Fact check · trading agents · checked 2026-10-01
Did a fired Citadel quant make $430,000 on Polymarket with Claude Fable 5 in 48 hours?
What was claimed?
“Citadel fired their best quant. He rebuilt their entire algo with Claude Fable 5 in 48 hours - and he's up $430,000 trading it against them. He didn't take a single file. He didn't need to - ten years of that logic lived in his head, and you can't raid a memory. Wallet proof: http://polymarket.com/@0xb55fa1296e6ec55d0ce53d93b9237389f11764d4-1777575277609?via=cvxv666”
In numbers: a gain of $430,000; an algorithm rebuilt “in 48 hours” with Claude Fable 5; a trader described as Citadel’s “best quant”, fired; a Polymarket wallet given as “proof”. No starting amount and no dates. The wallet link carries a referral code and the post also links a copy-trading page under the same code — the claim is also a funnel.
The Community Note on the post reads: “Claude Fable 5 was released on June 9, 2026, hours before this post, making it impossible to have rebuilt an algorithm with it in 48 hours. No reports confirm any Citadel quant recreating their trading algorithm this way.” We checked the timing ourselves rather than rely on it.
What did we check, and how?
The post names the wallet, so the record is public: polymarket.com/profile/0xb55fa1296e6ec55d0ce53d93b9237389f11764d4, a profile with no username (the page shows the address itself). We pulled its first recorded activity, its deposits and withdrawals as Polymarket classifies them, its current open positions, and its daily cumulative profit-and-loss from Polymarket’s PnL API. We read the post’s exact time and the announcement’s exact time from the two post IDs. Probe calls made 2026-10-01 (times in the recipe).
What does the record show?
| Question | The claim says | The public record says |
|---|---|---|
| How much | “up $430,000” | Cumulative profit-and-loss $419,109.12 on 2026-06-09 (post day) and $439,931.22 on 2026-06-10. The figure holds. |
| When the trading happened | After an algorithm rebuilt “in 48 hours” with Claude Fable 5 | First trade 2026-04-30 19:08 UTC (TRADE). Cumulative profit-and-loss $165,797.28 on 2026-05-15, $338,138.40 on 2026-06-01, $408,475.22 on 2026-06-07. Claude Fable 5 was announced 2026-06-09 17:08 UTC; the post followed at 20:29 UTC. 95% of the claimed gain existed before the model did. |
| Starting money | Not stated | Deposits recorded by Polymarket: 7 rows, $1,861.07, all on or after 2026-06-21 — after the gain. The initial funding does not appear as a deposit row (direct transfers are not classified as deposits), so the starting sum is not readable from this API. |
| How it traded | “thirty-two thousand times”; “71% of trades land green”; “Eighteen thousand is mathematics” | Through the post: 864,304 fills across 36,427 markets, $18,260,602.39 traded, 0 sells; 37% of fills in 5-minute markets, 36% in 15-minute. Lifetime: 3,409,177 fills, 159,007 markets, $63,406,612.39 traded; both outcomes bought in 80% of markets. A per-trade win rate is not recoverable from the ledger; Polymarket’s daily series shows 21 down days out of 154, the worst −$10,378.38 on 2026-06-07. |
| Who | Citadel’s “best quant”, fired | Nothing in a ledger identifies a person or an employer. Unverifiable by nature. |
| Since the post | — | Still trading at fetch time (2026-10-01 01:21 UTC (TRADE)). Cumulative profit-and-loss $559,329.60 on 2026-07-01, $1,039,494.75 on 2026-09-01, $1,248,150.60 on 2026-10-01. One withdrawal: $100,000.00 on 2026-08-07. Open positions $5,240.26. |
Profile page as displayed 2026-10-01: “Joined Apr 2026”, 152,944 predictions, “Total PnL +$1.2M”, “Total Volume 131M”, biggest win $6,737.67.
What does this prove, and what does it not?
- It proves the wallet is real and was up roughly $430,000 when the post said so; and that the gain was built over about forty days from 2026-04-30, with 95% of it in place before Claude Fable 5 was announced. An algorithm “rebuilt with Claude Fable 5 in 48 hours” cannot be the source of money earned before the model existed.
- It does not prove who runs the wallet, what software (if any) places its orders, or whether anyone was fired from anywhere. Those parts of the story have no public record and the status says so.
- Context the reader should have: the same account posted a different wallet with a different back-story eleven days earlier (a “risk quant in Singapore”, “$400 … was $479,401”), under the same referral code. We did not check that wallet for this page.
- What would change the status: a dated, checkable record of the algorithm’s change on or after 2026-06-09 that explains the wallet’s later gains — the post-announcement profit is real and large — or evidence tying a person to the wallet. We would re-run the recipe and amend this page visibly.
Recompute it yourself
- Open the wallet link in the post (polymarket.com/@0xb55f…-1777575277609) or polymarket.com/profile/0xb55fa1296e6ec55d0ce53d93b9237389f11764d4; the address is the proxy wallet.
- GET https://data-api.polymarket.com/activity?user=0xb55fa1296e6ec55d0ce53d93b9237389f11764d4&limit=1&sortDirection=ASC&excludeDepositsWithdrawals=false — the first row is the account's first activity (2026-04-30 19:08 UTC).
- GET …/activity?user=<address>&type=DEPOSIT,WITHDRAWAL&excludeDepositsWithdrawals=false&limit=500&sortDirection=ASC — Polymarket's own deposit and withdrawal rows.
- GET https://user-pnl-api.polymarket.com/user-pnl?user_address=<address>&interval=all&fidelity=1d — cumulative profit-and-loss per day; read the values for 2026-06-07 through 2026-06-11.
- The post's own timestamp is in its ID: (2064444808996024829 >> 22) + 1288834974657 ms = 2026-06-09 20:29:32 UTC; the quoted Claude Fable 5 announcement, 2064394146916229443, resolves to 2026-06-09 17:08:13 UTC.
Statuses used on this site are defined on the fact-check index. This page
carries ClaimReview structured data that matches its text.