The industry has become very efficient at creating tokens, but creating something durable and worth holding for more than a year is much more difficult.

I find it useful to think about durable token demand along two dimensions: the strength of the belief supporting the asset and the credibility of the economic value accruing to its holders.

Belief assets

These are assets with no cashflows which derive value from a common belief that they are worth something. They have value insofar as other people believe they have value. As a non-believer in Bitcoin, I will only buy BTC if I think enough people believe in it.

But what differentiates fleeting memes like WIF or BONK from a meme with permanence like BTC?

There are recurring features of durable belief assets1:

1. A figurehead. A cult leader. BTC had Satoshi. ETH had Vitalik. SOL had Anatoly arguing with ETH maxis on twitter every day. HYPE has Jeff. Leaders of “belief assets” become larger than life. When the token is mentioned, this person’s name needs to pop up in your mind instantly.

2. A near-death moment and resurrection. BTC had Mt Gox. ETH had The DAO. SOL was intertwined with FTX, yet all of them came back from the dead. HYPE had the JellyJelly attack which dumped it to single digits as people questioned if it would cease to exist. Emerging from a near-death encounter sifts out all but the most devoted believers. This forms the base of your most hardcore believers.

3. Make your early believers rich. VERY rich. Number go up is the best BD and the people you make rich become your most ardent believers. BTC gave its early adopters a 200x in 2013-18. ETH did a 40x in 2020-22. SOL pulled a 100x too. HYPE did 30x with size and gave a fat airdrop to early believers.

4. A cult community with memes and a unique identifier. Even with monetary incentives, you still need memes to spread quickly. You need inside jokes so people feel like they belong to an exclusive club. You need people supporting the “greater cause” blindly because they feel like they belong. BTC had laser eyes, ETH had “.eth”, SOL had the “◎” / “.sol”. HYPE had the annoying “.hl” guys.

5. A one-liner rallying cry. You need a simple one-liner to spread your religion. BTC was “digital gold”. ETH was the “world computer”. SOL was the “ETH killer” or “decentralized NASDAQ”. HYPE is the “house of all finance”.

In recent times, ZEC comes close with (2) from the shielded pool vulnerability, (3) as it ran from $40 to $1200 in a year, and (5) as “Private bitcoin” or “quantum-proof bitcoin (soon)”.

These characteristics cannot be manufactured. You can’t just call your community a religion (super cringe). Within this category, I like PRL aka “AI bitcoin” as an early bet that it can develop the characteristics above.

There is a more straightforward way to climb up this mountain of durable tokens…

Cashflow assets

Cashflow assets are rare because (i) it’s much easier to market your token/narrative for exit liquidity than build a business with positive FCF, and (ii) if you actually have a cash-cow business that prints money and don’t need to raise money for growth, you save a lot of trouble by NOT launching a token and just pocketing the profits.

Teams that are printing money, like DexScreener, Axiom, Trojan, are happy directing 100% of profits to themselves. It was this way until Hyperliquid came along and showed that you can create a business with strong cashflows, where accruing value to the token could give you a better outcome than pocketing short-term profits.

The key questions here are:

1. Economics: Is there durable economic surplus (profits)? HYPE and PUMP are the best examples here as they make $1-2m/day throughout the year.

   Perp DEX revenues tend to be more sustainable and less cyclical especially after the introduction of equity perps. The durability of launchpad revenues tends to be questioned. The market has been right about PONS, STONK, and GP (BonkFun) revenues not being durable, but PUMP has been proving it wrong.

2. Transparency: Can that surplus be verified onchain? The most pristine revenues are fully on-chain, like HYPE and PUMP, as they require minimal trust. On the other hand, you have projects like GRASS where revenues are fully off-chain (private contracts with frontier labs), where the market questions if the revenues exist, though I’m sure they do.

3. Token vs Equity: How does that surplus benefit the token? Is there an equity entity siphoning value? This has been a subject of debate recently as Venice raised money on its equity while giving out free tokens. GRASS also had to clarify that all value goes to tokenholders.

   The best counter-example is SHFL which is a great business that makes good revenues, but value accrues to the equity entity in Curacao, which received $18m in dividends, while tokenholders get an entry into a lottery2.

4. Value capture: Does value accrue directly to the token through buybacks or dividends? Self-explanatory, but automatic buybacks and dividends (HYPE, LIT, SKY) are much better than discretionary buybacks (PUMP).

5. Trust: If distributions are deferred due to investing in growth, can the team be trusted? This is most similar to investing in startups. Tokens like GRASS, ENA, CHIP and CARDS sit here. They are explicit that value accrues to their token only, but they are investing in growth. It involves a lot of trust in the team as there aren’t legal protections for tokenholders and often not even a board or financial statements. Hence, something like the Blockworks’ TTF is helpful.

   Demanding an early-stage project distribute all its surplus can undermine its ability to grow. But tokenholders have good reasons to prefer it. If they cannot establish how retained earnings will eventually benefit them, “investing in growth” is asking them to underwrite both the business and the team’s future willingness to share its success. Buybacks resolve part of that uncertainty today, and I commend CHIP’s recent improvement in clarity around token value accrual.

The golden examples here are HYPE, LIT, PUMP, and SKY. All four satisfy (1) through (5).

It is rare to find new tokens that tick all five boxes. PAPER was one of the best examples in recent history, as it could plausibly generate substantial revenue3. Unfortunately they scrapped the token because it created an incentive to exploit the exchange.

I think this category is ripe for innovation. HYPE pioneered the idea of a highly value-accretive token. PAPER gave folks a taste of what it could have been. Imagine launching a high-revenue business like Polymarket (makes ~$20m a month) and accruing all the value to a token4.

In fact, the Shuffle team have been talking about launching a fully on-chain casino. If such a casino existed and 90% of net gaming revenues go back to buying the token every day, that’d put SHFL all the way at the top of the list. Buying tokens near the bottom of the list is just a bet that the team will move upwards.

Looking ahead, I am eyeing the launch of Perp DEX tokens like VAR (Variational) and XTEN (Extended). Other projects worth noting include exchanges like ARCUS, ENTROPY, and POLYMARKET (if they ever TGE). If you are keen to build here, hit my DMs.

Conclusion

There are only two ways to become a durable, multi-cycle asset. BTC, ZEC, and HYPE achieved that by climbing on the “belief side”, while HYPE, LIT, and PUMP are climbing via the right side.

Across all assets, HYPE is the most comfortable hold, as it both is really strong as a cashflow asset with the potential of becoming a durable belief asset.

Many other tokens try and climb the mountain but fail, including new L1 tokens (game is overplayed now) and boomer DeFi tokens with little revenue and little to no value accrual. Narrative trades like AI try again and again to climb up the left side of the mountain, but gravity prevails.

Subscribe now

Marked safe from AI slop:

I get asked a lot about whether you should trade your personal money or join a hedge fund.

The decision is usually based on 6 factors below.

1) Cashflow

Analysts are typically paid $130-200k outside of the US. In the US, it probably gets as high as $250k base. This typically isn’t as high as big tech or investment banking, in order to incentivise analysts to make money by trading. Bonuses can be >$1m.

If you’re running $1m of personal money, you need to make 13-20% per year just to break-even on your monthly salary (which carries zero risk). The math only starts working out if you’re trading a >$5m personal book, which requires 3-4% pa just to earn an analyst’s base salary.

On top of that, not having a monthly salary to cover living costs will impact your psychology especially on losing months.

2) Capacity of your trading strategy

You join a hedge fund if your cut of the potential profits is more than what you’d make on your own book, and your strategy has large capacity.

If your strategy is running solana microcaps or small prediction market arbs, you’re out of luck. A pod shop with a $100m book isn’t going to be interested in putting $10k in an illiquid microcap, because even a 20x is only 0.2% of the book.

As an example, if your strategy has $20m capacity (e.g., an OTC SOL tranche from the FTX estate at $70), you expect a 2x return, and you have a 10% profit share arrangement from the hedge fund, you will make $2m. Whereas even if you (irresponsibly) put your entire $1m portfolio into the deal, you will only make $1m.

3) Trading is a lonely sport

Solo trading can be pretty depressing because you can’t discuss your edge with other people. Also, your trader friends will rightfully be slightly cagey about their strategies to prevent edge decay.

This makes it slightly more difficult to be around smart people.

4) Your edge is in infra, scale, or access

There are certain edges which require a lot of maintenance on infrastructure (usually systematic trading) or scale (special exchange relationships & rebates / deals due to the volume your market making algos do).

In terms of access, funds also have access to some institutional instruments that solo traders might not have (e.g., complex options products via MMs, forwards).

On the flipside, if your edge is onchain or moving fast, joining a hedge fund that cannot transact onchain due to compliance may render your “edge” useless.

5) Learning the ropes before starting your own fund

There is actually a lot to learn at an institutional fund around legal, compliance, trading operations, operational security, tax, fund structure, etc. which are transferrable when starting your own fund.

I did not expect to learn so much tacitly, from joining a large institutional fund.

6) Utilising fund connections and reputation

Projects don’t want to talk to a solo trader especially if you’re just starting out (don’t have a brand name yet) and/or if your book is small (<$1m). If you’re starting out you can leverage your fund’s name to get to rooms you want to be in.

Conclusion

With all that said, the perfect arrangement would be joining a hedge fund to learn the ropes, while running your small capacity strategies on the side in a way that is also beneficial for the fund (e.g., monitoring smallcaps, which gives you a better pulse on the market).

Do note that most firms have strict trading policies which either (i) require you to hold your trades for >3 weeks (usually more TradFi funds), or (ii) highly discourage you to trade your personal portfolio to focus your energy on making gains for the fund.

I think to build a long-term career in trading, you need to sit in a room with the top 0.01% of traders in the world. The quickest way to get there is to join a top fund.

The downside is that for most junior positions, you would get a “discretionary bonus” instead of a fixed % cut of your profit (usually a Portfolio Manager gets this). Hence it is important for you to join a firm where you trust the PM will pay you out fairly.

If you’re recruiting for or starting a prop fund or hedge fund, do reach out!

This goal of this guide is to get things done with minimal hassle.

Step 1: Buy a robot.

The cheapest you can get is the SO-101 ($300 here). It is a pair of robot arms: a leader arm which you control, and a follower arm which copies your movement.

Each arm is made of 3D-printed parts and 6 servo motors. It’s all open-source so you can 3D print and assemble it yourself, but paying $300 saves you time.

Each motor acts as a joint, so this is a 5 DOF arm (5 arm joints + 1 gripper ) which mimics a human arm. The joints are the shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper.

Step 2: Teleoperation to collect data

First you need a setup: Two light sources, a webcam+tripod, a cube and a box (the task). Put it all on a table that’s at least 120cm x 60cm (the 140x60 Lagkapten from Ikea works well).

Create a HuggingFace account and get your user access token. Everything here onwards runs on the HuggingFace open-source robotics library LeRobot (@LeRobotHF).

Clone LeRobot repo and install via the command line interface (CLI). Everything that follows are just lerobot commands on CLI.

CLI may be intimidating for non-technical folks, but you can just “claude --dangerously-skip-permissions” and chat with Claude. It troubleshoots very well.

Connect both robot arms + wrist cam + scene webcam to your laptop (you will likely need a USB hub). Run “lerobot-calibrate” to calibrate, then hit the “lerobot-teleoperate” command.

Now, you can start recording teleoperation data for your task (”pick up the cube and place it in the box”). Complete 50 episodes, vary the position of the cube and box on the table for each episode, and grab the cube from slightly different angles each time so the model can learn.

If you make mistakes such as failing to grip the cube or moving the box with the gripper, re-record the episode. It should take 30-40mins in total.

Make sure to keep the lighting and camera angle consistent (do not bump the camera!).

After completing 50 episodes, LeRobot automatically uploads your dataset to HuggingFace.

^ bare minimum teleop setup in my bedroom (before i got my lighting set up)

Step 3: Train or finetune your model

Rent a GPU on Runpod.io to either train the ACT model (Action Chunking with Transformers, released 2023) from scratch, or finetune the frontier model π0.5 by @physical_int (model open-sourced Sep 2025, note the latest π0.6 and π0.7 are not open-source).

ACT is usually the first model practiced on by beginners. The model is small, trains on a single GPU, and has no pretrained robotics knowledge to lean on. It learns to imitate teleoperation demos through plain supervised regression.

It introduced two new ideas back then: 1) instead of predicting one action at a time, it predicts a chunk of the next ~50 to 100. 2) it encodes multimodal human demos as a “latent style variable” rather than averaging them into mush.

π0.5 is a vision-language-action (VLA) foundation model. It is pretrained on heterogeneous data, robot demonstrations across multiple embodiments, web image-text, object detections, and high-level semantic subtask labels, co-trained so the policy learns not just how to move but what scene it is looking at.

Architecturally it is a VLM backbone driving an action expert that emits continuous motor commands. What we are doing here is finetuning it on additional 50 demos, so the general manipulation prior gets specialised into a single SO-101 cube-in-box policy.

For ACT you will need an RTX4090 (~$0.70/hr) and training takes ~2hrs for ~15-20k steps, costing ~$1.40 in total.

For π0.5, you will need an A100 80GB (~$1.40/hr). They can take ~8hours to complete ~15-20k steps, so total cost is ~$10.

Once your training or finetuning is complete, your model will be uploaded to HuggingFace automatically.

Pro tip: Just run Claude to help you install all the dependencies on the Runpod instance.

Step 4: Time to run inference!

This is the fun part. Time to see if your training / finetuning works!

For ACT, you can run inference on your laptop via CLI. If all goes well, success rate for in-distribution data (same object color, position, target, lighting as training data) should be 70-80%. Out-of-distribution data varies, but can be as low as 10-30%.

I was speedrunning this and used whatever I had at home -- a 2x2 rubik’s cube and an old reflective metallic box. The tests were fairly out-of-distribution given the colours/orientation of the cube, giving a success rate of ~20% (4 out of 20 tries).

This is somewhat expected for a small 50-episode dataset, and the multi-colored cube likely let ACT overfit to colour.

As for π0.5 you will need to rent an RTX 4090 on Runpod to run inference in the cloud (note how inference is far lighter than training, where we used an A100). The success rate is much higher, mine worked ~40% of the time (8 out of 20 tries).

