AYC Fund trades digital assets with rules-based AI - but the money stays in your own exchange account. Inside a one-person Y Combinator fund betting that custody is crypto's real problem.
Most crypto funds begin the relationship by asking for your money. AYC Fund begins by asking for a trading key - and only a trading key. Your coins stay in your own exchange account. The fund can place trades, but it cannot press withdraw. In an industry that has lost fortunes to custodians who took the deposits and disappeared, that single design choice is close to the whole pitch.
AYC Fund is an AI-native crypto hedge fund that came out of Y Combinator's Winter 2022 batch. It was started in 2021 by MJ Kim and runs out of Seoul. On paper it is tiny - the kind of team you could fit in a taxi - and yet it competes in a market where the counterparties are professional trading desks that never close. What it sells is not a hunch or a hot tip. It sells a system: software that scans digital-asset markets, generates rules-based long and short signals, and executes them without a human hesitating over the button.
How the money moves - or rather, doesn't. Custody never changes hands. The fund holds a key that can trade but cannot cash out.
Ask anyone who trusted an exchange that no longer exists: the moment your assets leave your control, you are betting on the honesty and solvency of a stranger. AYC's answer is almost boring, which is what makes it convincing. Clients connect through API keys with trading-only permissions. The keys are encrypted before storage. The fund can open and close positions inside your account, but the withdrawal switch stays off. If everything went wrong tomorrow, the worst case does not include someone walking out the door with your balance.
This is the plumbing beneath what the firm calls its V11.25 Safety Engine. The version number is the tell. Eleven-plus iterations of pre-liquidation protection and daily loss limits suggest a team that spends more time thinking about how a strategy fails than how it wins. In trading, that ratio is usually the difference between a fund that survives a bad month and one that becomes a cautionary thread on a forum.
It also quietly reframes what the client is buying. A lot of trading products sell the upside and bury the risk controls in a footnote. AYC does the reverse. The public materials say very little about specific returns and a great deal about permissions, encryption, and loss limits. That is an odd choice for a marketing site, and a revealing one. It reads like a firm that expects to be judged by sophisticated people who have seen enough blowups to care more about the guardrails than the top-line number.
Your funds stay in your exchange account.AYC Fund - product principle
Claims are cheap in crypto, so it helps that some of AYC's are checkable. The firm entered the Bybit AI Showdown - a live contest against professional traders - and finished first, posting a gain of 14.82% over 21 days. Separately, its AYC VWAP strategy climbed to the top of the six-month return rankings on the Cornix marketplace, with a publicly verified result of +91.06% and a low-to-medium risk rating. VWAP, for the uninitiated, stands for volume-weighted average price: a way of anchoring trades to where the real volume is trading rather than to a single flickering number on a screen.
A second product, AYC ALGO-100, widens the net. It scans the top 100 trading pairs in real time and produces pure long and short signals without human input. The philosophy across both tools is the same. No hero trades. No revenge trades at three in the morning. In a market that runs 24 hours a day, the advantage of a machine is not raw intelligence - it is that it does not get tired, bored, or scared. Discipline is the edge. The bot just enforces it.
A verified track record is worth pausing on, because crypto has trained everyone to squint at screenshots. Cornix results and exchange contests are not audited financial statements, but they are harder to fake than a testimonial - the platform is holding the score, not the fund. That is the useful thing about competing in public: the leaderboard does not care about the pitch deck. It records what happened, in a market the trader did not get to pick, over a window the trader did not get to choose.
Verified where it can be. Figures reflect publicly posted marketplace and contest results; bar widths are illustrative, not to scale against each other.
Here is the part most startups would never do on purpose: AYC Fund made its public site smaller. The self-serve sign-up flows, the return tables, the buttons that let anyone with a wallet start trading - most of that is gone. What remains is a review gate. The public pages now describe a private, systematic program and ask qualified clients to request access.
