BDC portfolio exposure (Jefferies). But the real number is almost certainly higher — Bloomberg and research found 250+ investments worth $9bn+ that weren’t labelled as software loans even though the borrowers describe themselves as software companies. And software now accounts for ~50% of all private equity deals over the past decade. The asset class bet enormous on one sector — and AI is now threatening to upend that sector’s entire business model.

2. Iran and risk-off. The geopolitical shock provided the trigger for anxiety that was already structural. Investors who were already nervous about valuations and returns used the macro deterioration as a reason to join the exit queue.

3. Returns have collapsed since 2022. US PE funds annualised ~5.8% between 2022 and Q3-25. The S&P 500 returned ~11.6% over the same period. The performance premium that justified the illiquidity premium has been disappearing — and the sector that drove it was software. The 2017-vintage tech fund IRRs are now barely above non-tech peers. The trade that defined a decade is unwinding.

The result: the sector is estimated to shed $45–70bn in assets over the next two years — reversing growth from $34bn to $222bn since 2021.

⏰ The Real Story: The 2026–2028 Maturity Wall

Everyone is focused on the redemption pressure. That’s the news. But the more important structural risk is the software debt maturity wall — and it’s closer than most people realise.

The numbers:

• More than $330bn of HY, leveraged loan and BDC-linked software and technology debt is coming due through 2028 (Bloomberg).

• $140bn of technology debt matures in 2028 alone. Companies will start trying to refinance from H2 this year.

• The average price of 2021-vintage/2028-maturity loans is already $83.40 — significant stress signalling before a single maturity has been tested.

• A third of these loans have credit dates still in 2021. The issuer has not demonstrated capital market access in years. They’ve never refinanced in a higher-rate, AI-disrupted world.

• $31bn in BDC software sector debt is coming due this year — 2026. Not 2028. Now.

And here’s the leading indicator the fixed income market is already pricing:

In the leveraged loan market, the premium traditionally seen in technology loans has completely broken down this year. Falling lev loan prices for software are a leading indicator for private credit — where marks can considerably lag public markets. Typically, when loan prices fall, a rising number of private credit borrowers subsequently face stress in the form of not earning enough to cover their interest costs (Kroll StepStone data).

Some private credit funds are already turning away software borrowers outright. A number of software company sales planned by PE have stalled. The exits are freezing before the maturities even arrive.

The debate is live on how bad this gets. Marathon Asset Management’s Bruce Richards estimates up to 15% of software direct lending could default in coming years. Goldman GSAM’s Vivek Bantwal pushes back: most lending is at the top of the capital structure and relatively insulated from restructurings. Both are probably right for different tranches of the market.

💭 My take: The 2028 wall is framed as a future problem. But refinancing runs on a 12–18 month pre-maturity cycle. That means the real stress tests start in H2 2026 — right now. The lev loan market is already pricing the dislocation. Private credit marks are lagging, as they always do. The gap between public and private pricing of these assets will close. The only question is whether it closes via orderly refinancing or via forced sales at distressed prices.

Subscribe now

🔍 The Hidden Mechanism: Adverse Selection

Here’s what nobody is writing about clearly enough.

When a fund faces redemption pressure, it doesn’t sell assets randomly. It sells what has a bid.

Cash goes first. Then the liquid, clean, easy-to-price names. Then the stronger credits that a sophisticated buyer will underwrite quickly, near par.

What’s left in the fund after that process?

• The 2021/2022 vintage heavily-leveraged paper (underwritten at 7–9x when rates were zero)

• PIK loans — where borrowers are paying interest by issuing more debt rather than actual cash

• Software loans with AI disruption risk hanging over them

• Opaque assets that secondary market buyers passed on

Goldman’s $1.4bn Blue Owl portfolio sale was celebrated as proof the system works. Look at who bought it: CalPERS, OMERS, BCI, and Blue Owl’s own Kuvare insurance affiliate. Four highly sophisticated institutions evaluating a pre-selected slice in a negotiated process. That’s the best possible execution on the most sellable assets in the book. For anything beyond that curated pool — buyers will not pay par.

The brutal irony: the patient investor who stayed, trusted the stable NAV, and didn’t panic is now holding what’s left after everyone else walked out with the good stuff. The quarterly report won’t tell them that. The NAV will look roughly the same.

Until it doesn’t.

💭 My take: This is why “NAV looks stable” is almost meaningless as a signal right now. It reflects the composition of what remains, not the quality of the original portfolio. The instrument panel can look fine long after the underlying has deteriorated. It usually does in private credit — until the forcing event.

🚩 The Red Flag Scorecard

I’ve been building this list for a year. Here’s where it stands:

🔴 PIK surge — “Bad PIK” (not in the original loan structure, added to relieve cash flow pressure) hit 6.4% of all private loans in Q4-25, up from 2.5% at end-2021 (Lincoln International). More than half of all PIK loans now show this dynamic. And it’s getting worse: companies with bad PIK have seen loan-to-value ratios surge from ~35% at close in 2021 to ~65% today. They’re more indebted on assets that are worth less.