^ fully autonomous robot using π0.5 finetuned on 50 teleoperated episodes

Learnings & Next steps

Overall I learned a lot about the practical difficulties and limitations of deploying robots in the real world: hardware limitations (max reach, slippery / transparency, contact-based tasks), data collection ops (manhour scales linearly, teleop vs egocentric vs UMI data), and software (long context window limitations, tradeoff between latency and edge compute).

Now that I have the basics, there are lots of things to experiment with:

• More complicated tasks: Conditional pick-and-place, stacking blocks, pouring beads from one cup to another, contact-rich tasks such as driving a screw, installing a drive belt, wiping a surface clean, peeling, plugging cables

• Try different robots (embodiments): Bi-manual, more DOF per arm (typically 7 DOF), longer reach, heavier payload. (Next cheapest one is YAM at >$3000)

• Try other models: VLA-JEPA apparently works well after

  finetuning with only ~13 episodes

All of this has also given me a profound appreciation of an infant’s ability to learn to manipulate objects through visual cues, proprioception, and force feedback. The human brain is truly amazing.

One of the uncles referenced a friend who is now 65 years old and has never worked a day in her life. She inherited a sum of money that wasn’t huge by global standards, but sufficient to live a decent life in Penang. Was that a life to be envious or be proud of?

That led to the question of: Should you leave an inheritance to your children? If so, how much, and when do you give them?

So firstly, what is the goal of giving your child an inheritance? The uncles agreed that this is to allow your child to “be happy” because they have a “meaningful life”.

So the question is then: how do you give a sum of money to your child to maximise the odds they live a meaningful life, so they will be happy?

What is the meaning of life?

Does the universe assign a purpose to your existence? Almost certainly not2. Biology gave humans a drive to survive and reproduce, and that drive feels like purpose. But it is a cause of why we exist, not something we exist for.

If meaning is not given, it has to be made. Many schools of thought3 exist, but I like the “eudaimonic” camp where the meaning of life comes from exercising your capacities towards something you find worthwhile.

This is in contrast to hedonic happiness which comes from pure pleasure, as hedonic adaptation makes experiences less pleasurable over time. The first time watching the Grand Prix live is much more pleasurable than the third time.

Hence the goal is to give them enough that they feel they can exercise their capacity towards something worthwhile, not so much that they can do nothing (and succumb to the hedonic treadmill)4.

What this means practically is to fund optionality so they can do whatever they find worthwhile. You can fund further education, a house, a venture, a floor under failure, rather than an income that makes work optional.

It is a tight balance between either giving them so much money that it creates a 22 year old with no reason to develop skills, or forcing them to waste their twenties on wage labor they do not need, purely so they “build character”.

There is obviously no single right answer as every child’s personality is different. Give a driven kid $10m and he builds an empire; give a lazy kid $10m and he parties every day for the rest of his life.

The amount you give, and when, matters a lot too.

I think there are three ways which might work.

Firstly, capital investment for specific purposes. This could be for building a business, starting a new venture, property, education. You would be funding something your heir finds worthwhile building, not just his lifestyle.

Secondly, age-gated tranches. This could look something like $200k at 25, $500k at 30, $1m at 35. You assume their identity and competence are formed through the years, and you are funding a reasonable increase in expenses such as children/mortgage as they grow older, so they have optionality to do something worthwhile which might not be rewarded by capitalism (e.g., musician, artist, wildlife preservation).

Thirdly, a match-earned income. An heir who earns $100k gets a $100k supplement, so they “deserve” what they get. If you think capitalism works and hence “meaningful work” tends to generate a bigger financial outcome, you should amplify it.

“Something you find worthwhile” is subjective. What your child thinks is “worthwhile” could be anything from creating music, running a pottery studio, saving wildlife, or advancing AI.

And you are choosing how much to give before you know who they become. You are making the decision blind on both counts: you cannot define the target of “worthwhile” work, and you cannot predict whether they will hit it.

The mechanisms of passing down wealth do not solve this. What they do is buy time.

A lump sum at 22 settles gives you the answer early, whether for good or for bad. Age-gated and conditional money refuses to settle on an answer early. It keeps the outcome resting on them, year by year, for as long as possible, with hopes that they grow into a someone who lives a meaningful life.

That is the most a parent can do: hand over the money in a shape that leaves the answer to the child rather than deciding it for them.

The rest is the child’s to answer, and you may or may not be alive to see which answer they give.

The crux of the issue is that crypto struggles with three key things: (i) lack of innovation, nothing new has come out in the past 2-3 years, (ii) advancements in Quantum tech which threatens the very existence of Bitcoin by 2029, (iii) LLMs like Claude Mythos increasing the number of attacks, rendering the r/r of DeFi unattractive.

That begs the question, what is there left to do in crypto?

VCs and Liquid Token Investors

Due to this innovation drought, the VC industry has been pretty quiet, especially on token deals. Every crypto VC will tell you how boring it’s been, unless they’ve been doing the larger Series B+ rounds or funding stablecoin payment startups.

The handful of good VCs I’m speaking to are investing in contrarian verticals, for example startups quantum (e.g., Project Eleven, Oratomic) or novel ideas (e.g., Shift Foundation, PostFiat, Ambient).

I think this is natural, there is fewer new exciting things as we’ve largely figured out what works and what infra is needed. Now we’re in the adoption phase for payments and remittances. The end-game and institutions are already here.

Similarly for people working in crypto, it only makes sense to be at Stablecoin Fintech firms (e.g., Circle, OpenFX, Tempo, Arc, Plasma), Trading platforms (e.g., Polymarket, Kalshi, Hyperliquid) or novel contrarian startups (as above). Working at an L1 Foundation is a great dead-end afk job with good pay but won’t do you any favours in the long run.

This lack of “cool new things” on the VC side means fewer good tokens will be launched into the market, and less VC capital to flow into liquid markets.

Hence, longer time horizon liquid token investors who assess tokens based on growth, fundamentals, and value accrual, now have maybe <10 quality names, and this seems unlikely to increase anytime soon.

There are only 12 token projects who make >$50m revenues annually. Only three of them have a value accrual scored >= 7 (HYPE, PUMP, JUP). Even if you assess based on “growth” and on teams to improve token value accrual, you could maybe add 5-10 more tokens to the list below (e.g., MORPHO, SYRUP).

OTHERS has also dropped from ~$450B to ~$180B, while stocks are going on a speculative mania across high-bandwidth memory, photonics, quantum, peptides, etc.

Discretionary and systematic traders

You might say that crypto is largely a narrative trade and momentum trade, so it’s meant for discretionary traders who do long/short, trade narratives or catalysts/news. This is where I sit and am most familiar with.

In the depths of the bear, it’s been much more lucrative to trade catalysts, though many require you to be awake and react quickly. There exists a lot of edge here and decent opportunities to trade.

Nevertheless, the past 3 months still had pretty eventful trades like…

• Shorting TAO on Templar subnet leaving, dumped $330 to $260 in 5hrs (Apr 9)… longing TAO when Chamath and Jason Cal were shilling (varying levels of success)

• no trade here but interestingly, WLD didn’t even pump on the Tinder and Zoom partnership announcement (dumped 8% across two days)

• Shorting AAVE on the KelpDAO hack which happened at ~1730 UTC but was only first reported on twitter one hour later at ~1830 UTC (Apr 19)

• Shorting AAVE on Marc Zeller’s ACI leaving (Mar 3)

• Shorting the TRUMP retrace after the Trump Dinner announcement (Mar 12)

• Shorting ACX to the “conversion price” after the spike due to token-to-equity conversion announcement (Mar 11)

• Shorting DRIFT on the exploit, it was down 40% in 1hr (Apr 1)

• Shorting RESOLV on the exploit, it was down ~10% (Mar 22)

• Longing ALGO on Google’s quantum breakthrough (Mar 31)

• Long LDO on the buyback announcement, up ~17% in ~5d (Mar 27)

• Long/shorting pump-and-dump scams like RAVE, SIREN, STO, PIPPIN, POWER etc. (something I prefer not to participate in)

Though with that said, OI is down ~60% since 10/10 and reactions to news tend to be small. You need to be selective of what news you trade and whether other people care about it, as there is little retail trading interest (usually people who respond slowest to headlines) so you are largely pvp-ing other news traders.

I am increasingly seeing discretionary traders spending more time in prediction markets and trading equities / commodities, which Hyperliquid has made a lot easier to pivot.

As for systematic traders and basis traders. As volumes and funding rates die down, traditional strategies becomes less and less lucrative. To keep themselves entertained, they would be doing HIP-3 markets, arbing prediction markets, trading Pendle PTs / Boros, or arbing new perp DEXs (limited liquidity).

Yield farmers

The increasing number of hacks in DeFi has also made yield farmers (or TVL deals on the institutional side) go dormant or fully exit this year as the last couple of good deals were Plasma and USDai. FlyingTulip which likely could’ve worked in a bull cycle, launched without much fanfare.

Usually the reason why a “yield deal” makes sense is because you could sell the governance token rewarded to you (as I’ve highlighted here), but this assumes there’s someone buying it from you and if the liquid investors / discretionary traders are not around to add fuel for the fire, it is difficult to make the r/r worth it.

This has been the case in the past (OHM), present (XPL, ENA) and likely needed in the future (USDai’s CHIP). The hurdle for onchain defi yields used to be 15-25% APR to account for T-bill rate at ~0% and a ~10-15% chance of a hack.

The reason why OHM was worth it back then wasn’t because the risks were lower, but because the rewards were much higher.

Now the hurdle is probably closer to 50-60% APR given the increasing number of hacks ($795m hacked in DeFi in 2026 four months YTD) plus Claude Mythos potentially causing more hacks and quantum risks mounting. With no one buying emitted tokens, it is hard to achieve a justifiable r/r.

Most rational farmers have almost fully gone offchain as even yields from STRC at 11.5% of tradfi Fixed Coupon Notes provide much better risk-adjusted yield (15-20%).

see: Rami poker, CBB, Sisyphus, delucinator, misaka

Onchain trencher

Last but not least… the infamous onchain trencher: the guys who buy at $1m MC and sell at $100m. I think these guys will still be around, given the trenches are still the only place where you can pull a 100x.

It also feels like onchain trenchers are the WW2 soldiers stuck in the cave, not knowing the trenching era is largely over.

At this point needless to say that memecoins have peaked when our dear POTUS and the First Lady launched their coins. We’re towards the end of the Euthanasia rollercoaster where things don’t pump as much anymore and there are just too many value extractors across the board (“FNF groups”, “LA vape cabal”, serial rug factories, and extractive trading fees).

I still don’t think this segment just fully dies though, the small sliver of hope keeps people in the trenches. Over the past few months, we’ve seen…

• $GAS: GasTown went from $100k MC to $60m in 3d, then back to $1m in 3d, now $50k MC. (15 Jan)

• $RALPH: RalphWiggum went from $500k to $55m in 2wk, then to $3m (-93%) in 12h when the dev disowned it, now trades at $50k MC. (21 Jan)

• $PENGUIN: Nietzchean Penguin went to $170m in three days (24 Jan), but now trades at $3m.

• $MOLT hit $120m in one day (31 Jan), but now trades at $1M

• $WHITEWHALE hit $200M (10 Jan), but now trades at $7M

• $ASTEROID hit $200m yesterday (19 Apr) after Elon tweeted that it could be SpaceX’s mascot. It will probably trend to zero after a few more days.

Although these show that from time to time, things do pump, the odds of you hitting them is probably <10%, and the absolute maximum potential is likely a net 10x ($10m → $100m) as things don’t pump much past $100m anymore. And now you have only a few hours to sell it after the peak as things drop >90% in a few hours.

So… what am I doing?

I’m maintaining my arbitrage strategies on Polymarket which generates ~15% APR with capacity of up to ~$250k max. It only makes ~$3.5k/month fully deployed. Arbitrage opportunities have also decreased since Polymarket introduced trading fees, and with the recent npm package poisoning, I’m feeling increasingly that the r/r isn’t there anymore (esp. post-airdrop). Current plan is to kill it post-airdrop.

I’m continuing to trade crypto but not as active as I used to be. I had the idea to “start an AI hedge fund” but my exploration has brought me to the conclusion that AI is not creative enough to do discretionary trading yet (e.g., for idea generation), but it is great if you give it a task or methodology which you have previously used and requires a bit of logical reasoning.

Hence, I’m spending time automating certain “hedge fund analyst” tasks, for example automating the onchain snooping I used to do for insider wallets on Polymarket, which is something that is fairly process-based but requires a little bit of logical thinking. A task perfect for Claude.

I’m also increasingly looking into the finetuning side of AI models, particularly from the angle of crypto and finance data. I’ve been reading outside of crypto which I’ve found incredibly intellectually stimulating on topics such as the AI stack (societal implications and stocks across the stack), Physical AI (world action models, visual language models, and the data problem), and the idea of an “AI rollup” (PE rollup but with AI), but have not found a problem interesting enough to dedicate my life to solving yet.

If you have any interesting things you’d like to discuss, feel free to reach out!

Subscribe now

The concept is simple. On aGDP.io, there is a weekly prize pool of $125-150k which comes from a 0.3% tax on trading all Agent tokens, and a 20% tax on all payments made to Agents for services.

Last week’s prize pool was $137,340!

This prize pool is allocated to Agents that top the leaderboard, ranked by their “Agent Score”. Half of your prize money is given to the Service-Selling Agent as USDC, and the other half of the USDC prize is used to buy the Service-Selling Agent’s token and distributed to the Buying Agents.

Looking at the way the prize pool is allocated, it doesn’t pay to win the #1 spot by too much. The goal should be to spread your resources (total $ budget) to get as many agents as possible into the top 5 or 10, who get a disproportionately large reward.

Hence the way to maximize profits is to maximize your “Agent Score” across ~5-10 agents such that the sum of your % rewards is the highest.

The best way to get your Agent Score up is to have each buyer spend a lot (Buyer Spend), spread revenue across many buyers evenly (Buyer Diversity), and each buyer spending evenly across many sellers (Spend Diversity).

For this article I will assume you have 5 Seller Agents (this is the optimal number of agents N>1 to maximize the per-agent score) and you are spending the same amount on each agent.

With 5 sellers, your Spend Diversity = 0.39 assuming equal spending across buyers. The higher the # buyer agents, the higher the Buyer Diversity, but you hit diminishing returns fairly quickly. At N=50, you’re at 91% of theoretical maximum, and at N=100 you’re at 96%.

