Fact check · trading agents · checked 2026-10-01
Did a “Clawdbot” make $150,000 from 8,894 Polymarket trades?
What was claimed?
“🚨 BREAKING… the fastest 5m Polymarket Clawdbot setup just leaked. This is a REAL framework with REAL results. … He started with a modest base, let Clawdbot manage execution, and scaled it to nearly ~$150K in returns … Profile - polymarket.com/@PBot1 … 5-minute BTC & ETH micro-arbitrage … no predictions, no directional exposure, no bias … 8,894 trades. Individually insignificant. Collectively, they compounded into just ~$150K in gains”
The next morning @rohanpaul_ai carried it to a wider audience (about 983.5K views): “Clawdbot setup is fuelling a quiet Polymarket automation race. It exploits brief glitches in 5-minute BTC/ETH contracts where Yes+No drops below $1, buys both for risk-free arbitrage. Ran nonstop, made 8,894 trades, earned ~$150K without human input.” On 2026-02-21 CoinDesk wrote that “A fully automated trading bot executed 8,894 trades on short-term crypto prediction contracts and reportedly generated nearly $150,000 without human intervention,” citing “a recent post circulating on X.” Neither the amplifier nor the article named the account.
In numbers: ~$150,000 gained, 8,894 trades, 5-minute Bitcoin and Ether markets, both sides bought when “Yes” plus “No” priced under $1, “no directional exposure”, no stated starting amount, no stated dates. “Clawdbot” (also “Moltbot”) is the name the post gives the software.
What did we check, and how?
The original post links a Polymarket profile, @PBot1. That link now returns a 404, because the account
was renamed: the wallet behind it is still live as polymarket.com/@pbot-1. We tied the two
together through the Internet Archive’s snapshot of @PBot1 from 2026-03-02,
whose page source carries the wallet address 0x88f46b9e5d86b4fb85be55ab0ec4004264b9d4db — the same address Polymarket’s API returns today
under the name “PBot-1”.
We pulled the wallet’s complete activity from Polymarket’s data API — 916,827 rows: 881,482 trades, 34,875 payouts, 21 deposits, 46 withdrawals and the rest rebates and merges — and its daily cumulative profit-and-loss from Polymarket’s PnL API. The claim gives no dates, so we read it as “everything up to the moment of the post”, 2026-02-15 14:15 UTC. Fetched 2026-10-01 00:46:35 UTC.
What does the record show?
| Question | The claim says | The public record says (through 2026-02-15 14:15 UTC) |
|---|---|---|
| How many trades | 8,894 | 8,898 distinct markets traded — Polymarket’s “Predictions” counter. Those markets took 133,810 separate fills. The claim’s word “trades” matches the markets figure, four apart. |
| How much gained | “nearly ~$150K” | Polymarket’s cumulative profit-and-loss: $142,694.14 on 2026-02-14, $147,827.30 on 2026-02-15, $149,776.86 on 02-16. |
| Starting money | “a modest base” | First deposit $50.00 on 2025-12-17 16:44 UTC; $3,000.00 25 minutes later; 14 deposits totalling $34,769.08 by the post. Withdrawals by the post: 22, totalling $148,434.00. |
| Which markets | “5-minute BTC & ETH” | 7,749 of the 8,898 markets (87%) were 15-minute “Up or Down” markets; 598 (7%) were 5-minute; 551 were hourly. By fills: Bitcoin 86%, Ethereum 12%, Solana 2%, XRP under 1%. |
| How it traded | Buys both sides when Yes + No < $1; “no directional exposure” | In 3,773 markets (42%) the account bought both outcomes; in those, the two sides’ average prices summed to a median $0.98, under $1 in 80% of them. The other 58% of markets were bought on one side only. 2,747 fills were sells. |
| Since the post | — | Trading continued to 2026-07-14 16:28 UTC; cumulative profit-and-loss $183,298.25 on 2026-03-02, $234,804.31 on 2026-06-01, and $235,476.42 since 2026-07-15. Lifetime: 21 deposits $87,264.08, 46 withdrawals $322,678.37, open positions now $0.00. |
Profile page as displayed 2026-10-01: “Joined Dec 2025”, 43,810 predictions, “Total PnL +$235.3K”,
“Total Volume 21.1M”. Summing usdcSize over every TRADE row gives $10,115,651.41;
Polymarket’s displayed volume follows its own definition.
What does this prove, and what does it not?
- It proves that the account linked from the post is real, and that on the day of the post its public record carried the two numbers the post gave: about 8,900 markets and about $148,000 of profit. The profit had largely been withdrawn.
- It also shows what the post left out: the money came in as $34,769 of deposits over two months, not a token sum; most markets were 15-minute, not 5-minute; and more than half the trades took one side, which is directional exposure by definition.
- It does not prove that “Clawdbot” or any software placed the orders, that it ran “without human input”, or that the strategy is “risk-free”. A ledger shows an account, not a hand. The Telegram “copytrade” channel linked from the post is a separate product; we did not test it.
- What would change the status: evidence that the wallet at
0x88f46b9e5d86b4fb85be55ab0ec4004264b9d4dbis not the account the post linked (the Internet Archive snapshot is the tie), or a different reading of the post’s numbers with a stated date range. We would re-run the recipe and amend this page visibly.
How did the claim change as it spread?
- The original post names the account; the amplifying post and the CoinDesk article do not, so readers of those could not check.
- “Nearly ~$150K in returns” became “earned ~$150K without human input” and then “reportedly generated nearly $150,000 without human intervention” — the software’s autonomy hardened with each retelling; the number stayed.
- The profile link carried a referral code (
?via=…) and the post a copy-trading channel: the claim was also a funnel.
Recompute it yourself
- Open the Internet Archive snapshot of polymarket.com/@PBot1 from 2026-03-02 (link in Sources); the 0x address repeated in its source is the proxy wallet. Confirm it against polymarket.com/@pbot-1 today.
- GET https://data-api.polymarket.com/activity?user=0x88f46b9e5d86b4fb85be55ab0ec4004264b9d4db&limit=500&offset=<n>&start=1&excludeDepositsWithdrawals=false&sortBy=TIMESTAMP&sortDirection=ASC — page by offset; when the API refuses a high offset, restart with start=<last timestamp seen> and drop duplicate rows. Expect about 917,000 rows.
- GET https://user-pnl-api.polymarket.com/user-pnl?user_address=0x88f46b9e5d86b4fb85be55ab0ec4004264b9d4db&interval=all&fidelity=1d — Polymarket’s cumulative profit-and-loss per day.
- Cut at the post’s timestamp (2026-02-15 14:15:34 UTC). Markets = distinct conditionId among TRADE rows; fills = TRADE rows; deposits and withdrawals by type. Market length is read from each title’s time window (“12:30PM-12:45PM ET” = 15-minute).
- Both-sides test: per market, sum usdcSize and size of BUY rows for outcome 0 and outcome 1; a market counts as both-sided when both have shares; the price sum is (usdc0/shares0) + (usdc1/shares1).
Statuses used on this site are defined on the fact-check index. This page
carries ClaimReview structured data that matches its text.