The Exit Machine

I seriously researched running a fund. The mechanics are well understood, the playbook is legible, and there's real money in it. Then I looked at how the markets actually operate. The structure produces the same outcome regardless of intent.

Venture funds deploy capital early via SAFTs and token warrants. The thesis is rarely the product. It's the launch event. The token is the exit.

In March 2023, Blockchain Capital led a $50m Series A into EigenLayer, with a16z, Polychain, and others participating, at roughly $500m implied fully diluted valuation. Fourteen months later, EIGEN launched at a peak FDV of $7bn. On paper, a 14x return. The lock-up made this look like a long-term bet. Eighteen month linear vesting. Investors aligned with the project. Everyone in the same boat.

A SAFT restricts token transfers. That's smart contract logic, not business logic. The underlying economic exposure is still just paperwork. And paperwork has always moved freely between funds, between desks, between anyone willing to sign.

A collection of funds puts $5m into a $25m seed round. The expectation, loose but shared, is a $100m follow-on before TGE. At that follow-on, the position has 4x'd on paper. The funds OTC away enough to recover their initial $5m — say 20% of their allocation — retaining 80%. Their remaining position is now free carry. They are not exposed. They are not aligned. They have recovered capital and hold optionality. The lock-up still shows them as committed holders. On-chain, nothing has moved.

Whales Market processed over $300m in pre-launch OTC volume in 2024. Aevo ran perpetuals on EIGEN before the token was tradeable. Pre-market OTC on EIGEN cleared at roughly 40–50 cents on the dollar relative to expected listing price in the weeks before TGE — that discount is the price of liquidity, paid willingly by someone who needs to be out before the lock-up technically expires.

EIGEN is now down 88% from peak. The retail buyer who purchased at TGE is sitting on that loss. I don't think these are separate events.

Worldcoin raised $240m in total — a16z led a $115m Series C in May 2023 at roughly $3bn implied valuation. WLD launched in July 2023 with approximately 1% circulating supply against a peak FDV of $12bn. A circulating market cap of around $120m against a fully diluted valuation of $12bn is not a price. It's a forecast retail is being asked to fund. WLD is now down roughly 90% from peak. The unlock schedule is still running.

In May 2024, Binance Research published "Low Float & High FDV: How Did We Get Here?" — one of the more precise public analyses of the problem. Of the major 2024 token launches surveyed, most launched with under 15% circulating supply against multi-billion FDV. The report noted that tokens in this cohort had underperformed both Bitcoin and the broader market significantly in the months following TGE.

The FDV to circulating ratio is the tell — but the mechanism is more precise than it looks. When 5% of supply is circulating, price discovery is happening on negligible volume. The FDV number that results — $7bn, $12bn — is not a cleared market price. It's a single marginal transaction extrapolated across the entire theoretical supply. It's mathematically fictitious at inception. The overhang isn't just large. The starting price was never real.

Bitcoin was up significantly over the same period. That's not proof of mechanism. Correlation isn't causation, and a trader will tell you so immediately. But it does remove the most obvious alternative explanation. A broad crypto bear market would account for the losses. A bull market running simultaneously makes supply the more parsimonious explanation. The pattern isn't definitive. It's consistent.

I find it hard to believe anyone constructing these tokenomics is unaware of what that ratio implies.

The precedent is cleaner than most people in crypto want to acknowledge. Between 2019 and 2021, SPACs raised over $160bn in US markets. Sponsors received founder shares at near-zero cost, listed a blank-cheque vehicle publicly, and had 24 months to complete a merger. Retail bought in believing they were getting institutional deal access. In 2022, Michael Klausner, Michael Ohlrogge, and Emily Ruan published "A Sober Look at SPACs" in the Yale Journal on Regulation. Their finding: of SPACs completing mergers between January 2019 and June 2020, sponsors had extracted so much in fees and dilution that the SPAC held a mean of just $4.10 — median $5.70 — in net cash per share at the time of merger, from $10 originally raised. The vehicle was legal. Disclosure was technically adequate. The structure transferred wealth from public investors to sponsors with mechanical consistency. The SEC moved on SPACs in 2022. Volume collapsed within twelve months.

Variable Prepaid Forward contracts allow corporate insiders to pledge pre-IPO shares as collateral, receive cash upfront, and technically retain ownership through lock-up expiry. Functionally identical to selling. Doesn't appear as a sale in public disclosure until settlement. Legal, documented, used routinely. Pre-market token OTC is the same instrument with less regulatory infrastructure around it.

The lock-up was supposed to be the mechanism that made token launches different from this. I'm not sure it is.

Retail capital that survives the initial dump doesn't leave. It rotates. Into the next narrative, the next points programme, the next protocol accumulating TVL from locked capital that makes the next Series A look reasonable. Each rotation is the next project's adoption metric. The metric justifies the round. The round produces the next TGE. The cycle requires new believers at every step, and as far as I can tell, the structure manufactures them.

Pump.fun is this made visible. Since launching on Solana in early 2024, it has generated approaching $1bn in cumulative protocol fees. No product, no team attribution, no pretence of long-term value. The token is pure speculation, undisguised. Capital flows through it in minutes rather than months. Everything else in the cycle is this with more steps and a better story.

None of this is technically a fraud. The on-chain data is public. The vesting schedules are disclosed. The pre-market instruments are available to anyone who knows they exist. Charles Ponzi promised fixed returns. Bernie Madoff fabricated records.

What I do think is that the structure creates conditions where the returns at the top require continuous inflows at the bottom to exist. Because there's no other way for the numbers to work.

The honest version of what the bear cycle reveals is that the floor price — the price at which a token stabilises when sentiment normalises and inflows stop — is probably the closest thing to intrinsic value the market produces. EigenLayer's restaking primitive is genuine infrastructure. Worldcoin's biometric identity approach is genuinely novel. The technology is real. But without continuous capital inflows, the token price has no mechanism to reflect that value. It's speculative in exactly the way a pre-revenue IPO is speculative — except that loss-making IPOs operate under tight controls around insider trading, lock-up enforceability, and disclosure of material interests. The S-1 exists precisely to close the information asymmetry that pre-market OTC exploits.

In a traditional venture-backed company, the fund absorbs the risk of being wrong. The bet is illiquid, the timeline is long, and the exit requires the company to perform. In crypto, that risk has been transferred to the public. The fund's exposure is retired early. The public buys the narrative at peak. The projects continue, or they fade, and the people left holding find that the floor was always a long way down from where they entered.

The SPAC market required regulatory intervention. Pre-market OTC in crypto operates in a gap that hasn't been addressed. The strategy is legible enough now that I'm not sure even the best actors can ignore it and remain competitive.