Hence it makes sense to just do with 100 Buyer Agents.

Now that we have worked out that the optimal number is 5 Seller Agents and 100 Buyer Agents, we just need to write 100 bots using Virtual’s openclaw-acp to start trading with each other. You can run them locally. This is trivial with Claude Code.

Note that the “Agents” have nothing to do with “AI”. Your agents can and probably should just reply “gm” once they receive payment.

The only thing left to do is allocate a budget, say $10k. Use disperse.app to send 100 USDC to each of your 100 Buyer Agents. Set each Seller Agent service cost to 20 USDC. Write a script such that each Buyer Agents spends their 100 USDC budget, on each of the 5 Buyer Agents.

After this, each Agent will have $2k in “revenue”, totaling $10k revenue across 5 agents. Your cost of farming is 20% * $10k = $2k which Virtuals take as fees.

Go back to the leaderboard to see where you rank. And see if you can earn >$2k back in rewards, so that it’s profitable.

THAT’S IT.

In the last Epoch, there were two teams that took all of the top 11 spots.

Capminal (#1), Mechamind (#5), RoboSphere (#6), Captain Dackie (#8), Synapse Robotics (#9) and AutoForge (#11) are the same person. They had $212k in revenues, which means they spent $42.5k, and had $60k in rewards. Making +40% in one week, or +$17.5k/week → potential for $70k a month!

Hyperbet (#3), ERC-8183 Inspector (#2), Validator (#4), Oracle (#7), Escrow (#10) are the same person. They had $195k in revenues, which means they spent $39k, and had $57k in rewards. Making +46% in one week, or +$18k/week as well!

One interesting observation is that these guys are optimising for having multiple agents on the top of the board, hence why they have 5-6 Seller Agents.

The reason I know they are all the same person/team is that all the Buyer Agents transact with the same 5 or 6 agents. They all also have 100-200 unique buyers every epoch (also the optimal number).

If you’re going into this, it is important to note that this is more like a poker game rather than a risk-free farm.

It is a poker game in the sense that:

1. Depending on other farmer’s scores, you can decide to sacrifice your 4 Seller Agents, and push 1 Seller Agent to the top by concentrating all your money to the 1st agent as revenues.

2. You can boost your “revenues” at the final minute to move to the top of the list, not letting other people have a chance to react. Note that this is actually what happens and the Total Agent Revenue doubles in the last few hours of the Epoch. Your Agent servicing speed is largely limited by how quickly you can ping ACP, so I’d recommend not starting at the final hour.

3. 2/3 of the prize pool comes from the aforementioned 20% fee on revenues, so the more people farming this, the bigger the prize pool, which in turn attracts more people to farm. There will only be ten Top 10 folks obviously, so it is kind of a zero-sum game.

4. You could spend $30k in costs and not rank in the Top 10, thus having small rewards and ending up with a net loss instead.

5. Alpha here is that the aGDP.io dashboard is does not refresh instantly. So in the final hours, if you don’t track other agent’s revenues, it’s hard to know where you rank and thus how much more money you should sink into it to get rewards. Code something up to track other people’s revenues, calculate their agent scores and thus get their rank.

6. Virtuals team can tweak the Agent formula, or make it harder to farm it. They’ve been great at shaping incentives, so I wouldn’t be surprised if they tweak it.

The obvious question is, if this is so profitable, why don’t I do it. Why am I sharing this lol.

The answer is that I did it. I did ~$5k revenues ie ~$1k cost to test if my hypothesis works. It did. I had five agents ranked top ~50 with only 2-3 hours of vibecoding effort.

But this is a temporary zero sum game, and I want to focus my efforts on something more scalable and long term. Hence I'm sharing this here so at least someone else can compete and try and make some money!

Anyway, good luck & have fun!

Subscribe now

The human skill tree is being replaced

Think of a human as an RPG character with a skill tree: physical strength, fine motor movement, movement to transport things, language, reasoning / judgment, creativity, social / emotional intelligence, calculating, pattern recognition.

Everytime we invented something, we only replaced one skill:

• Cars & Planes → replaced human movement

• Typewriter → fine motor movement (writing)

• Calculator → human brain for calculation

• Agriculture machines → physical strength applied in a specific manner

And each time, the displaced workers could fall back on the remaining skills. When steam engines replaced muscle power, you still needed human judgment to run the factory. When spreadsheets replaced bookkeepers, you still needed human reasoning to interpret the numbers. There was always an escape route into the un-automated parts of the skill tree.

But now, Claude is on track to replace all soft skills within the skillset, including the most important ones — reasoning, judgment and social intelligence.

And once we achieve Physical AI which similar to LLMs can generalize across tasks, we replace all physical skills within the human skill tree too.

The optimist says “modern telephones replaced switchboard operators; agriculture machinery replaced the farming workforce; the invention of cars replaced horse-related jobs”. Humans will always invent new jobs to create value and earn a living.

But I’d argue this time it’s different. When cars were invented, horses became completely useless outside of occasional entertainment — horseriding, equestrian, zoos — and they just spend time lazing around.

Humans are the new horses.

The Post-Claude Era

I think 2026 is a crucial year which starts the “Post-Claude era1” which is an inflection point in humanity’s productivity.

Every day I wake up and there are just endless things to build. The last time I felt this energised was when I tried using DeFi and stablecoins in 2021.

Matt Shumer’s viral X article covers it quite well. Something big is happening. But only to those paying close attention, just like during COVID times.

I would stress that we are very early. It’s only materially impacted software engineers so far. Other professionals in Law, Finance & Audit are still living in the pre-Claude Code era.

Within SWE, CC is already in charge of 4% of all Github commits (100% of mine). At the current trajectory, Claude Code will be >20% of all daily commits by end of year.

And this is just the beginning. Claude Cowork hasn’t even gained as much adoption as Claude Code, and yet it could be even more disruptive as Claude Cowork is simply Claude Code packaged for jobs which do not interface with code (i.e., everything except SWE / quants).

In a few months, we will also get the “Claude Cowork” moment, just like how we had this “Claude Code” moment. And it will be even more impactful.

What gets replaced by Claude?

In my view, Claude almost fully replaces jobs that are largely rules-based, logical and repetitive (AI agents have a slight flexibility to reason around the rules) and jobs that can be done to a “good enough standard”.

These are jobs2 like:

• Customer service agents (Sierra.ai, Decagon)

• Insurance claim agents

• Financial auditors (rip PwC, EY, KPMG, Deloitte)

• Legal paperwork (great X article)

• Admin assistants (a nicely packaged clawdbot)

• Back office automation (compliance, expenses, regulatory reporting, KYC/KYB onboarding, AML screening)

• Coding (translating specs to code) and financial analysts (translating a company’s financial statements to a model)

Anthropic own data makes it pretty clear that all white collar categories (i.e., non-physical) work will soon be replaced by Claude.

Hence I expect unemployment to slowly increase, starting from fewer fresh grad jobs (we’re already seeing this in UK and Singapore). The hypothetical Citrini’s 2028 Global Intelligence Crisis scenario is less likely on that timeline, but plausible if you extend it to 2030+.

I don’t see it happening by 2028 due to multiple bottlenecks in getting sufficient power, chips, and data centres to increase compute capacity. There simply isn’t enough compute to fill in the blue gaps in the chart yet.

Supply-demand Seesaw

This is basically a seesaw between supply and demand.

Sustainable adoption will come from agents replacing large expensive teams such that cost savings > cost of deploying agents (e.g., customer service agents) and where incremental profits > cost of deploying agents (e.g., hedge fund analysts).

We’re already seeing this as per Chamath’s tweet:

Since November 2025, our AI costs have more than tripled and we are now spending many millions per year trending to $10M+ per year.

That, in and of itself, feels very scary to me running a small startup.

Mostly because I do not yet see an equivalent uptick in productivity or revenue... while their revenues may be doubling and tripling every month, ours are not so this is starting to eat into margins.

Companies are spending aggressively on AI but the ROI hasn’t caught up yet for everyone.

On the supply side, Claude is currently heavily subsidising subscriptions — a $100/month Claude subscription gives you >$2000/month of API-equivalent usage.

What’s going to happen is that as demand increases, then Claude sub price increases until only those getting their money’s worth will subscribe. Then we’ll build out power, data centres and GPUs to produce more compute so the prices come back down.

How to position yourself

I think there are only four areas where it makes sense to have a “job” in:

1. Physical jobs. This one is obvious but very unsexy. Going back to the skill tree, all the physical jobs are yet to be replaced… until we solve Physical AI and generalised humanoid robots

2. Join AI. Harder, but join a top-tier AI Research lab, create data for the latest models, or help implement AI improvements in companies and make them leaner.

3. Jobs that require relationships and secrets. Partners in law, consulting, business. Trading where secrets are what drives alpha. By definition if an LLM knows it, then everyone knows it, then there is no edge. The moats here are trust, access, and information asymmetry. These are things that aren’t in a model’s weights.

4. Entrepreneurship. The cost of building has dropped 90%, hence you now only compete on other factors i.e., distribution and taste (knowing what to build). I expect the % of population being entrepreneurs to increase greatly. Companies need to hire less, which is great for entrepreneurs.

And above all, you need to OWN AI.

Own the AI Stack

There will be only a handful of companies who can control and own the AI stack. This small class of AI owners are more powerful than world leaders.

As we have seen, the Department of War went all out in pressuring Anthropic on safeguards. Why would they do that if they could just swap Anthropic for OpenAI or Gemini? Anthropic is clearly crucial for national security and to win wars.

If we can’t join the AI overlords directly, perhaps we could hold equity in the AI stack (models, compute, energy, and everything in between) so we can benefit from the value accrual to the handful of AI companies.

This logic extends to countries too. Every nation will eventually need to on-shore their AI capabilities: their own LLMs (or access to open-source ones), GPU data centres, and power generation.

Because if the US decides to sanction you and Anthropic stops serving you overnight, you lose all your employees (AI Agents) and your white-collar GDP goes to zero.

If someone bombs your data centres, your GDP goes to zero.

If you get energy from a neighboring and country and they cut you off one day, your GDP goes to zero.

As all employees of the future are AI Agents, controlling your AI stack becomes the #1 national priority.

Some parting thoughts…

I have a friend from university who works at a LLM lab in SF and he’s been saying since 2022 that we’ll achieve AGI in a few years and there will be Universal Basic Income so there’s no need to worry about anything in life.

I used to think he was crazy, but now I’m starting to see it.

The question isn’t really whether AI replaces most cognitive work. It’s whether the transition is fast enough to be disruptive (the COVID analogy) or slow enough to be absorbed (the agriculture analogy).

As always, the answer is somewhere in between — faster than the optimists expect, slower than the doomers predict.

Either way, if you want to make yourself Claude-proof: Work in the right areas, build things, own the stack.

Humans are the new horses.

I started using Cursor along with Sonnet/Opus but as someone who hasn’t written production code professionally, there is still quite a lot of inertia as you face many system errors and have to repeatedly prompt the AI to fix it (yes, I’m complaining as a privileged post-Claude vibe coder).

One fine day I finally decided to try Claude Code, and everything changed. The biggest difference to me is that Claude Code was able to:

1. Use your computer via CLI to execute commands, install new packages, create & edit files, manage git, SSH to an AWS instance, etc.

2. “One-shot” tasks by writing a detailed prompt via Planning mode, troubleshooting system-level issues via CLI, and running unit tests to ensure the output was correct, otherwise it would recursively fix the code

Within two days, I had the first version of my Telegram bot which alerts me if a potential arbitrage has appeared. I would click trade it, as I wanted to make sure the bot is correctly identifying arbs first.

After a week of adding arbitrage strategies and refining the code, I managed to get it to a point where I was confident enough in the reliability of the algo and also my understanding of market microstructure (I was also getting tired of manual click trading). So I gave it $500 to play around with.

It quickly deployed that $500, then I topped it up to $1000, then $3000.. and after topping up for a week, it was managing ~$150k autonomously. The strategy has a ~15-20% APR and a maximum capacity of probably ~$200-250k.

Ideally it would be a y=mx line but the start of the Iran war on the last day of Feb brought about insane volatility which made my algos go a little crazy (only one leg manages to execute), so I had to make quick fixes right before a wedding. This volatile moment ended up being very profitable, even though the algo was lost money at the start of the war.

Whenever you have a profitable strategy, you should always ask why it exists.

My best guess is that the strategy is slightly hard to code and maxes out at ~$30k/yr in profits1 which is not interesting enough for a quant engineer to do (they get paid >$200k/yr at a fund). But Claude Code lowers that fixed cost to ~$300 in credits2, making the long tail of small arbs suddenly viable (even to the unemployed kek).

This got me thinking a lot about how the trading industry will change with Coding Agents.

As the cost to build and maintain infra decreases, arbs with small capacity become viable, until more participants drive those inefficiencies away too and so the market more efficient.

More broadly, we’re in the early innings of AI reducing headcount in the “implementation layer” across trading (e.g., analysts building dashboards and running backtests). A PM with an army of AI agents can now run a wider ticker universe than a PM with a team of human analysts ever could.

This pushes everyone up the stack. Every software engineer will have to grow into a Product Manager, and every trader/analyst will grow into a Portfolio Manager — managing a team of AI agents doing the grunt work while you make the higher-level decisions.

Anything that can be done to a “good enough” level will eventually be handled by AI agents: translating specs into code, auditing, compliance, paralegal work. But trading and investing is largely a relative, zero-sum arena, so being “good enough” doesn’t make money, only being better than your counterparty makes money.

That’s why I don’t think it’ll be fully automated by AI Agents. Instead of competing on who has the best analysts, you’re competing on who has the best AI agent setup with the most differentiated insight. Because if everyone has that alpha, it’s not alpha.

Three months ago I couldn’t ship a working bot, and now I have a live bot managing a decently profitable strategy. It really feels like a one-person hedge fund doesn’t sound so crazy anymore.

(i) Insider Risk

On (i) the insider risk is simply the # of potential insiders (e.g., team members working on a crypto project or tv series) * the propensity to leak (i.e., how much crypto knowledge they have such that they will come and trade on Polymarket).

Given prediction markets still aren’t governed by the CFTC, insider trading is in a grey zone (even though retrospective enforcement is possible). I think these markets will always exist, but they will have limited two-sided liquidity as market makers won’t quote here.

