In July 2026, researchers from Stanford University and Singapore Management University published a working paper titled "Settlement Manipulation in Prediction Markets." Its conclusion is blunt: Polymarket's 5-minute Bitcoin Up/Down contracts contained a structural vulnerability that a small group of traders exploited systematically, at the expense of everyone else.
If you trade these markets — manually or with a bot — this matters more than any strategy article. Here's what the study found, what it doesn't say, and what a retail trader should actually do about it.
What the researchers found
The paper, by David Dai and Ruizhe Jia of Stanford's Department of Management Science and Engineering and Shihao Yu of Singapore Management University, examined roughly 16,000 five-minute Bitcoin contracts from their launch on 12 February 2026 through April.
The headline numbers:
| Finding | Figure |
|---|---|
| Suspected manipulators identified | 821 traders |
| Estimated profits to those traders | ~$8.2 million |
| Estimated transfer from retail traders | ~$1.28 million |
| Binance volume spike at settlement | ~3.9× normal |
| Settlement matching Binance direction | ~85% of the time |
The mechanism is straightforward once you see it. In the final seconds before a contract settled, Binance spot order flow spiked sharply, prices moved — and then reverted almost immediately afterwards. The researchers characterised this as a temporary push to move the spot price rather than trading on genuine information.
Why the design allowed it
Polymarket's 5-minute contracts settle against a Chainlink oracle that reads the spot price at a single moment in time. That single point of reference is the vulnerability.
You don't need to control the Bitcoin market to influence one instant of it. You need enough capital to move the price briefly, at a precisely known moment, while holding a position that pays out if you do. Because Binance's volume made it an effective proxy for the oracle's aggregate — the researchers found settlement matched Binance's direction around 85% of the time — the exchange to push was obvious.
Two conditions made this profitable:
A known settlement instant. Everyone knows exactly when a 5-minute window closes. That's normally a feature — it's what makes these markets automatable. It also tells a manipulator precisely when to act.
A single-point price reference. If settlement depends on the price at one moment rather than an average over a period, moving that one moment is enough.
The crucial detail: 15-minute markets were largely unaffected
This is the finding most coverage buried, and it's the most useful part for traders.
The manipulation pattern largely disappeared in the 15-minute contracts.
The economics explain why. Moving Bitcoin's spot price meaningfully costs real money, and the cost scales with how much you need to move it and how long you need it to stay moved. Over a 5-minute window with a single settlement point, a brief push can be enough — and cheap relative to the payout. Over a 15-minute window, positions are larger, more capital is required, and the manipulation has to overcome more genuine order flow. The trade stops being worth it.
The researchers' recommendations follow directly: extend the settlement window, or use a time-weighted average price (TWAP) instead of a single-point reference. Either change makes a momentary push useless, because a brief spike barely moves an average.
Polymarket's response
Polymarket has said it does not believe manipulation occurred as described. At the same time, the platform has indicated it plans to move toward average-price settlement, which is precisely the fix the researchers recommend.
It's also worth noting Polymarket tightened its market integrity rules in March 2026, covering market design standards, resolution criteria and data sourcing.
Read that how you like. The practical takeaway is the same either way: the settlement mechanism is changing, and until it does, single-point settlement on very short windows carries a structural risk that longer windows don't.
What this means if you trade 5-minute markets
Let's be measured rather than alarmist.
This does not mean every 5-minute market was rigged. The study identified patterns consistent with manipulation in a subset of contracts, driven by a small number of traders. Most windows resolved normally.
It does mean the tail risk is asymmetric and invisible. You cannot see a settlement push coming, you cannot price it, and no amount of strategy work protects against it. It's not a risk you can manage — only one you can choose to be less exposed to.
It also means some losses weren't your strategy's fault. If you've traded 5-minute windows since February and found results worse than your testing suggested, part of that gap may be structural rather than a flaw in your approach. That's cold comfort, but it's worth knowing before you rip up a strategy that was actually working.
What a sensible trader does now
Weight toward 15-minute windows. This is the clearest, evidence-backed adjustment available. Same instrument, same style of strategy, materially less exposure to the documented issue.
Don't abandon 5-minute markets on reflex — reduce exposure. They still offer more opportunities, and the manipulation affected a minority of contracts. Sizing down rather than switching off is the proportionate response.
Watch for the settlement change. When Polymarket moves to average-price settlement, this specific risk largely disappears and 5-minute markets become considerably more attractive again. That's a change worth tracking.
Judge any backtest covering Feb–April 2026 carefully. Historical results from that period include whatever effect this had. Anyone showing you spectacular 5-minute results from those months should be asked whether they've accounted for it.
How this affects Metazen Pulse — and what we're doing
We run strategies on both timeframes, so this applies to us directly rather than being someone else's problem.
Our 15-minute strategies are structurally safer, and that's now an evidence-based statement rather than a marketing one. If you're choosing which of our strategies to enable, the 15-minute set carries less exposure to the mechanism the study describes.
Our 5-minute strategies operate in the affected markets. We're not going to pretend otherwise. They remain available because the manipulation affected a minority of contracts and the timeframe still produces genuine opportunities — but if you're risk-averse, weighting toward 15-minute strategies is a reasonable choice, and the platform lets you enable or disable each strategy independently.
We're reviewing our own 5-minute performance against this period. Any strategy tested across February to April 2026 sits in the window the researchers studied.
Being straight about this costs us nothing worth keeping. A trader who finds out later that we knew and stayed quiet is a trader we've lost permanently.
Frequently asked questions
Is Polymarket itself unsafe to use? The study is about the design of one specific contract type, not the platform's solvency or honesty. Polymarket processed billions in volume in June 2026. The finding concerns settlement mechanics on short-duration crypto contracts.
Are 15-minute markets completely safe? Safer, not immune. The researchers found the pattern largely disappeared at 15 minutes because manipulation stops being economically worthwhile — not because it becomes impossible. Longer windows and TWAP settlement both raise the cost of trying.
Should I stop trading 5-minute markets entirely? That's your call. The proportionate response is reduced size rather than complete withdrawal — the affected contracts were a subset, and the settlement mechanism is expected to change.
How would I know if a contract I traded was manipulated? You wouldn't, in real time. The researchers identified it statistically across thousands of contracts by looking for order-flow spikes followed by immediate reversion. No individual trader can spot that during a five-minute window.
Does this affect Kalshi or other prediction markets? The study examined Polymarket specifically. The underlying issue — a contract settling against a single-point price reference that someone could move — is a general design concern for any short-duration market, not unique to one platform.
The bottom line
The study is a genuinely useful piece of research, and the honest reading is neither "prediction markets are rigged" nor "nothing to see here."
It's this: a specific contract design created a specific exploitable incentive, a small number of well-capitalised traders exploited it, and the fix is already understood. Longer settlement windows largely neutralise it today. Average-price settlement will neutralise it structurally when it arrives.
Until then, if you trade Polymarket's short Bitcoin markets, favour the 15-minute windows — and treat anyone showing you glowing 5-minute results from spring 2026 with appropriate scepticism.
Metazen Pulse runs automated strategies on both 5-minute and 15-minute Polymarket Bitcoin markets, each independently switchable so you control your exposure. Every non-premium signal is published live and free on our Telegram channel.
Related reading: Is it safe to give a Polymarket bot your private key? · How Polymarket's 5-minute BTC markets actually work
Trading prediction markets involves real risk of loss. Past performance does not guarantee future results. Nothing here is financial advice.
See the strategies before you trust anything
Every non-premium signal is published live and free on our Telegram channel — follow the calls in real time and check them against actual market outcomes.