What Is Wash Trading in Prediction Markets?
Wash trading in prediction markets is the practice of executing offsetting trades against yourself or a colluding party to inflate volume without taking real economic risk — and a 2025 Columbia University study estimated up to 25% of Polymarket's three-year volume was wash-traded.
Wash trading in prediction markets is the practice of executing buy and sell orders against yourself — or against a colluding counterparty — so the trades cancel out economically while still printing as volume on the exchange. It is illegal on regulated U.S. derivatives venues, prohibited by the platforms themselves, and increasingly cited as a structural risk to the credibility of the prediction market category as a whole.
How Wash Trading Works
A wash trader funds two or more accounts (or coordinates with another participant) and trades back and forth on the same contract. The economic position is roughly flat at the end of the day, but the tape shows volume — sometimes a lot of it. The motive is almost always to manipulate one of three signals:
- Headline volume used in marketing, valuation pitches, and league-table comparisons against competitors
- Liquidity perception that pulls in real traders who assume tight spreads and deep order books
- Probability signal itself, when wash trades are layered with a directional bias to push the implied odds before resolution
On order-book venues like Kalshi, wash trading typically requires self-matching across accounts. On AMM and hybrid venues like Polymarket, traders can route trades through the automated market maker or between wallets they control on-chain, which is why the activity has been easier to study using public blockchain data.
What the Data Says in 2025–2026
The conversation moved from theoretical to concrete in late 2025:
- Columbia University study (November 2025). Researchers analysing on-chain Polymarket activity estimated that roughly 25% of three-year cumulative volume was artificially inflated by wash trading, with some weeks reaching as high as 60% inauthentic volume.
- Kalshi vs. Polymarket public dispute (April 2026). Kalshi executives publicly alleged wash-trading rates as high as 70% in select Polymarket markets while defending their own volume figures. Polymarket disputed the methodology. The argument is unresolved but has put volume integrity in front of regulators and journalists.
- CFTC scrutiny. Wash trading is a defined prohibited practice under the Commodity Exchange Act, and the CFTC's 2026 prediction-market rulemaking explicitly contemplates surveillance standards designed to detect it.
The practical effect: any operator quoting "Polymarket volume" or "Kalshi volume" in a pitch deck now has to defend the methodology, not just the number.
Why Wash Trading Matters for Operators
For B2B operators evaluating prediction-market integrations, wash trading is not an academic concern — it changes unit economics and compliance posture.
- Liquidity you can't hedge against. If 25% of displayed volume is wash, the real liquidity available to absorb your users' orders is materially smaller than the public tape suggests. Operators planning to lay off risk on the underlying venue can be surprised mid-event.
- Probability signal contamination. The intelligence layer use case for prediction markets — pulling implied probabilities into sportsbook trading or risk models — only works if those probabilities reflect real money. Wash-traded markets feed noise into your pricing.
- Regulatory transfer risk. When a state regulator or the CFTC sanctions a venue for wash-trading failures, downstream B2B partners get pulled into the news cycle. Diligence on surveillance controls is now table stakes.
- Marketing claims become liabilities. Repeating an upstream venue's volume number in your own materials — without independent verification — is the kind of claim that surfaces in litigation and regulatory filings later.
How Venues Detect and Deter It
Mature prediction-market venues use a stack borrowed from traditional derivatives surveillance:
| Control | What It Catches |
|---|---|
| Self-match prevention | Two orders from the same account or owner crossing each other |
| Cluster analysis | Wallets or accounts that consistently trade only against each other |
| Statistical fingerprints | Abnormal first-digit, round-number, and trade-size distributions |
| Beneficial-ownership checks | KYC linkage between accounts that look independent on the surface |
| Position vs. volume ratios | Volume that never produces a meaningful net position over time |
On-chain venues have an advantage and a disadvantage: every transaction is publicly inspectable, but pseudonymous wallet creation is cheap. Off-chain venues see less but can enforce identity at the account level.
What This Means Going Forward
For any operator using prediction-market data as part of an intelligence layer, the right posture in 2026 is to assume a meaningful share of headline volume is non-economic, and to build pricing and risk models that lean on derived signals — open interest, position changes, settlement flows — rather than raw volume. Operators that treat surveillance as a feature, not a cost, will be the ones positioned to plug into licensed prediction-market frameworks as they emerge.
Last verified: May 2026