The easiest way to spot insiders is when they bet +$100K on something at 90c odds, because why would you risk $100K to win $10K unless you’re 100% sure. From their counterparty’s perspective, they are gambling on a 10x hence why they take the trade.

You would think that insider markets would be much smaller given you need a lot of suckers to take the other side of the trade. But on Polymarket there are usually a lot of suckers (see: “uninformed flow”) partially because there isn’t a “THIS MARKET IS RISKY” label and also partially because these insider bets work best when going from 90c → 100c, i.e. the sucker is gambling for a >10x convex payoff.

“What’s cool about Polymarket is that it creates this financial incentive for people to go and divulge the information to the market and the market to change, and all of a sudden it’s trading at 95 cents.” - Shayne Coplan, Founder of Polymarket (source)

Here I classify some markets based on insider risk. I think it is important to note that they should be “directionally correct” and not the hard truth. Also, it is hard to know if it was definitely an insider, or if the person had some other edge in finding the info early via scraping webpages, API, etc.

Honestly I don’t think it’s +EV trading insider information just to make a few thousand dollars with the risk of getting legal risk retrospectively.

Examples1 with twitter posts linked:

• “What day will Lighter airdrop be?” 29th Dec insider placed +$125k in bets (it was delayed to 30th)

• “Will Stable launch a token in 2025?” insider placed +$285k in bets

• “What day will Gemini 3.0 be released?” Google insider now 5-0 on Google-related markets, won +$1M so far

• “Time 2025 Person of the Year?” less so insider but was leaked one day before

• “Nobel Peace Prize 2025” what people thought to be an insider, likely wasn’t

• “Maduro out by..?” insider invested $30k one day before the US caught Maduro, to make +$400k.

(ii) Bounty Markets

On (ii), there’s the question of how all “prediction markets” eventually become “bounty markets” or “manifestation markets”.

The most egregious form of this is when the OG prediction market Augur in 2018 became an assassination market.

These were “prediction markets” to bet on whether prominent politicians, entrepreneurs and celebrities would be assassinated. There were also markets to bet on mass shooting and terrorist attacks with a minimum number of casualties.

As an example, if someone wanted to put a $100k bounty on Hilter, they could in theory create the market “Will Hitler be die of non-natural causes by 31st June 2026?”, then put a limit sell of 100k YES shares at 1c. This is equivalent to limit buy of 100k NO shares at 99c, which costs $99k.

The potential bounty hunter i.e. the assassin just needs to spend $1k to buy up these 100k shares at 1% odds, then go kill Hitler and wait for the market to resolve to claim his $100k. If the prediction market integrate a fully anonymous crypto, this could all be done privately.

Another type of “bounty” is in the form of manipulation of the underlying price. This usually occurs when the derivative market becomes bigger than the spot market, then you can manipulate the underlying spot market prices to make larger gains on the derivative (the prediction market, or sometimes the perp market by “dodgy market makers”).

The “manipulability” of a market is simply how easy it is to manipulate (requires physical presence? or large sums of money?) and the incentives to do it (is there sufficient $ reward to manifest the market?).

Here I classify a few2:

• “Total commitments for Solomon public sale?” (hard to pin this down to “manipulation” as ICO commitments could just be over-eager participants)

• “Will Kabuto 1st Edition card hit $100”, relatively easy to manipulate this by buying up all the Kabutos, and incentives were there (prediction market winnings > cost of buying up Kabutos)

• “Logan Paul Pikachu Illustrator Sale Price?”, easy to manipulate. But the example given in the tweet doesn’t work because there’s not enough liquidity, i.e., prediction market winnings < cost of buying the multi-million dollar Pikachu

• “Will the Gavle Goat burn in 2025?” bounty market, spoiler alert, ~$5k was not enough for someone to go to Sweden and commit arson

• “Dildo thrown at WNBA game?” bounty market, this went very viral in August such that WNBA now has a “no bag” policy. Here legal stakes were low enough that the incentives (+5.5k profit) justifies it

• “What will Coinbase say during their earning call?” went pretty viral as Brian Armstrong simply read out the list of words in that mention market. It is easy to manipulate but the incentives for high-status folks to game it is low.

• “Lakers vs Pelicans” I put a sports game here just as an anchor point, especially given the recent FBI arrests on Rozier, Chauncey Billups in the rigged sports betting

• “Israel Strike on the West Bank by ___?” which I put at the absolute bottom because there’s no way the prediction market becomes big enough to influence geopolitics and war.. right?

Prediction market platforms need to do ensure that “assassination markets” or any markets that could become a bounty for doing harm are not created. The real debate is around “war markets” such as “Will Israel strike Gaza on 5th Jan 2026?” are a slippery slope to incentivising violence, especially if the prediction market size is big enough.

As for non-violent bounty markets, I think Polymarket should put on appropriate risk labels so the degenerates know that they are just burning money where the more you buy on the YES side, the more likely the opposite is to happen.

As for fulltime prediction market traders, the idea should be to have a clear framework on classifying markets. Then, focus on generating alpha on markets with lower risk of insiders and manipulation. And if forced to play high risk markets, spot the insiders early, and tail them into events.

If you combine the two charts above, you should arrive at something like this.

Overall I still think that using Polymarket to trade is the most +EV thing you can do right now due to the potential airdrop (confirmed by their CMO). Given that Polymarket has raised at $11B FDV, even a 10-20% airdrop at half that valuation would be a $500m-$1b wealth generation event.

Also, it’s generally pretty fun to learn about the world and putting money where your mouth is.

Glhf!

Maplestory is a 22-year old game franchise owned by Nexon (Tokyo exchange listed company). It commands cult-like following especially in South Korea (KMS), Taiwan (TMS), and Singapore (MapleSEA).

Nexon recently invested $100M into the web3 version of Maple, Maplestory Universe (MSU), which is being developed by its crypto arm @Nexpace_HQ, all powered by the Nexspace token ($NXPC).

What exactly is Maplestory Universe?

Maplestory Universe (MSU) is an ecosystem of games and apps utilizing Maplestory's IP NFTs and $NXPC.

Within MSU there exists IP NFTs (e.g., "Ember Sword NFT"). Anyone can create UGC games or apps which utilize these IP NFTs. You can integrate the NFTs into your game by minting them from the MSU Item Universe using NXPC.

The first flagship title in MSU is MaplestoryN (MSN).

Its core game loop is the same as the original franchise - grind your character to lv250 while completing quests, killing monsters and bosses, and doing party quests, to secure the in-game currency (NESO) and equipment.

Nexon basically invented F2P gaming 22 years ago by integrating somewhat gambling-related aspects into (gachapon tickets and % probability upgrades). This has now become the norm that most F2P games nowadays have some sort of blind boxes integrated.

The game devs are incredibly good at dopamine hacking. The game revolves around (i) daily boss raffles and (ii) slot-machine-style equipment upgrades with % chance success rates.

MaplestoryN blends crypto into the game in a way that makes the most sense:

1. $NXPC acts as a crypto-economic balance system. Games like Maplestory suffer from inflation, continuously decreasing the value of items. In MSU, there is a natural economic equilibrium state as a basket of items can be converted back into $NXPC if there is severe item inflation, or vice versa if there's too much user demand, you can mint the items by using $NXPC. There is also an underlying dynamic pricing system (dutch auction) for improving the Star Force and Potential of your equipment, which ensures the game is not flooded with equipment of high stats. Hence good equipments will retain value.

2. $NXPC as a payment rail. The original Maplestory games were similar to Runescape in the sense that there was a huge p2p market where players would trade in-game currency (mesos) or rare equipment for real money. The marketplace was rife with scams as one party had to transact first (one tx in-game, one tx via a bank outside the game). It was incredibly common to get rugged.

What is the NXPC token even for?

The NXPC token distribution is healthy - no VC allocations.

• 80% for Contribution Rewards, i.e. for those who create content for MSU, including the MSN team. Minting of NFT equips and NESO dropped from in-game monsters comes from this bucket

• 16.3% for the Early Community (incl. 2% airdrop to players and the rest airdropped by CEXs to their customers & ~5% used by the team for Fission to provide equipment NFTs at launch)

• 3.1% for Team & Advisors

• 0.6% for Gas Sponsorship and Liquidity Provision

The NXPC sinks / utility are:

• Item Enhancements (Potential & Star Force): This is by far the largest token sink. The end-game is to max out the Starforce and Potential of your equipment, and these upgrades resemble gambling where there is only an x% chance for you to increase your stats.

• Fees from NFT marketplace (5% fee on NFT volumes)

• NXPC Fission (to be launched): You can use NXPC to mint a basket consisting of one of every equipment in MaplestoryN. This is not released yet, but likely to be used only by arbitrageurs / market makers as it will likely require >40K NXPC to mint the equips and sell them in the marketplace. It follows a bonding curve so the Fission price increases as more people mint NFTs.

• Gas token for the Henesys L1 which is an Avax subnet (small relative to the first three). Gas used is returned to the gas sponsorship wallet.

The NXPC emissions (faucets) are:

• NESO drops in-game from killing monsters / bosses

• NXPC Fusion (to be launched)

What do the on-chain metrics show?

To decrease the NXPC circulating supply, you need to ensure the Sinks >> Faucets. Hence, you want "Sinks - Faucet > 0", or...

(NXPC spent on Enhancements) + (Fees from Marketplace) - (NESO Emissions) > 0

The beauty of the blockchain is that you can track the entire economy transparently onchain.

First, let's look at Enhancements. NXPC spent on enhancements keeps increasing as more and more players reach the "end game" (~lv150+) where you have obtained the top equipment and need to enhance them to improve your stats.

MaplestoryN is removing ~76M NXPC from circulation from just item enhancements alone.

Secondly, let's look at NFT Marketplace Volumes.

Anecdotally it's an incredibly active marketplace where even specific equipment such as Earring NFTs transact >300 times a day, way more than traditional PFP NFTs in crypto. The top NFTs in the marketplace go for more than 888M NESO or +$12K.

In the 26 days since launch, they've done ~7.32M $NXPC in volume. The 5% fee they collect equates to ~0.37M NXPC so far or ~5.2M NXPC annualized.

Now for the emissions, which currently only come from monsters in-game.

The best way is to follow the NESO Emissions Wallet on Henesys L1 (0x18) which started with 900B NESO (minted with 9M NXPC from "Contribution Rewards") and now has ~753B NESO, meaning they've emitted ~147B NESO so far. The revenue wallet (0xcd) has also collected ~605B NESO (roughly tallies with the 580B we calculated above).

Hence, NXPC is on pace to remove 65M tokens or 6.5% of total supply (~22% of y1 supply) from circulation over the next 12 months. This is +$100M in net revenues for the game.

This assume zero growth, while games if successful are usually explosive and have exponential growth (remember Axie). The sink could grow exponentially while emissions remain constant, making the net reduction much higher.

At $1.3B FDV, this translates to a mere ~13x P/E, for a growing game that hasn't hit its inflection point yet.

The final cross-check you can do is looking at the net inflow from Henesys L1 to Avax C-chain as CEXs only accept NXPC on Avax-C. The net removal of 4.58M NXPC comes from people buying on CEXs and bridging it into Henesys L1 to spend, which you can clearly see from this Dune dashboard by the chads at @hashed_official.

The best part about this is that they are already burning 6.5% of total supply annually, with more catalysts lined up:

• NXPC Fission / Fusion which is a large sink for $NXPC as it'd require 40K NXPC to mint a single basket of equipment, and this price goes up following a bonding curve

• Launch of other MSU apps such as character lending (cf. Axie) and the "meta-gaming" layer (e.g., in FIFA apps, adults spend more time modding their team and collecting cards rather than playing the game football game itself). Some UGC has started popping up based on Henesys L1 NFT data (as seen here).

• Marketing to their Korean audience (>50% of all South Koreans have a Maplestory account)

It is still very early. The key here is for the team to communicate further about their plans to utilize the funds from the sinks that currently just sit in the revenue wallet. This is similar to the HYPE Assistance Fund. The team needs to decide how much of it to burn permanently or re-inject it into the ecosystem. So far, the team is still quite well-capitalized by Nexon.

Putting the financials aside, the organic user numbers are also looking good. All metrics are up and to the right, including unique actives, user stickiness, and new users.

The Maplestory franchise (GMS + KMS + TMS + MS Worlds) as a whole has 6-10M DAUs. MSN has currently grown to 70-80k DAUs, but if it can reach just ~10% of their DAUs (~600k-1M) just by converting players, that'd be huge for MSN.

New users generally only start engaging in revenue-generating activities (equipment enhancements and NFT trading) above level 100, which will take time for many users to reach.

Hence I think we'll likely see a strong continuation in revenue growth from the cohort of new players who are in the process of progressing to the late game.

Subscribe now

Airdrop fractals & MSU's challenges

So you're telling me all the numbers are up and to the right, for a game with growing user numbers (20k -> 50k DAU), +$100M annualized net revenues (sinks) with more catalysts in the pipeline... and the chart looks like that?

Yes, and IMO the biggest mistake the team has made is the CEX catch-22. CEXs ask for large allocations so the CEXs can airdrop to their users. For example, CEX users get NXPC for staking BNB tokens or generating volumes on Binance. They probably don't even know what NXPC is used for, so obviously they'd dump it for cash.

From @Jayyqueu's analysis you can see that 76M NXPC (7.6% of total supply) were given to CEXs and presumably sent to their users free-of-charge. Case in point is that when Bybit's trading competition ended and their users received their NXPC rewards, the charts started printing red candles only.

People complain about VC allocations and forget that retail airdrops with no lockups can be worse.

Even though early signs are looking good, the team is still facing multiple challenges, such as to attract players from KoreaMS, MS Global, and MS Worlds into MSU.

To do this, they need to slowly make MaplestoryN more feature complete (rolling out new jobs progressively), advertising to existing Maple players (currently zero marketing spent in Korea due to regulations), and catching headlines with high payouts (MapleLands in MS Worlds went viral because of this).

The high payouts are difficult to balance as you need both farmers and whales to join to maintain the equilibrium. Hence you need a strong unique selling point to migrate KoreaMS whales to MSU.

I've met few whales in-game who've spent +$50K across multiple years in KMS who have moved over to MSU and spent +$10K since launch. Their reasoning is straightforward.