The audience shifted with it. Where the firm once courted retail traders through a bot marketplace, it now speaks to high-net-worth individuals, single-family offices, and the advisors who sit between them and their money. Access follows a suitability and operational review. Detailed risk, structure, and reporting materials are shared privately, and the site is careful to repeat that its public content is informational, not an offer to sell anything.
AYC Fund is now a VIP-only public site.From the company's own repositioning
It is a hard pivot to pull off - from many small self-serve users to a few carefully vetted relationships. But it tells you something about where the operator thinks the durable business is. Fewer, better-fit clients you can actually serve beats a crowd you cannot. For founders watching from the outside, the lesson is portable: your first traction channel does not have to be your business model. The marketplace was the proving ground. The fund is the plan.
Before the doors narrowed, AYC published a fee structure worth studying regardless of whether you ever wire it a dollar. It charged a small one-time management fee - around 2% or a fixed amount - and a performance fee, roughly 15 to 30%, collected only when the fund actually made money. Withdrawals opened after a three-month lock-up, during a weekly window. Assets ran on Bybit, Binance, Bitget, and OKX.
| Term | What it meant |
|---|---|
| Management fee | ~2% or a fixed amount, one-time and upfront |
| Performance fee | ~15-30%, charged only on profits |
| Custody | None - funds stay in your exchange account |
| Access | Trade-only API keys, encrypted, no withdrawal rights |
| Lock-up | 3 months, then a weekly withdrawal window |
| Exchanges | Bybit, Binance, Bitget, OKX |
The performance-only fee is the interesting part. It aligns incentives in the plainest way possible: the fund eats first only after the client does. That structure is not unheard of in traditional asset management, but it remains rare enough in crypto that its presence says something about how much confidence an operator has in their own returns. Under the current VIP model, terms like these are no longer posted publicly - they are shared only through definitive documentation after a fit review.
Zoom out and AYC Fund is standing at a crossroads of two crowds. On one side are the automated-trading and copy-trading tools - the bots you find on platforms like Cornix, 3Commas, and Pionex - which are cheap, self-serve, and largely leave you on your own. On the other side are traditional crypto hedge funds and digital-asset managers, which offer a human relationship and institutional wrapping but usually want custody and a much larger check. AYC is trying to occupy the seam between them: the systematic engine and verified numbers of the first group, delivered with the private, vetted relationship of the second, minus the custody.
There is one more thread that hints at a wider ambition. AYC also describes itself as an early-stage Web3 venture fund, backing token projects and helping them expand into Korea and Southeast Asia. The bet underneath is a familiar one for any fund working across borders: in a market you do not understand, capital plus a local introduction is worth more than capital alone. Seoul is not an accident of geography here - it is a bridge the firm is offering to founders trying to reach Asian users and liquidity.
Those two activities - trading managed accounts and backing token projects - are not as unrelated as they first look. Both depend on reading the same markets closely, and both put the firm in a position to see which projects have real liquidity and which are decoration. A quant desk that also writes venture checks has an unusual vantage point: it watches order flow all day and can tell, faster than most, when a token's activity is genuine. Whether AYC can turn that vantage into an edge on the venture side is unproven, but the logic of pairing the two is sound.
AI-Native Engine. Institutional Grade.AYC Fund - headline positioning
Whether the seam holds is the open question. A one-person fund is efficient, but it also concentrates the strategy, the risk framework, and the client relationships in a very small number of hands. The right question for any allocator is not whether the software works - the contest and the leaderboard suggest it does something - but what the team does that the software does not, and how much of that survives a market the backtests never saw. AYC's answer, at least in how it presents itself, is that the durable product is the risk engine and the review process, not any single winning streak.
For now, AYC Fund is a compact, verifiable, and unusually disciplined entry in a category that has burned a lot of people. It won a contest against pros. It posted numbers you can check. And it built its whole model around a promise most of its industry cannot make: whatever else happens, it never gets to walk away with your coins. In crypto, that is a smaller claim than the marketing usually reaches for, and a more honest one.