Kevin Zhou + Galois Capital
Saw Through UST 5 Months Early, Only Made $15M. Here's Why

On the morning of May 7, 2022, UST visibly depegged for the first time — from $1 to $0.985. Just a 1.5% deviation. Plenty of people in the market said "the peg mechanism will pull it back." One week later, the entire Terra ecosystem was $50 billion in vapor.

But 5 months before that, someone was already warning about it publicly. His name is Kevin Zhou, co-founder of Galois Capital. He didn't guess. He saw the structure. This article is the full breakdown of that trade — including what he actually made (not the "hundreds of millions" everyone assumes) and why he deliberately capped position size at 10% of AUM.

1. From Shanghai to Berkeley — The Math Background

Kevin Zhou was born in Shanghai and moved to the US with his parents at age 3. His father went to the University of Hawaii for a PhD in math; his mother got a master's in electrical engineering. He grew up in Hawaii, and was a math prodigy — Johns Hopkins Talent Search, #1 in Hawaii and top 120 nationally.

He majored in math and economics in college (graduated 2009), got an economics master's at Boston University, then went to UC Berkeley for an MFE (Master in Financial Engineering). That MFE cohort is where he met his future co-founder — a person whose name has never publicly surfaced. Every outlet just describes him as "a co-founder met in financial engineering school."

His first job after the 2009 financial crisis was as a back-office quant at a credit rating agency. Boring. In 2013 he jumped to Buttercoin, a Y Combinator-incubated crypto exchange, as a market maker. The company collapsed quickly. In 2014-15 he joined Kraken, helped them build the OTC desk, and survived the brutal 2014-15 bear. Those two stints are where he actually learned what the market looks like.

In 2017 he left Kraken and co-founded Galois Capital with that math-background partner, raising about $5.5M to start. Positioned as a market-neutral hedge fund — OTC, basis arbitrage, yield farming. No directional bets. AUM peaked around $200M.

2. The Guy Who Was Warning Publicly in Late 2021

On the Unchained podcast, Zhou left this line:

In plain English: between December 2021 and January 2022, he was already publicly asking "this thing is a top 10 coin, why hasn't it already unwound?" Bloomberg independently corroborated this: "began publicly warning about Terra in early 2022" and "started shorting Luna in early May."

He published a piece on the Galois Medium titled "State of the Market"(February 19, 2022) where he compared DeFi's "composability" directly to the CDO-squared structured products from before the 2008 crisis. At the time, most people were still farming the 20% Anchor yield.

His specific objections:

• Anchor offered ~20% APY on UST deposits — that yield was subsidized by the Terraform Labs / LFG reserves. Income was way below expenses. It was a burn mechanism.
• The mint/burn between LUNA and UST is reflexive — LUNA falling makes UST depeg pressure worse, and vice versa.
• LFG used Bitcoin as a reserve (cumulatively up to 80,394 BTC). Under peg pressure, that reserve flips from a natural bid into forced sells, amplifying downside risk in BTC itself.

In multiple interviews he uses one analogy: "LUNA is a perpetual motion machine"— looks like it can spin forever, but it needs constant energy input, and once the energy is cut, it stops.

3. The Trade — A Three-Stage Position Structure

Most people assume Zhou just shorted LUNA. The real position had three layers:

This structure has a name: "pick up nickels then short the steamroller" — collect the pennies until you can't anymore, then turn around and short the steamroller. Zhou is explicit in interviews: he wasn't a "die-hard short" the whole time. He captured the yield while the mechanism still worked, and the moment the first crack appeared (the 30 bps depeg), he flipped sides.

4. May 7-12 — The Five-Day Timeline of $50 Billion Vaporizing

The core data from the collapse:

One thing worth noting: the LFG BTC dump on May 9-10 — 80,394 BTC → 313 BTC, nearly the entire reserve liquidated overnight — dragged the broader crypto market down with it. Even BTC holders with no Terra exposure took major losses. This was the live demonstration of what Zhou had warned about in February 2022: "reflexivity will spread into BTC itself."

5. Five Signals Any Retail Trader Could Have Monitored

The reason this collapse was "predictable" is that every signal was public on-chain data. All five of the following were visible in real time. Most people just weren't looking:

Put those five together and you have an early-warning dashboard for "algorithmic stablecoin + subsidized yield" structures. The same framework still applies today: the USDe / sUSDe funding spread, the DAI / sDAI rate gap, any structure with "abnormally high stable yield + reflexive collateral"deserves the same monitoring setup.

6. What Happened to Galois Next — FTX, and the SEC Settlement

The UST trade is the brightest moment in Galois's public history. Six months later, the entire fund got dragged down by the FTX collapse:

• November 2022: Zhou informed investors that roughly half the fund's assets ($40M) were frozen on FTX.
• February 20, 2023: Publicly announced the wind-down of the flagship fund. Investors got back 90% of the non-FTX portion + the FTX bankruptcy claims later sold for roughly 16 cents on the dollar.
• September 3, 2024: SEC fined Galois Capital Management for violating the Custody Rule — assets weren't held with a qualified custodian (FTX wasn't one), and the redemption notice timing misled investors. $225,000 civil penalty, all paid to harmed investors.
• 2024-2026: Zhou shifted to early-stage investing / angel deals, and stated he won't be opening a new fund anytime soon (it would "unfairly reset the high-water mark").

7. Three Key Takeaways

1. Right direction + intentionally small beats "all-in once" every time. $200M AUM × 10% cap = $20M of risk exposure, $15M of P&L. The point isn't "he could have made more." The point is that discipline kept him alive after UST (even if FTX eventually got him).
2. Collapses in reflexive structures can be flagged early. UST went from a 30 bps deviation to $0.22 in 6 days. All five on-chain indicators were public. Any one of them crossing a threshold should have been a reason to reassess the position. The same framework transfers to USDe, the next algo stablecoin, the next "high-yield pool" — same playbook.
3. Even if your research is right, custody risk can still wipe you. When FTX collapsed, half of Galois's assets were locked up. This had nothing to do with "the algorithm being right" — the underlying assumption about centralized custody was broken. Non-custodial isn't a moral stance. It's structural risk control.

This article is not investment advice. All numbers cited come from public interviews, SEC filings, on-chain data, and media reports — full source list available on request after the RelatedPosts section below. Crypto trading carries high risk. Please assess your own risk tolerance.