KoreaMS has infinite item creation (as with most web2 games), so the price of items trend to zero very quickly. Because of that, KMS devs introduced a trade count restrictions for each item to ensure there are item sinks (e.g., "sword" item can only be traded 10 times). This is a temporary solution as items with fewer "trade counts" remaining depreciate very quickly.

As a whale, a game with guaranteed limited supply enforced my smart contracts with unlimited trade is way better.

You can tell the demand for MSU equipment has been high and has recently found an equilibrium. This Noble Ifia Ring (boss equipment set) has increased in price from ~150k NESO (~$6) at the launch of the game, to ~7M NESO (~$80) now.

Whatever you spend in web2 Maple, you might walk away with 10-15% of what you spent when you quit / decide to sell. Within MSN, it's likely 80%+ or even profitable, considering the demand for the limited number of rare items. - some whale I met in-game

Another huge issue is around botting / hacking. This has always been a constant cat-and-mouse game in the original Maplestory franchise but now that a lot more real money is involved, the stakes are higher. There has been recent reports of botting and hacking, but the team's swift responses to the incidents have been super impressive. Even the community is impressed.

And finally, the UX. The game requires a VPN (US, UK, Korea banned), a Metamask/Rabby wallet, a Windows PC, and a complex process if you wanted to spend money (i.e., buying NXPC via CEX to withdraw to Avax-C chain, then bridging to Henesys L1 to convert to NESO).

The team is already working on their own wallet with improved UX, but the VPN requirements are due to regulations. However, there's been some tailwinds on that front as South Korea just got a new government last week and the new government's Chief Policy Officer is the former CEO of Hashed Open Research (@hashed_official).

Final Thoughts

Nexon and Maplestory are the perfect candidates to pioneer web3 gaming. They created F2P gaming which is now the most profitable and common game monetisation method.

Maplestory has a diverse and global user base that has generated +$5B in lifetime revenues through digital item sales and all the users are used to using "digital currency" in the form of Gift Cards ("NX Cards") you buy at a physical shop (eg CVS).

Nexpace is also in an extremely advantageous spot because they can focus 100% of their time on game mechanics and design, instead of building out the game because they have a battle-tested off-the-shelf game and IP that can be deployed off the shelf.

Despite facing multiple challenges in the few weeks post-launch, user metrics are looking good and up only. The game has already generated +$100m per year in profits / net revenues at this rate and there's still insane room for growth.

Ultimately, I think MSU is Web3 Gaming's only hope. If it doesn't work, perhaps Web3 Gaming is truly fubar. But if it works, MSU could become the poster child of web3 gaming, and be one of the real world use cases of crypto.

If you're keen on following MSU updates, follow them: @MaplestoryU, @nexpacetime (Chief Strategy Officer), @msu_boss, @towelholderA (Nexpace CEO), @domidom88.

The MSU Community is also incredibly active. Also mostly @cowooding (leader of the #1 guild @limelight_gg), @eth_apple, @kss4319, @scv0510, @universeguild yapping.

Shout out to fellow MSU enjoyers for some great discussions: @mattmaximo1, @cptn3mox, @JayQueu, @0xlaiyuen, Gabriel @ Republic Digital.

Also if you're in game add me on the Ain server, ign is noobie

Disclaimer: This post is for general information purposes only. It does not constitute investment advice or a recommendation or solicitation to buy or sell any investment and should not be used in the evaluation of the merits of making any investment decision. It should not be relied upon for accounting, legal or tax advice or investment recommendations. This post reflects the current opinions of the author and does not necessarily reflect the opinions of the author’s employer. The opinions reflected herein are subject to change without being updated.

It doesn’t matter whether your $1,000 sits in a Citibank account or a Bank of America account — it’s functionally the same dollar. That’s because all commercial bank money (like "citiUSD" or "boaUSD") is backed by the central bank and seamlessly interchangeable under a regulatory umbrella.

This is what economists call the singleness of money.

Let me translate it into crypto terms: When you deposit $1,000 of physical cash into your Citibank account, you are minted citiUSD. When you transfer your funds from your Citibank account to your Bank of America account, your citiUSD gets swapped to boaUSD, worth exactly $1.00. The "citi" and "boa" portions have been abstracted away from your "wallet".

You never notice this swap because the government guarantees fungibility and par value. The value and utility of that dollar is the same, no matter who issued it.

Not all dollars carry the same risk

Even though physical cash and “cash” in your bank account feel the same, one is safer than the other.

The dollar takes a few forms, from least to most risky:

• Physical cash (a direct claim on the central bank)

• Narrow bank deposits (e.g., stablecoins since issuers don't lend)

• Fractional reserve bank deposits (typical banks with credit risk)

Bank deposits are IOUs from private banks. If the bank fails, you may lose access to your funds. Well unless it's a systemically important bank (GSIB), that the government deems too big to fail, then they may intervene.

Stablecoin issuers, if regulated, function like narrow banks. They don't lend out your deposits; they hold only cash and high-quality liquid assets (HQLA), like short-term Treasury bills.

So all things equal, stablecoin dollars are safer than bank deposits. They carry no credit risk and have lower bank-run risk. The main bank-run risk is on the bank deposits held by issuers (e.g, Circle holds $6.7bn out of $60bn in bank deposits for liquidity reasons, which in 2023 was held at SVB, Silvergate, and Signature Bank, among others) because they cannot hold reserves directly at the Federal Reserve as they lack a Fed Master Account. Only ~4,500 entities (banks, credit unions, certain government entities, etc) have access and none crypto-native. (@bridge__harris goes into a lot more depth here).

This provides an interesting scenario where holding Paypal's $PYUSD is safer than holding Paypal's dollar on their app. Both dollars are issued by the same company, but one is superior to the other.

Ideally, stablecoin issuers would hold reserves directly with the Fed, earning risk-free yield and eliminating reliance on commercial banks. This would effectively turn stablecoins into synthetic CBDCs (called “synthetic” because the central bank doesn’t deal with retail users directly).

But why hasn't this happened?

As Bridgett notes, the Fed is unlikely to grant stablecoin issuers Master Accounts—just as it rejected The Narrow Bank’s application and denied crypto banks like Custodia. However, it is possible that Circle has a sufficiently close relationship with their partner banks for a master account to not be a significant improvement in money movement.

Ultimately, the US system is built on fractional reserve banking, and synthetic CBDCs would disrupt that model by draining deposits from commercial banks into the central bank. A more compatible approach? Let banks issue tokenized deposits themselves. This may soon be required: under the GENIUS Act, stablecoin issuers with over $10B in circulation would need to come under Federal Reserve oversight.

From the above, it might seem like we can only have singular money via tokenized deposits or Synthetic CBDCs (multiple stablecoin issuers that have Fed Master Account access).. if we're not creative.

In Crypto, $1 ≠ $1

In crypto, stablecoins such as USDC, USDT, PYUSD, DAI look similar on the surface but aren’t actually fungible. Try swapping $10mn USDT for USDC in one go. You’ll lose ~5 bps due to slippage (~$5k) as you trade on an orderbook / AMM.

Obviously USDC <> USDT shouldn’t be 1-1 because USDC acts as a narrow bank with only bank deposits & HQLA reserves, while USDT is only 82% backed by bank deposits/HQLA, and the rest in BTC (5.5%), secured loans (5.7%), precious metals (3.7%) and "other investments" (2.8%).

But does that mean USDC <> PYUSD should work given they're both backed by HQLA (albeit a different composition of T-bills)? Also no. They're not interchangeable at par value on-chain, simply because they're issued by different parties without a singular off-chain infrastructure.

This is the core problem: there’s no singleness of money in crypto. 1 USD stablecoin is not worth 1 USD across the board.

There are a few ways to solve this, such as:

• Synthetic CBDCs and Tokenized Deposits (boring)

• Stablecoin Consortium where members accept each other's stablecoins at par and settle in the background (somewhat similar to the Global Dollar Network which involves Paxos, Visa Robinhood, Kraken, Galaxy Digital)

• Slightly more complex series of unsecured / secured loans with different DeFi protocols albeit with some size constraints, such as with

  MakerDAO PSM (detailed explanation by Seb here)

• M^0.

I think the first three are self-explanatory, so let's dive into M0.

(Note: Neither myself or my employer are investors in M^0 and I have no vested interest in it. It is simply something that sparked my curiousity, tweeted this a few weeks ago.)

M^0: The On-Chain Central Bank

M^0 is a "stablecoin-as-a-service" platform which acts as a coordination layer that lets multiple trusted institutions mint the stablecoin $M, which can then be used to back their own-brand stablecoin. It separates monetary issuance from risk-taking — just like how the Fed operates in TradFi.

The system revolves around three actor types that must be whitelisted by governance:

• Minters: whitelisted issuers who deposit eligible collateral (<180d T-bills) to an eligible custody solution to mint $M

• Validators (e.g., @ChronicleLabs): act as "auditors" who verify reserves in the eligible custody solution, and timestamps transactions to authorize the minting of $M

• Earners (e.g., DeFi protocols launching stablecoins, institutions, CEXs): whitelisted entities who can earn yield by holding $M

This structure allows for multiple issuers of a single digital dollar, all backed by shared collateral and common rules.

Minters pay a Minter Rate on the $M they mint, currently set at 4.15% through governance. Earners earn the Earner Rate on the $M they hold, currently also set at 4.15% through governance.

The new Stablecoin Bill does not allow yield to be passed onto stablecoin holders. Hence M^0's model only passes on yield to permissioned earners.

These earners are institutions, stablecoin project teams, and CEXs from approved jurisdictions. It is then their duty to ensure they are following local regulations when providing yield to their end customers.

As an example, Binance could be an Earner, earning 4.15% on $M held on their exchange. Then, they decide how much of that they want to pass onto their users on the Binance Earn programme. Similarly, the $M that Noble (an Earner) holds will earn yield, and Noble will decide how much of the yield they want to pass on to $USDN holders.

Given that not all $M will be held by Earners, the remaining gets sent to a Distribution Vault where $ZERO holders earn revenues.

Distribution is King in the Stablecoin Wars

Since Circle's S-1 filing, it's never been clearer that distribution is king for stablecoin issuers. Circle made ~$1.7bn of revenues in 2024 while paying out a whopping ~$900mn to Coinbase as a distribution partner (who then passed on most of it to their end users). Circle also paid Binance a one-time $60M fee, plus a monthly incentive fee to hold USDC on Binance.

Tether on the other hand has established itself as the stablecoin with deepest liquidity and most CEX pairs via Bitfinex and its early use for cross-CEX arbitrage. For this distribution network effect, it made $13bn in profits in 2024, with ~$140bn of USDT issuance.

Anyone with distribution will want to launch their own stablecoin to earn the juicy ~4% net interest margin (NIM) with zero cost of capital. The problem new stablecoins (call it newUSD) face are:

1. Lack of newUSD-USDC-USDT liquidity. Newer stablecoin protocols typically emit protocol tokens to incentivize users to bootstrap the USDT-USDC-newUSD Curve 3pool. This is usually not a great use of emissions as it attracts mercenary stable-farmers.

2. The off-chain setup requires heavy legal, compliance, and operational expertise. Hence, new stablecoins would consider using B2B stablecoin providers, just like how Paxos set up PYUSD and what used to be Binance USD (BUSD).

M^0 solves these two problems.

All stablecoins fully backed by $M will be interchangeable via $M (similar to Maker's USDC PSMs). Hence to go from newUSD to Noble's $USDN, you simply get an atomic 1-1 swap from newUSD > $M > $USDN. This solves the singleness of money for all stablecoins in the M^0 ecosystem.

Needless to say, there is an immense network effect where if there are more stablecoins in the M^0 network, liquidity increases because these stablecoins are interchangeable 1-1 by routing via $M, making it a more attractive proposition for new issuers. Currently, they are working with @noble_xyz, @usualmoney, @USDAI_official, and @KAST_official.

Note that to go from any $M-backed stablecoin to USDT or USDC, M^0 still needs to bootstrap the M-USDC or USDT liquidity. Otherwise we'd face the same issue of slippage. True singleness is only achieved when M^0 achieves network effects. The team has seeded $18mn in M/USDC liquidity on a single tick at $1.00 on Uni V3 as seen below.

The second point is also an interesting business model problem for Stablecoin-aaS companies. These companies typically charge a small 20-30bps fee on the AUM, making only ~$2-3m on a customer with a BILLION in stablecoin issuance.

The entire stablecoin market cap is $233bn, but excluding USDT ($145bn), USDC ($60bn), and decentralized stablecoins leaves us with an existing market size of $10bn. Capturing 100% market share here would only provide $20-30mn in revenues.

These businesses require a large AUM for the business model to make sense. Yet, when their customers get big enough, they'd be incentivized to run their off-chain operations themselves to get that additional margin. M^0 side-steps this by having a more creative revenue stream (yield from non-Earners).

Besides that, the programmability of stablecoins allows its distributors to stream T-bill yields to incentivize certain actions. For example, L1's could direct interest earnings to devs who burn the most gas, or exchanges like Hyperliquid could stream interest earnings as rebates to builder code businesses who are pushing Hyperliquid distribution (hint hint 👀).

M^0's Dual-Token Power Structure

M^0 marries the best parts of centralization (holding tradfi risk-free T-bills) and decentralization (governance-voted set of Minters, Earners, Custodians, and Collateral).

Governance involves $POWER (1mn initial supply) and $ZERO (1bn initial supply).

$POWER is the management token used to vote on day-to-day proposals. If you vote, you receive a pro rata share of 10% $POWER inflation + 5mn $ZERO every epoch (30 days). The non-voter's emissions are auctioned off in a Dutch auction. Unclaimed $ZERO tokens are not minted.

This mechanism heavily incentivizes $POWER holders to participate in governance, as you prevent dilution and earn "revenues" via $ZERO emissions which start at 0.5% of total supply (5mn vs 1bn $ZERO supply) and slowly decreases in %. You can delegate your $POWER to someone else to vote, and receive the 10% $POWER inflation, but the $ZERO is emitted to the person doing the governance.

$ZERO is the guardian token used to vote for proposals which cause big changes to the protocol, such as Reset Power, and setting the $POWER and $ZERO thresholds to pass proposals. The Reset Power switches the management token to a new token (eg $newZERO) and distributes it pro rata to the current $POWER holders, effectively reclaiming management rights to themselves.

The utility of $ZERO is to receive revenues generated from yield on $M held by non-Earner addresses, the spread between the Minter Rate and Earner Rate (currently zero), and Fees (payments for $POWER token auctions, proposal fees, and penalties).