🔴 Wrong labels — Bloomberg found 250+ investments worth $9bn+ not labelled as software loans, even though the borrowers describe themselves as software companies. Investors are flying blind on sector concentration.

🔴 Rating shopping — Issuers increasingly routing deals to smaller, less established rating agencies to obtain more favourable ratings on identical assets. The incentive is obvious: lower required capital against the same risk. This is not an edge case — it’s becoming structurally embedded in how the industry operates.

🔴 Industry concentration — Half the BDC industry is run by just 4 firms (Blackstone, Blue Owl, Ares, BlackRock). BCRED has never held so many PIK loans. Weighted average spreads in the two largest private credit BDCs reached series lows at end-Q3. When something goes wrong in the largest managers, it goes wrong at scale.

🔴 Structural mismatch — Non-traded BDCs were designed to solve a distribution problem for asset managers, not a liquidity problem for investors. Quarterly redemption windows against 5–7 year loan maturities was never going to hold under sustained stress. We watched exactly this play out in non-traded REITs in 2022. Same structure. Same redemption gates. Same outcome.

🔴 Banks repricing the collateral — This week’s JPMorgan collateral markdown is new and important. It’s not a realised loss. But it reduces borrowing capacity for managers under redemption pressure and can force additional collateral — tightening the spiral exactly when it’s already tightening.

⚖️ Is This 2008? The Honest Assessment

Let me try to be genuinely balanced here, because the “this is subprime 2.0” narrative and the “nothing to see here” narrative are both wrong.

The bears are right that:

• PIK is surging, adverse selection is hollowing out fund quality, and NAVs are fiction until forced to clear in the open market

• $160bn+ in private credit software loans mature by 2028–2029 and need to refinance in an AI-disrupted world. The average 2021-vintage/2028-maturity loan is already at $83.40

• $31bn in BDC software debt matures this year — the first real test is now

• Some BDCs already saw NAV drops of -15% to -20% in H2-25 (Oxford Square Capital, BlackRock TCP, OFS Capital) — showing how fast marks can move when a book is forced into the open

• JPMorgan’s collateral markdown is the first sign that the infrastructure supporting private credit leverage is quietly tightening

The bulls have a point too:

• Private credit is far less leveraged than banks were in 2008 (1–2x vs 15x for banks)

• Only ~15% of direct lending AUM sits in the semi-liquid retail vehicles subject to these gates

• Most large-cap PE-backed software lending is at the top of the capital structure — relatively insulated from restructurings (Goldman GSAM)

• Uncalled LP capital ($310bn) and bank exposures ($410–540bn) are manageable in isolation relative to the broader system

• The current stress is primarily a retail asset-liability mismatch plus AI uncertainty — not a broad macro credit crisis (JPMorgan, Cliffwater, Blackstone’s Jon Gray)

Share

The real wildcard isn’t the funds. It’s the investors in those funds. Insurers, pension funds, and retail investors who borrowed to make their private credit allocations. If they face liquidity stress — not the funds, the investors — the forced selling cascade hits assets well beyond private credit. That’s the channel that’s harder to model and harder to track.

And it’s the one nobody has a clean answer for.

💭 My take: This isn’t 2008. The leverage ratios are different, the systemic interconnection is different, and the size relative to the overall financial system is different. But “not 2008” is a low bar. The adverse selection mechanism is real, the maturity wall is real, and the JPMorgan collateral move suggests the stress is migrating from the funds to the broader credit infrastructure. I’d rather be early than late in acknowledging that.

💡 The Contrarian Signal: Wall Street Is Pouncing

Here’s the thing: while retail investors are queuing to exit, the largest institutions in the world are doing the opposite.

• Morgan Stanley is launching North Haven Strategic Credit Fund — a new interval fund with 5% quarterly redemptions, designed specifically to deploy capital at improved terms from the current dislocation.

• JPMorgan has filed with the SEC for a new interval fund with 7.5% quarterly redemptions — 50% above the standard 5% cap. They’re betting they can build a more liquid version of the same product.

• Goldman Sachs and JPMorgan are offering hedge fund clients tools to short the private credit market — a market that barely had hedging instruments a year ago.

The read: the smart money sees an opportunity. But they’re not buying the existing vehicles — they’re building new, better-structured ones. And the fact that GS and JPMorgan are simultaneously creating short products tells you they’re not convinced the bottom is in.

Consolidation is coming. Strong, seasoned platforms will gain share. “Tourist” managers — funds that piled into private credit during the 2021 boom without the infrastructure or credit expertise — will be shaken out. The question for investors is which side of that divide their manager sits on.

🌍 EUR Credit: Insulated, But Not Immune

Here’s the important question for European investors: how much does this matter for EUR credit?

The direct exposure is genuinely limited:

• Software & IT services = only 4% of the Euro iBoxx HY index — versus ~20% of US BDC exposure and ~50% of PE deals

• BDCs represent just 5% of European private credit (versus 1/3 in the US)

• European banks’ direct exposure to US private credit is small and largely absorbable — KBC, ING, UniCredit all below €50m

But the transmission channel isn’t capital. It’s sentiment.

