A new academic working paper has put the booming prediction market industry on notice. Researchers from Stanford University and Singapore Management University — David Dai, Ruizhe Jia, and Shihao Yu, in a study titled “Settlement Manipulation in Prediction Markets” — found trading patterns strongly suggesting that sophisticated participants manipulated Bitcoin’s spot price to win Polymarket’s popular five-minute prediction contracts, pocketing millions largely at retail traders’ expense.
The Exploit: One Price, One Second, One Nudge
The vulnerability lies in the contract design. Polymarket’s five-minute markets ask a binary question — will Bitcoin finish above or below a set threshold when the window closes? — and settle against a Chainlink price feed referencing Bitcoin’s spot price at that exact moment. Because the entire payoff hinges on a single timestamp, anyone holding a large enough position has an incentive to shove the spot market in their favor during the closing seconds. Co-author Shihao Yu described it as a structural flaw: the contracts settle on a price that traders can themselves move by trading the underlying asset.
Analyzing roughly 16,000 contracts over two months, the researchers found exactly the fingerprint that theory predicts: concentrated bursts of buying or selling on Binance in the final seconds before settlement, followed by rapid price reversals once the window closed. In contracts poised near 50-50 heading into expiry — where a tiny nudge decides everything — Binance order flow surged to about 3.9 times normal settlement-period volume. The moves were directional, precisely timed, and temporary: textbook settlement manipulation.
The Damage: 821 Wallets, $8.2 Million
The paper flags 821 suspected manipulator wallets, estimated to have earned around $8.2 million in total, with roughly $1.28 million effectively transferred from ordinary traders to these participants during the study window alone. One important caveat the authors stress: they could not directly prove the Binance accounts and Polymarket wallets belonged to the same people, so the evidence remains circumstantial — compelling patterns, not courtroom proof.
You might also like: Bybit Goes Local: Exchange Giant Launches Regulated Indonesian Platform After NOBI Takeover
The Fix Is Surprisingly Simple
The researchers’ proposed remedies are modest. When contract durations were extended from five to fifteen minutes, the manipulation signature largely vanished — longer windows make a momentary price shove too expensive to sustain. They also recommend settling on a time-weighted average price (TWAP) rather than a single snapshot, which would dilute any one-second spike into irrelevance. Polymarket, for its part, denies that manipulation occurred but has confirmed plans to introduce average-price settlement for some markets over the next year. Binance said it monitors its own platform but cannot dictate how third-party prediction markets settle contracts.
Bigger Than Bitcoin
The authors argue the warning extends well beyond crypto. As Cboe expands event contracts tied to the S&P 500 and Nasdaq pursues similar products, any contract that settles on a single price snapshot inherits the same exploit. And the stakes are growing fast: DefiLlama data shows Kalshi processed about $9.4 billion in June volume and Polymarket International roughly $4.3 billion, turbocharged by more than $5.4 billion in combined wagering on the expanded 2026 FIFA World Cup. All of this lands amid an escalating regulatory fight, with US states challenging Kalshi and Polymarket while the CFTC asserts primary authority over federally regulated event contracts — a dispute now working through federal courts that could ultimately reach the Supreme Court.
The message of the research is less “prediction markets are broken” than “settlement design is destiny.” Fix the snapshot, the authors argue, and one of finance’s fastest-growing arenas becomes dramatically harder to game.
Sources Consulted
Coinpedia via TradingView (original reference article); “Settlement Manipulation in Prediction Markets” working paper by Dai, Jia & Yu (Stanford / SMU); Yahoo Finance; CryptoTimes; BigGo Finance; KuCoin News; crypto.news; Cryptonomist; MoneyCheck; Cointribune; DefiLlama data as reported.






