Final Thoughts

Stablecoins, if structured well, could become safer than traditional deposits. But without singleness, we’re stuck in a fragmented, inefficient state.

M^0 offers a compelling attempt to bring the singleness of money to crypto. It separates monetary issuance from risk-taking — mirroring how the Fed operates in TradFi — and creates a neutral base layer ($M) that stablecoins can build on.

Ultimately, what M^0 proposes is not just a new stablecoin — but a new monetary coordination layer. One that asks: What if crypto dollars had the credibility of T-bills, the neutrality of the internet, and the programmability of DeFi?

The answer may define the next era of digital money.

Special thanks to @lucaprosperi, @sebventures, @fiege_max on the discussions and feedback.

It’s incredibly tough to find people who understand both crypto stablecoins AND traditional banking, hence you should definitely follow:

1. @sebventures, founder of @SteakhouseFi, writes the CryptoBanking blog explaining stablecoins with double-entry accounting

2. @sytaylor, writer of FinTech Brainfood, covers primarily Fintech and occasionally the usage of stablecoin tech in Fintech

3. @jp_koning, writer of the Moneyness blog, monetary economics nerd and wrote

   about bitcoin in 2014

4. @lucaprosperi, founder of M0, writer of dirtroads substack

5. @nic__carter, GP / founder of @castleislandvc who co-authored the stablecoin report

6. @cambelljaustin, stablecoin expert & writer

   ---

Disclaimer: This post is for general information purposes only. It does not constitute investment advice or a recommendation or solicitation to buy or sell any investment and should not be used in the evaluation of the merits of making any investment decision. It should not be relied upon for accounting, legal or tax advice or investment recommendations. This post reflects the current opinions of the author and does not necessarily reflect the opinions of the author’s employer. The opinions reflected herein are subject to change without being updated.

For crypto founders, attracting initial liquidity is often the biggest hurdle to overcome & kickstart the flywheel. "TVL bootstrapping" from funds (or whales) offers a tactical way to kickstart the liquidity flywheel, signal trustworthiness, and build early momentum.

The core idea is simple: the project offers a small percentage of its tokens in exchange for funds to temporarily boost TVL. Founders should ensure (i) they’re not giving away too much supply since that creates sell pressure, and (ii) that they can attract sticky TVL who are genuine users of the product, so it sticks around after the deal ends.

Nowadays TVL bootstrapping deals are priced very well given how common they have become. It's become a lot more competitive for funds, but great for protocols since they won't have to give out too much of their tokens.

The yield to a TVL provider ("fund") is simply "Base APR" + "Token APR".

Typical "Base APR" for yield deals

Base APR ranges between 4-20% depending on credit risk, roughly speaking from safest to riskiest:

• T-bills in bankruptcy remote structure ~4% APR (@OpenEden_X, @0xCoinshift)

• Overcollateralized defi lending ~5-10% APR (@eulerfinance, @MorphoLabs)

• Secured overcollateralized lending ~11% APR (@maplefinance) and tokenized basis trades ~6-10% depending on market environment (@ethena_labs, @ResolvLabs)

• Unsecured lending to reputable companies with good cashflow 12-14% (@WildcatFi)

• Lending to emerging markets, younger firms, GPU farms etc. ~15-20% (Goldfinch, GPU AI protocols)

Most participants in these deals come from venture or discretionary trading backgrounds — they want yield but often don’t have the credit expertise to properly underwrite risk.

Hence the "base APR" is mainly the focus of crypto-native credit funds (very few), while the main interest for yield farming and discretionary liquid funds will be for the Token APR.

How "Token APR" works

There are three ways to quote Token APRs:

1. More mature projects would quote a fixed token APR (e.g., 1% APR added onto Base APR). Then, the project would give the fund the exact amount of tokens based on a 7d TWAP which would provide you exactly 1% APR on your TVL.

2. Pre-TGE and younger projects would typically give you a fixed % of supply of tokens (say 0.5% of total supply if you provide $10M for 9 months) OR a points programme.

3. Projects with lots of interest could have an auction mechanism where they allocate x% of token supply and take bids for how much TVL funds/users are willing to provide for a period of time.

Note on (2), as a project there's no need to advertise ridiculously high token APRs with a small footnote that the APR assumes your token trades at a some crazy $10B valuation.

What's more helpful is if to share your previous fundraise valuations for TVL providers to anchor to as a base case. Funds would then build their personal bull cases based on their estimation of market demand for your token, the market regime, and token lock-ups/vesting.

Points programmes on the other hand are a bit riskier for funds as the "APR" is obfuscated, making it more trust-based. Projects are inherently incentivized to delay TGE and continue dilution of points to retain TVL, and that's how we've seen it play out across multiple protocols.

Note on (3), it is the most capital efficient way to deploy your tokens, but only works if you have sufficient bidders, hence works well if you're a well-capitalized L1 for example. However it could encourage collusion between the biggest TVL providers to strike the best deal.

Also note that all the above still apply if you're doing your TVL programme through TVL aggregators like @roycoprotocol or @turtleclubhouse who aggregate smaller ticket sizes to get better deals.

Deal terms to think about

A few considerations / parameters you might want to think about upfront:

1. Asset provided: Some funds only provide USDC/T while others may use their idle BTC / ETH / SOL.

2. TVL lock-in period: Some liquid funds, as the name suggests, are unable to lockup their funds due to their mandate and LP withdrawal requirements.

3. Structuring: It could be useful to structure the TVL provisioning as a loan from a fund, as some regulated funds may find it difficult to transact on-chain.

4. Documentation: Provide clear docs on your website (not as a TG message!). The most useful thing I've always found is a flow of funds diagram. Remember, funds need to do DD on legal, compliance, tech, counterparty risks, etc.

Implications for Builders, TVL providers, and investors

For builders, make sure you're not giving away too much tokens in this phase. Think about it as trading equity for a loan (convertible loan). Try and kickstart your flywheel ASAP to attract real use & sticky TVL. Pay off the loans as much as possible in USDC. Also obviously, you can customize your deals for each investor depending on their size, lock-up period etc.

For TVL providers, make sure you do proper diligence on the risks involved as TVL provisioning an endeavor with capped upside, but uncapped downside (total loss). Also, see through the "Token APR" shenanigans and mark them to your own token price projections. Customize deals so they fit your fund mandate, e.g., TVL lock-ups vs LP redemptions.

For investors, be aware that many protocols with large TVLs may have done TVL deals and hence may experience large cliffs of TVL drop-off. Ensure you are not blindly doing a "TVL multiple". Figure out how much of it is sticky TVL from real PMF (eg Aave).

Hope this was helpful for anyone navigating the TVL raise landscape. Goodluck!

(Thanks @vc_larp & @dataoverdose for bouncing ideas!)

The idea was simple: a new, “fairer” way to distribute TAO emissions across subnets.

But barely a month in, the cracks are already showing. Turns out, what looks good on paper doesn’t always survive the free market.

How dTAO works

As a refresher, here is a simplified explanation on how dTAO works:

1. Each subnet has its own subnet token ($SN) which exists in the form of a native TAO-SN UniV2-type pool. Confusingly, instead of “swapping” TAO for SN, they call it “staking” TAO for SN. They are functionally the same, except that you can’t add liquidity into the LP and you can’t trade between SN tokens, eg SN1 → SN2. But you can still do SN1 → TAO → SN2.

2. emissions are allocated proportionally to subnets based on their price. They use a moving average price to smooth out price spikes or attempts to manipulate price.

3. SN tokens also have high emissions with a 21M supply cap, just like TAO and BTC. A portion goes into the TAO-SN pool, and the rest goes to the subnet’s stakeholders (miners, validators, subnet owner).

   1. The amount of SN token that goes into the TAO-SN pool is the amount required to balance off the TAO emitted into the pool, such that SN price in TAO terms remains constant, while liquidity increases.

   2. However, if the math above works out to a number where the subnet receives more tokens than the maximum allowable SN emission (based on the SN emission curve), then it is capped at the max possible SN emission, and hence the price of SN token increases in TAO terms.

The underlying assumption in point (2) of the mechanism is that subnets with higher price generate more value for the Bittensor network, hence they deserve more TAO emissions.

But let’s be honest. In crypto, the tokens with the highest price are going to be the ones with the most mindshare, hype, ponzi, and marketing. This is why it’s always been L1’s and memecoins that have the highest relative valuations.

The intentions behind the mechanism design are good as it assumes that subnets that create value via generating revenues will use a portion of revenues to buy back SN tokens, hence have higher price, and deserve more TAO emissions. But I think this line of thinking is a little naive.

The memecoin subnet & broken tokenomics

Prior to the launch of dTAO, I was discussing with a few crypto analysts about the obvious flaw in dTAO tokenomics — namely higher price != higher revenues or higher value creation.

But little did I know, theory would come into practice right away. The free market works in beautiful ways.

Right before the upgrade, an anon took over Subnet 281 and turned it into a memecoin subnet named “TAO Accumulation Corporation” aka “LOL-subnet”. This obviously has nothing to do with AI.

In the (now removed) Github, it said…

Miners do not need to run any code, and validators score miners based on their subnet token holdings. The more a miner holds, the more emissions they receive.

What actually happens here is that speculators buy SN28 token → SN28 price goes up → SN28 gets more TAO emissions → SN28 price goes up (if it breaks the subnet’s token emission limits) + the SN token emissions passed onto “miners” in proportion to how much SN tokens they hold → people buy more SN to get more TAO → higher price → ponzi goes on.

TAO emissions have officially gone to funding… memes! At one point, it was even the subnet with the 7th highest MC.

But why hasn’t SN28 taken over Bittensor? Centralization saved the day.

Within a couple of days, the Opentensor Foundation used their root stake to run customized validator code to incentivize people to dump SN28 tokens, sending it down 98% in a few hours.

Essentially the Foundation acted as a centralized entity to stop what the free market wanted to do with the dTAO mechanism. This centralized takeover is only possible now because there is a slow transition period where the emission mechanism transitions from the old TAO emission mechanism to the new dTAO mechanism.

TAO’s old mechanism and transition to dTAO

TAO’s old mechanism basically allows the 64 validators who have the most TAO staked on SN0 (“root subnet”), to vote on who received TAO emissions.

This presented its own set of incentive problems driven by the power the large validators have, such as the Opentensor Foundation, DCG Yuma, Dao5, Polychain, etc.

Many potential conflicts of interest would arise, for example they could in theory direct TAO emissions to subnets which they have invested in or incubated, or to subnets where they run a validator for and are personally receiving TAO rewards for.

Hence, moving away from that is a good step towards decentralization. I applaud the team for choosing a more decentralized reward mechanism, even if it means they could lose out on emissions.

When SN28-fiasco happened, it was only ~1 week into the introduction of dTAO, hence SN0 (blue lne below) still had control of ~95% of emissions, and the Opentensor Foundation could intervene.

In ~1y, the power of SN0 to direct emissions diminishes to about ~20%. This means that if something like SN28 takes off again, it is unlikely that anyone can intervene via SN0. In this case, Bittensor could cease from being a “decentralized AI” project, but a memecoin incentivization network.

Let’s say it’s not all memes

Even if we assume that people are altruistic enough not to ape into memes in this bear market, there is a good chance that Bittensor might become a generalized incentivization network, completely unrelated to AI.

Consider the thought experiment where someone launches a subnet specifically to mine Bitcoin in a decentralized manner (This is not a very novel idea). The idea is for the subnet to incentivize mining bitcoin in a resource efficient manner, while using the BTC mined as recurring revenues to buyback their SN token to receive TAO emissions.

Hence, TAO goes from a decentralized AI project to a generalized incentivization project, where TAO emissions just goes to substitute the most random OpEx costs of businesses, instead of advancing towards a specific goal.

This should technically be a-okay given that the Yuma consensus mechanism was meant to form consensus around any “subjective” work, which need not be AI. But the lack of this specific goal kind of makes it all.. very much pointless.

Final thoughts

We’re only 1 month in and the cracks are already showing in the new dTAO model.

The free market incentives dictates that without any centralized force, Bittensor could stop being an AI project, but an “attention network” dominated by memecoin subnets, or a “generalized incentive network” dominated by revenue-generating businesses subsidizing their OpEx with TAO emissions without improving the Bittensor network.

I think what the network works is one true “objective function” which could align all subnets to a singular goal. However, it is obviously quite difficult to have one true objective in AI (AGI?) as we’ve seen how difficult it is to run fair LLM evaluation frameworks…. which is why the Yuma consensus for “subjective” work was created in the first place…

Tell me the incentives and I’ll tell you the outcome. Peace!

Disclaimer: This post is for general information purposes only. It does not constitute investment advice or a recommendation or solicitation to buy or sell any investment and should not be used in the evaluation of the merits of making any investment decision. It should not be relied upon for accounting, legal or tax advice or investment recommendations. This post reflects the current opinions of the author and does not necessarily reflect the opinions of the author’s employer. The opinions reflected herein are subject to change without being updated.

This time it’s different.

Everyone’s eagerly waiting for the “alt szn” (blue line to shoot above the orange line), where every single alt goes parabolic like in 2021-22. Yet, the BTC-TOTAL2 gap has been getting larger and larger since the launch of the Bitcoin ETF in Jan 2024.

In previous alt seasons, everything pumped indiscriminately as investors moved up the risk curve as they realize their BTC gains. Leading to the classic flowchart.

Nowadays, BTC lives in a world of its own, with inflows mainly driven by (i) ETFs which collectively hold 5.6% of all BTC, (ii) the perpetual buying machine Microstrategy which holds 2.25% of all BTC, (iii) macro, mainly interest rates, and politics ie. a US Sovereign Wealth Fund or other countries buying BTC.

Meanwhile on the outflow side you have (i) the US government who holds ~1.0% of all BTC and has signaled they might not sell, (ii) BTC miners, and (iii) BTC whales who are up ~5x from the 2023 lows.

Clearly these drivers are completely separate from the rest of crypto.

Altcoin flows: Are there enough players in the casino?

I like to think about the “altcoin market” as a casino.

You should only be playing when the casino has a bunch of money sloshing around (high net inflows), and you want to pick the right table to play at (asset selection).

The inflows to the casino are:

• New $$$ liquidity pumping into the casino. This can be either via..