In the social media era, a crisis of confidence moves faster than the underlying losses justify. Barclays has already provisioned £110m on Tricolor and holds £500m exposure to the collapsed UK mortgage lender MFS. If JPMorgan’s collateral markdown approach is replicated by European banks, that creates a sentiment shock for EUR financials that has nothing to do with the direct numbers.

The yield curve angle matters too. If this contributes to CB repricing (as I discussed in my rates vol note a few weeks ago), the flattening trade would be credit-negative via the structural buyer base dynamic — exactly the 2022 playbook.

The private credit stress is primarily a US problem. But sentiment travels. Position accordingly.

The Bottom Line

Private credit isn’t broken. But the non-traded retail vehicle built on top of it is. And the adverse selection mechanism means the investors still inside those vehicles are holding something that looks exactly like what they bought — right up until the quarter it doesn’t.

The $31bn BDC software maturity wall hits this year. The $140bn wall hits in 2028. The refinancing cycle for the 2028 maturities starts in H2 2026. The leveraged loan market is already pricing the stress. Private credit marks are lagging, as they always do.

The cockroach count is rising.

Thanks for reading. If you found this useful — share it with someone who should be thinking about this.

👉 Follow me here & Twitter @ZeGoodTrader for daily takes on Credit, Rates and Financial Markets

ZeGoodTrader is an independent credit market commentary. Views are personal and do not represent those of any employer.

Three weeks into the Iran conflict and credit is still absorbing the shock better than almost any other asset class. EUR IG spreads ended around 102bp — only 20% above YTD tights — even as oil went haywire, rate hike expectations exploded across every major central bank, EUR HY funds bled their largest outflows of the year, and the supply window essentially slammed shut. The resilience is real. But so is the asymmetry: limited upside from here, several tail risks un-priced, and the market structure has changed in ways that deserve more attention than they’re getting. This week’s post is a full war-game through the fault lines.

1. 🛢️ The Energy Shock: What’s Priced and What Isn’t

The first and most important thing to understand about this week: Brent crude is NOT reflecting how tight the oil market actually is.

Commodity strategists estimate shut-ins could shortly hit 12 million barrels per day as storage capacity dwindles. The current Brent price still reflects comfortable inventories built up heading into the year and long shipping lag times from the Gulf. But if the Strait stays effectively closed, the inventory draw-down makes a non-linear reprice inevitable.

And then came the escalation nobody had in their base case: Israel hit Iran’s South Pars gasfield. Iran retaliated by striking Qatar’s Ras Laffan LNG plant — 17% of capacity damaged, with a 3–5 year repair timeline. That’s 20% of global LNG supply at risk. Even when the war ends, long-run gas prices stay elevated. The energy shock now has a permanent component.

The Oil “VIX” hit 120 this week — a 5.5 standard deviation move, far above its 2022 peak. Brent moved $36 peak-to-trough in a single day. Oil ETF volumes surged to a record $16.3bn. Yet EUR IG? Just +4bp wider. That disconnect is telling — but it has limits.

The IEA arithmetic: Emergency reserves sit at ~400 million barrels, release rate ~1.2mbd. Potential supply losses in a prolonged Hormuz closure: 12mbd. At that release rate, reserves cover roughly 33 days of losses. After day 33, upstream shut-ins begin and prices move non-linearly. That is the real tail risk — and it’s not in the price.

Oil is now the de facto volatility signal for the entire risk spectrum. Intra-day correlations to crude have surged since the conflict started: EM equities show the highest correlations, the S&P and CDX IG right behind, with the Stoxx 600 relatively insulated. Credit has effectively become an oil derivative.

France vs Italy — the energy trade nobody is talking about: France’s nuclear-heavy energy mix makes French IG names significantly less exposed to the gas price shock than Italian peers. In a prolonged disruption scenario, the sovereign/quasi-sovereign and utility spread differential between OAT and BTP-linked credits should widen from here. This is a structural relative-value opportunity that is still underappreciated in positioning.

My view: Don’t anchor to Brent’s current level as the signal. Watch the IEA day-count, the Hormuz re-opening timeline, and the gas price — not oil alone. The permanent LNG supply reduction from the Ras Laffan damage is a game-changer that the market is still processing.

2. 🇮🇷 Iran: “Mission Accomplished” is Getting Harder to Believe

When the conflict started, my base case was a rapid de-escalation: historical precedent, military imbalance, OPEC+ flexibility, and the Trump political calculus all favoured a fast fade of the risk premium.

We are now in week three. And conviction in that scenario is eroding.

The ground-truth scorecard:

• Strait of Hormuz remains essentially closed

• Iran continues targeting ships AND oil infrastructure across the region

• US deploying additional forces to the region, including 82nd Airborne

• Trump rhetoric has shifted from “very complete” early in the week to decidedly hawkish by Thursday

• “Mission Accomplished” from the US does NOT guarantee free passage through Hormuz