  • Newly on-ramped dollars eg mass market retail “onboarded” into crypto in 2021. I do not think there is enough new money onboarded via Phantom / Moonshot nor the TRUMP token launch + increase in USDT/C MC

    Note: A specific table in the casino could also get inflows due to rotation from other parts of the casino (e.g., a midcurve who doesn’t invest in memecoins, starts investing into “AI memecoins” as they are more justifiable)

  • Dollars created via leverage either through DeFi (Aave, Maker/Sky) or CeFi (BlockFi, Celcius). From the institutional side, the epic collapse of lenders has made the CeFi side much less rampant compared to 2021. On the DeFi side, the IPOR index is a great way to track USDT/C borrow rates which have dropped from ~20% in December to ~8% now.

• Token buybacks / burns (the casino owners use their revenues to make the casino player’s chips worth more)

  • The best example is the HYPE insurance fund which has bought 14.6M HYPE, worth ~$350M at $24/HYPE.

  • Although usually crypto projects do not achieve sufficient PMF to print revenues that can perform buybacks that impact the price (cf. JUP buybacks)

The outflows from the casino:

• Liquidity $$$ cashed out from the casino (large off-ramping events)

  • We had TWO “once-in-a-lifetime” extraction events with TRUMP (0→$75B → $16B) and MELANIA (0 → $14B → $1.5B) in Jan. These collectively siphoned conservatively $1B+ out of the ecosystem. (i.e., if you made $10M+ on a single trade, you’d definitely off-ramp >50% of that)

  • The small but constant extraction via tools such as Pumpfun ($520M cumulative rev in ~1y), Photon (~$350M cumulative revs), Bonkbot, BullX and Trojan each also having ~$150M cumulative revs

  • Cabal extraction and pre-sales meta tend to mark the end of cycles because a handful of people extract large sums of money and just off-ramp them. These cycles become shorter and shorter towards the end of the cycle. PASTERNAK lasted ~10hrs, JELLYJELLY lasted ~4hrs, while ENRON pump lasted for literally 10mins. Truly a euthanasia rollercoaster experience.

  • VC unlocks such as TIA, where they cash out of the crypto casino into dollars to return DPI to their LPs

  • De-leveraging, i.e., lowering USDT borrow rates

• Token emissions of existing projects (e.g., Bittensor emits 7200 TAO per day or at ~$330 per TAO, or $870M per year) diluting existing players to the benefit of miners who presumably sell

Altcoin selection: Choosing the right table to play at the casino

When the casino is open (ie lots of players), your potential returns are higher.. assuming you choose the right poker table (ie the right token).

I call it poker, because these are inherently zero-sum games as projects either (i) generate no revenue / value or (ii) accrue that value to the token.

The only maybe-arguable exclusions from this are frequently used L1 tokens (eg SOL, ETH), or highly revenue-generating products (eg HYPE).

Note: Some “fundamentals-based” investing folks are betting on teams generating sustainable revenues in the future, but I have a more cynical near-term view.

The casino in 2025 is A LOT HARDER to find the right tables to play at, because there are way too many concurrent poker tables.

Every single day 50K unique tokens are launched via Pumpfun. In fact, Pumpfun has launched +7M tokens since its inception, with about 100K unique tokens graduating to Raydium.

There is no way we have enough players to sit on all these poker tables. Hence, altcoin returns exhibit strong dispersion.

Choosing the right table to play in is a whole artform itself, but generally it’d be something along the lines of (i) team / product, (ii) narrative, (iii) virality / marketing. Kel wrote a great article on that topic.

So what are the implications?

1. Given alts are not “higher beta BTC” anymore, the old portfolio theory of “I hold alts and no BTC” might not work anymore.

2. Alts asset selection is more important than ever when the barriers to launch a new token is almost zero thanks to Pumpfun. A rising tide does not lift all boats.

3. I know describing altcoin “investing” as poker is very nihilistic, but it is what it is in its current form. Perhaps I’ll write something about the real long-term use cases of crypto some other time.

Have we topped? Idk, probably for now.

Wen next alt szn? I think the four-year cycle breaks down now as alts detach from BTC. Alts will sporadically spring to life when triggered by something unexpected like GOAT.

Over a long enough time horizon, it’s never been more bullish with a potential US SWF, pro-Bitcoin administration, and stablecoin bills.

GLHF!

Subscribe now

Generally you need either novel tech or a new token distribution mechanism (i.e., method to speculate) to kickstart a proper bull run.

In 2013/14, we had the PoW craze (way before my time), novel tech.

In 2016/17, it was ICOs, new distribution mech.

In 2020, it was the first time DeFi was possible and we went crazy with farming literal yams, novel tech.

In 2021, we had NFTs (essentially illiquid sh*tcoins) introduced via free mints, and launchpads, IEOs, and IDOs with bonding curves, and so-called “DeFi 2.0” with (3,3) game theory, ponzis and APY’s that don’t even fit the length of your screen. New distribution mech.

In 2022-24, the most popular meta were the second coming of airdrop farming fueled by Jito (along with the growth of industrialized sybil farms), points farming (popularized by Blast/Blur), pre-sales 2.0 (i.e., just sending ETH / SOL to someone’s addy), and LBPs on Fjord Foundry (still remember those?). Arguably these are all just slightly modified distribution mech and nothing net new.

So what are the slightly more interesting “tech” / distribution mech I’m seeing in 2024?

• ERC-404 (PANDORA), the short-lived craze which is somewhere between an NFT and a token (tech?)

• “ERC-4D” (INCEPT) which is an ERC-6551 token that acts like a wallet which can own other tokens / NFTs. Reminds me of FuturePrimitive (tech?)

• Baseline’s programmatic liquidity (thick liquidity at “floor price”) and their “up-only” tech which, spoiler alert, did not go up only (remember $YES?)

• Node sales were hot for a minute but are very similar an ICO with tiered FDVs and a vest. Aethir raised ~$150M, Xai raised ~$40M, Sophon raised ~$60M, etc.

• I think zkTLS could be an interesting way to distribute tokens and introduce non-Web3 users into the Web3 world. I thought Tea protocol was pretty interesting as it was giving out tokens to web2 open-source contributors.

• “Mining” meta, mainly with AI protocols. There are a few protocols going the “21M BTC no investor / team allocation” route. You probably need to set up your rig via Runpod, Paperspace etc., but these suffer from insanely high inflation (as with BTC in the early days). There are also protocols like ORE which congested Solana at its first launch, and can be mined with CPUs

• And of course the most obvious new tech / distribution was Pump.fun’s “unruggable” way of launching new tokens, which is only possible on faster chains like Solana

All this is just to say that I’m sitting here waiting for a new on-chain innovation to pique my interest.

This post was inspired by this tweet by DeFiIgnas, imo one of the very few CT personalities worth following who doesn’t shill you some small cap coin.

Disclaimer: This post is for general information purposes only. It does not constitute investment advice or a recommendation or solicitation to buy or sell any investment and should not be used in the evaluation of the merits of making any investment decision. It should not be relied upon for accounting, legal or tax advice or investment recommendations. This post reflects the current opinions of the author(s) and does not necessarily reflect the opinions of the author’(s) employer. The opinions reflected herein are subject to change without being updated.

Quick life update: I’ve joined Brevan Howard Digital to do liquid crypto investments. I’ll still be attending a few conferences and meeting crypto folks, and will be layering on a more systematic / data-driven and macro lens to crypto investing. Whether you’re a trader, investor, or working on a token project, feel free to reach out!

I took a short break between jobs to enjoy the Olympics. This year’s Olympic games felt very different as you could follow the journeys of multiple Olympians leading right up to the Olympics..

• Allie Ostrander (US 3000m Steeplechase) qualifying for the Olympic Trials two months prior, but unfortunately finishing 7th at the trials. Not often do we get to see the pov of someone who didn’t make it.. (She gained ~70K subs though!)

• Noah Lyles’ (US 100m, 200m sprinter) both on SPRINT (on Netflix) which covers the sprinters (gives me absolute chills watching it!) and he makes great vids on Youtube. You really need that almost absurd level of confidence to excel at this sport, and really enjoy the showmanship

• Tara Davis-Woodhall’s (US long jumper) vlogs.. from the day she won the US Olympic trials to the day she won gold at the Olympics. The emotions that an athlete experiences when working hard for something for years and finally achieving it is insane. Her husband Hunter is also a 3x Paralympics medalist.

• Then there’s Ilona Maher (US rugby), Tom Daley (British diver), and Devin Booker’s (US basketball) reels and vids giving an inside-look of the Olympic village, many with 10-20M views!

What a time to be alive.. loving how the vlogs give a realistic glimpse into the lives of Olympians.

I made a weekend trip to Paris to catch the badminton finals. And boy was it worth every penny. The Malaysians had one of the craziest comebacks in the history of the sport — losing the first game 17-21 and down 16-20 on the second game, one point away from losing their spot on the podium.

They proceeded to win 6 straight points, taking the second game 22-20, and then winning a tight third game 21-19. The entire arena went wild. For a small nation like Malaysia, a bronze medal means a lot.

Malaysia also won bronze in the Men’s Singles, making it the nation with the most Olympic medals without a gold (15).

Some of the other highlights / things I found interesting:

• Every Olympic host can add and remove 5 events. Paris chose breaking (or breakdancing), sports climbing, surfing, and skateboarding. These replaced karate, softball and baseball introduced in Tokyo 2020. LA 2028 is adding baseball, cricket, lacrosse, flag football (?) and squash. (Breaking is removed!)

• Obviously you can’t mention breaking without the viral RayGun. But what I wanted to highlight was the misinformation going around that the Australian Breaking Association was founded by her husband who manipulated the selection process. It was all fake news!

• Noah Lyles winning the 100m by 0.005s, then getting covid and winning the bronze in the 200m (he has never lost a single 200m race since Tokyo in 2021), getting wheeled out in a wheelchair and back in the club a few days later.

• Quan Hongchan’s perfect dive with absolutely zero splash. She’s only 17 and won 2 golds in Paris. Also won 1 gold in Tokyo when she was 14. Wild. She also has a younger brother who makes absolute zero splash too.

• I realised how rage-baiting the Threads app is. Given it’s owned by Facebook I’m not surprised. It was full of Americans questioning why [insert non-American sport] are in the Olympics, when those sports are played by hundreds of millions of people in the east! Also bad takes all around. (I have uninstalled it again)

• The Chinese swim team including Pan Zhanle who broke the world record in the 100M Freestyle sparked controversy. Being on both western and eastern social media provides a very interesting perspective. On Twitter and in the Senate, Michael Phelps and the western world are highly skeptical of WADA “turning a blind eye” on Chinese doping. Yet, on Xiaohongshu (Chinese instagram), the posts have the tone of “oh yes, whenever China wins it’s doping but when the US wins it’s clean”, and “the Chinese swimmers were tested twice as much prior in the Olympics”. Whichever side you are on, I’d highly recommend reading the backstory around the Chinese doping scandal during Tokyo 2020.

• USA Basketball almost losing to Serbia in the Semis, down 17 points, but the King decided that he didn’t want to lose. Then in the finals, Steph Curry nailing every single three and putting France to sleep in the final minutes of the Finals. It was brutal..

• Also shout out to the French Women’s basketball who were down by 3 in the literal final second of the Finals, accidentally stepped on the 3-point line while sinking a buzzer-beating three, losing 66-67, one point away from a potential Gold.

• Sifan Hassan winning gold in the Women’s Marathon on Sunday after winning bronze in the 5K on Monday and bronze in 10K on Friday. All on the same week. Everyone who’s run an all-out 5K or 10K can tell you how draining it is to your body… and to only rest for <48h between the 10K and 42K? Madness.

• The two marathon legends 39-year-old Eliud Kipchoge (DNF) and 42-year-old Bekele (39th) in the Men’s Marathon. This is the end of an era!

• Carolina Marin bawling her eyes out during the Women’s Singles semi-finals as she ruptured a ligament on her right knee, knowing it’s probably a career-ending injury at this point. It was really hard to watch. She’s previously won Gold in Rio 2016, and missed the Tokyo Olympics due to an ACL injury.

• Of course there were way too many highlights this year… Vivian Kong (HK fencer) comes back from a 6-point deficit to win Gold, Djokovic finally winning an Olympic gold, Andy Murray never really liked tennis anyway, 41-year-old Cuban wrestler who’s won 5 straight Olympic medals (since 2008 Beijing Olympics!), the viral Korean and Turkish shooters, Italian woman boxer describes Algerian boxer’s punch being “the hardest punch she’s ever felt”, giving up after 47 seconds vs Imane Khelif who has previously failed “unspecified gender eligibility test” by the International Boxing Association, Remco Evenepoel shouting for a bike just 3.5km away from the finish line after riding 273km + the legendary finishing photo, Simone Biles’ usual 3 Gold 1 Silver brilliance, Leon Marchand’s 4 Gold 1 Bronze dominance in swimming, Tebogo winning Botswana’s first Gold in the Men’s 200m, Julien Alfred (Women’s 100m) with St Lucia’s first Gold, Duplantis (pole vault) shattering yet another world record, GBR’s Amos Yee comeback for Gold in the Triathlon, and the many medalists who have a freaking day job.. GBR’s Georgia Bell (bronze in 1500m) who works at a Cybersecurity firm and Olivia Coffey (US rower) who works in private equity.

And I have to share my fav vid of the Paris Olympics:

As with every Tings newsletter, I need to include something about my running journey to hold myself accountable.. In mid July, I hit my 10K goal of sub-55, improving my PB by 5mins at the Saucony London 10K. The vibes were immaculate and the PB now stands at 53:02.

I’m currently working towards a sub-50, which seems almost impossible at my current fitness level ngl. The next 10K race will likely be late Sep/early Oct, so wish me luck! Slowly working my way up the half and full marathon.

This is the first edition with no crypto, but there will be more crypto coming soon!

Subscribe now

Note: I’m using GnosisPay as an example here because it’s one of the more well-know self-custodial crypto debit cards.

I explained in detail how the GnosisPay card works in my previous article, but the TLDR of the user flow is that:

1. User sends 100 GBPe from their Metamask to their Safe wallet

2. User taps their GnosisPay card at Starbucks for a £3 coffee

3. Then, 3 GBPe is automatically transferred from their Safe wallet to GnosisPay’s address (0x4822521e6135cd2599199c83ea35179229a172ee)

This means that you can see every single “tap” on every GnosisPay card.

And more scarily, if you see someone using a GnosisPay card in front of you at Starbucks, you can work out their entire spending history and potentially investment history just by knowing either the exact time of the tx or the exact amount paid.

How to dox all GnosisPay users

To get my data source, I looked up the GnosisPay payments address on GnosisScan and downloaded the tx data as a csv. This shows me every single tx, the time, amount, source address and EUR/GBP amounts.

Here I show the top 10 spenders on GnosisPay. As an example, I will pick on the #3 top spender, Safe address “0x3b” who has spent ~€23K over the past 3 months.

Slapping this address onto your preferred blockchain stalking tool (I prefer Zerion), you can see every single time this person has tapped their GnosisPay card. Both the exact second (just look at the csv) and amount.

This isn’t even that bad.

You can then find out what this person used to fund their wallet, which is typically an EOA, aka a “metamask account”.

In this case, I found that user “0x3b” funds his account through this “0x17” EOA. He sent a €10 test tx before sending the full amount, €6,354.

Looking at “0x17”, it only has ~$300 of assets. Phew, he has good operational security. Privacy prevails!

But wait a minute… looking at tx history from last year, it seems like he always funds this wallet from another EOA “0x16” which holds ~$340K worth of crypto and an ENS…

And voila, now I can see his entire “investment” history all the way back to October 2021. And I know every single $DEGEN and $FRIEND trade he’s made and every single protocol he’s interacted with.

The thing is, this didn’t even take much effort. 15mins tops. He could be a millionaire and this is just a small shitcoin wallet, but if I were to pull a zachxbt I could probably find out this person’s real identity and first pet’s name.

I chose this example because he still remains pseudonymous. There was another address I chose which was attached to an ENS which gave me the person’s twitter, linkedin (real name), location (Spain), where he works, etc.

I did another exercise with a VC friend who told me he bought a coffee in the morning on a certain day with his card. I could trace his tx even though he didn’t tell me the exact amount or time, and even managed to dox his main wallet!

So what do we take away from this?

Takeaway one: Have better operational security.

You should either fund a fresh new EOA through an international CEX such as Binance or Monerium (the EURe/GBPe issuer), then send the EURe / GBPe to your Safe on Gnosis chain. GnosisPay also has “IBAN integration” on its roadmap so you can do it directly.

There are actually a few GnosisPay funder addresses I looked up which were clean and only funded with ETH and USDT through MEXC. In this case, I could only see the person’s full GnosisPay tx history (which is also pretty bad), but not their entire investment history or figure out their identity.

Takeaway two: GnosisPay should make it extremely clear to their users about the dangers of using a non-custodial crypto card. GnosisPay users are susceptible to a targeted $5 wrench attack, at least before GnosisPay transitions to a private L2.

Takeaway three: You need to run self-custodial payments on a blockchain with built-in privacy.

To their credit, GnosisPay announced in July last year that they will be transitioning to the GnosisPay zkEVM L2 based on Polygon zkEVM which will allow for private transactions.

Note that there are also other live solutions such as Railgun, zkBOB and Payy.link (payments app on their ZK-based L2). Also, given that GnosisPay users are already KYC’d, you could in theory make a compliant privacy pool out of them. Hence, many potential solutions out there.

I remain bullish on crypto payments, especially cross-border payments where money is exchanged across multiple Real Time Gross Settlement (“RTGS”) systems. This is an area I’m seeing many people building in, and I think will become one of the very few cash-generating crypto businesses that create real world value.

Shout out to @GoldenbergLior and @0xSkyMine for the idea!

Subscribe now

1. The consensus among investors seems to be a bull market until Q4 this year, though conviction was not as strong as in Dec/Jan. Hence now is an awkward timing for startup fundraising as token projects typically have a 6-12-month lock-up with 2-4 years of vesting, meaning investors would vest into a bear market. Some good investors are already preparing for the next cycle.

2. Private markets are still overheated — high volume of low quality deals, and good deals being priced expensively (we’re talking $100-500M FDV for pre-product protocols, especially in AI). Part of the reason is also because deals are getting benchmarked against some coins trading at high FDVs, without effectively pricing in lock-ups and illiquidity discounts. Regan wrote a great thread about this.

3. The (in)famous money flow and path to “alt-season” this cycle will be muted given BTC price movements were largely driven by the Bitcoin ETF. These Bitcoin ETF gains could flow to a future Ethereum ETF, but not to alts since you can’t buy your favourite alts through the same stock broker.

   

4. It seems like we went straight from Bitcoin to Solana memecoins and skipped everything in between. But it’s actually two separate things. The Bitcoin rally brought excitement to Ordinals, BRC-20’s and Runes on the retail side and Bitcoin “L2s” on the private side, while it was the Solana rally that brought attention to memecoins, starting a positive feedback loop. You should know that every VC that wants exposure to Solana already has exposure by now, whether that’s through open markets, funding Solana protocols, or through the FTX estate.

5. Memecoin supply has far outstripped the demand. It’s now too easy to deploy new tokens with pump.fun and there are just too many new memes gunning for attention. Unless you’re in the trenches sniping new memecoins, it’s much better r/r to long your favourite mid-to-large cap memes. And for the love of god, please unfollow influencers, you are just their exit liquidity. (Or follow them to see what coins to avoid).

   

6. The two biggest things on Ethereum right now are Eigenlayer and Ethena. The market loves anything that generates high yield and looks somewhat novel technically so in hindsight the hype that followed was a no-brainer. I kinda worry that there isn’t enough innovation on Ethereum.

   1. It remains to be seen if Eigenlayer ($EIGEN) can generating sustainable yield from the AVS. Even if we assume the tech runs smoothly, there is still a possibility that users will withdraw en masse after the airdrop period, when they realise they’re only getting a few % above “risk-free” staked ETH yield. I generally wouldn’t recommend farming Eigenlayer and LSTs if you’re not a whale, as they’re way too over-farmed.

   2. Ethena ($ENA) is just a tokenised hedge fund marketed as a “stablecoin”. It’s basically if Jane Street had a fund that does basis trading, tokenised it at $1 per share and let it trade without KYC. I think it’s a great crypto-native way to “tokenise a fund” and create a stablecoin not dependent on traditional banking rails, but the 60%+ yields were clearly not sustainable (seems like it’s dropped now to ~15%). I think it’s great we had Luna so everyone is watching any “stablecoin” very closely. IMO the key risk here is getting wrongfully liquidated on weird CEX wicks, even if they use Clearloop etc., but they do have good relationships with CEXs.

7. The other two most active ecosystems are Solana (memecoins) and Base (consumer apps). I think Solana has successfully graduated to become a religion, joining ETH and BTC. I really enjoy seeing cool stuff being built on Base like CryptoTheGame, Farcaster, and Fantasy.top. Blast and Berachain are hyped but seem too PvP to me. Other ecosystems I’m watching are Sui, Ton, and the aoComputer. (+ Monad whenever that launches)

8. There is this meme that in bull markets, FDV is a meme and Market Cap is king. This makes sense because inflows of capital buying is typically a lot higher than the token inflation rate. But if you’re buying into the typical “VC coin” that is low-float high FDV, be wary of unlocks + airdrop dumping. It’s really hard for markets to absorb unlocks, eg with the upcoming STRK, AEVO, and ARB hitting the markets 15-16 May.

   

9. “Tell me the incentives and I’ll tell you the outcome”. There is a huge incentive (mis)alignment problem as token vesting for team and investors is based on time (“1y cliff + 2y vest”) instead of milestone-based. This means there are many mercenary teams that (i) optimise for launching a token asap to start the vesting clock, (ii) optimise for KOL rounds and marketing instead of the product to increase token prices during unlocks, and (iii) are short-term oriented instead of creating long-lasting value and cashflows over a 5-10y horizon. I think this is the single biggest problem in crypto that needs to be fixed before we can go anywhere.

10. I think most “airdrop farming” is really negative EV unless you’re a whale, this includes Eigenlayer, LSTs, and the long list of Solana protocols (MarginFi, Drift, Kamino, etc.). The handful of protocols I think are worth playing around with (“farming”) now are the ones that are harder to “farm”, such as providing compute to decentralized inference protocols or using the aoComputer which requires using CLI. But more importantly, just go out and play with different protocols with no expectation of returns, that’s much better for your mental health lmao

11. Portfolio construction is one of the most important aspects but still seems overlooked. Assuming a sub a 5-6 fig portfolio, and the goal to 10x it, it makes more sense to have 3-4 concentrated bets (15-20% each) instead of a basket of 20 assets because it’s much easier to keep track and your size should accurately reflect your conviction. More importantly, if any one coin does a 10-15x then you double your portfolio (an amount I assume would be meaningful).

12. You should always hold a % of your portfolio in BTC or ETH. This % depends on the size of your portfolio — larger portfolio and more risk-averse = more BTC/ETH, say +50%. If you’re still feeling sidelined (what are you doing?!), just market buy ETH and deploy from ETH.

13. A good heuristic to “do I buy now?” is to compare the pain of a [50%] down move vs a potential [5-10x] up move. Replace the brackets with your numbers. Which one feels worse?

14. Embrace speculation. Most of crypto is just a huge casino because this is exactly what the underlying tech enables — crypto is the “internet of value” and people from across the globe can agree that the digital thing they are trading are the same and is provably scarce.

15. I am cautiously optimistic — the BTC ETFs are starting to see outflows, the ETH ETF is almost certainly delayed especially due to the Consensys v SEC lawsuit, and we’re still lacking a novel primitive for speculation. We had the ICO of 2017, DeFi yield farming of 2020 and NFTs of 2021, but nothing truly novel in 2024 — memecoins have been around since the $DOGE days. This is the bull run we don’t deserve. I am optimistically hoping something novel comes out of AI, Social, DePIN.

If you enjoyed this commentary, drop me a “like”, it’s cheaper than buying me a coffee 👀. Thanks for reading!

Subscribe now

1. You’ve probably seen the viral Tiktok of a janitor in blue jumpsuit pranking gym-goers by lifting heavy weights with his small build. He is actually Vladimir Shmondenko, aka “Anatoly” and has 13M followers on Instagram. He grew up in a small town in Ukraine with only 200-300 people and the story of how hard he’s worked to become successful is truly awe-inspiring.

2. I (almost) got pickpocketed in Paris, which got me thinking… Where is the perfect city in terms of (i) safety — prefer somewhere without rampant phone-snatching, (ii) weather — prefer somewhere that’s not hot and humid year round, (iii) English-speaking, so I don’t need to learn a new language. Obviously everything is relative and if you’re rich enough you can hide in your Beverly Hills mansion. What city would you put in the middle of this Venn diagram and what other criteria would you add?

   

3. I always get recommended restauranteur videos and watching this video about New York’s most popular fried chicken restaurant just reminds me how the best founders and investors have the same defining qualities as the best chefs and athletes. It’s the absolute fucking obsession over perfecting their craft. This Kobe video is one of my all-time favourites — You want first place? Come play with me. You want second place? Go somewhere else.

4. Traba is one of those startups that has an “Olympian” culture. Founder Mike Shebat always had a chip on his shoulder. He has a 996 work ethic — 9am-9pm six days a week, popular in China — and expects the same from all joiners. He always finds ways to achieve something — eg. he saw that Keith Rabois (Founders Fund) likes to go to Barry's. When he LinkedIn DM’d Keith, instead of asking him for coffee, he asked to do Barry's the next day and then go for brunch. It’s a long read, but I feel Traba is one of those businesses that intuitively makes sense, and will be a huge cash cow in the future.

6. Are there any billionaires that fit the criteria above? My dad used to tell me “don’t envy those rich people, their relationships with their wife(s) and kids are always very bad. Their families are always a mess”. For some reason that seems to be the case most of the time. Perhaps it takes a certain type of character to become a billionaire, and “moral people” tend to stop after they get a few $100m. (Side note on the article: obviously it’s hard to quantify “morals”, i.e. divorces might just be due to lifestyle differences)

7. The “hardest geezer” ran across Africa (south to north) from the southern tip of Cape Town, South Africa to the northern tip of Tunisia, running 16,500km and 1.5 marathons (63km) every single day through the Sahara desert, military zones, the Namibian desert, and rainforests. He’s pissed blood, robbed at gunpoint at Angola, almost got killed in the forest, and raised £200K for charity.

8. I went to Sui Basecamp in Paris to see if I was missing anything… and I don’t think so. Sui’s team, tech, and devrel are undoubtedly good. It’s just missing a “10x improvement” and a community. I like the zkLogin and zkSend features — I even managed to access my wallet via their app with my gmail without my seed phrase. I don’t think that’s enough though, as it’s quite copy-able to existing L1’s. My problem with Sui is that (i) it’s still the same ~10-ish projects on Sui since its mainnet, (ii) there still isn’t USDT or USDC, (iii) there still isn’t anything #OPOS (only possible on.. Sui). Things like Movement Labs also don’t really help Sui.

9. Token2049 is one of those events I have a love-hate relationship with — it’s probably the noisiest conference with the most scam projects, but it’s also where everyone congregates so it’s easy to meet people from all around the world. Despite the flood, the energy was really great. The key themes there seemed to be AI, Memecoins, Bitcoin DeFi and RWAs.

10. Why The Winners Won. Sometimes we just look back and say Lido, Uniswap and dYdX became successful just because they were first movers, but this glosses over a lot of details. Uniswap wasn’t the first DEX, EtherDelta was; Uniswap won on distribution. Lido also wasn’t the only liquid staking token in town and wasn’t even the only token called “stETH”, but they had a strong bias to action. dYdX’s first product was Expo, but they iterated quickly based on user feedback.

11. There’s no podcast as transparent as Moneywise about how rich people spend their money. How much money do these hundred 9-fig exited entrepreneurs spend? How much money do they think is enough to retire? Do you actually just get a bank transfer of $100M when you sell your company? Spoiler: On average $80K pm, but some $5K pm and some $250K (of which $150K pm are donations). At $80K pm it’s $25M liquid net worth. And, yes.

12. You know there needs to be a running update section now — I ran a 58:20 for my maiden 10K, smashing my sub-60 goal. I’m actually most proud of the fact that I managed to maintain a steady pace throughout. The next goal is to break 55min and work my way up to a half marathon!

If you enjoyed reading this issue, drop me a “like”, I need that ego boost 👀. Thanks for reading!

Subscribe now