Mean Reversion

How prediction market prices can overreact to news and subsequently revert

By Prediction Pilot Research · Published April 2026 · Last reviewed May 2026

What is mean reversion?

Mean reversion in prediction markets is the tendency of overreaction-to-news prices to correct back toward the pre-event level within hours or days. Academic studies of binary betting markets find reversal effects of 3-7% within 24 hours of high-magnitude news shocks (Avery & Chevalier 1999; Croxson & Reade 2014). On Kalshi, the trade is to fade extreme price moves on low-information news in markets with normal volume — not headline events where the move is justified.

In prediction markets, mean reversion occurs when a contract price moves sharply on news or sentiment, overshooting the rationally implied probability, and then partially reverses as calmer analysis prevails.

The overreaction hypothesis

De Bondt and Thaler (1985) published one of the most influential papers in behavioral finance, documenting that stocks which had performed extremely well over 3-5 years subsequently underperformed, and vice versa. They attributed this to investor overreaction — markets initially overweight dramatic new information.

The mechanism in prediction markets is similar but compressed in time:

1. News event occurs — A political scandal, surprising economic data, or unexpected weather pattern
2. Initial reaction — Market participants rush to reprice, often based on emotional or heuristic reasoning
3. Overshoot — Prices move beyond what the new information rationally justifies
4. Correction — More deliberate analysis brings prices back toward fair value

Evidence in prediction markets

Research on prediction market overreaction is more limited than the equity literature, but several studies are relevant:

Political markets

Rothschild (2009) studied prediction market accuracy for elections and documented that market prices tend to overreact to polling news. A single poll showing an unexpected result can move a contract 10-15 cents, only for the price to partially revert as subsequent polls confirm or deny the trend. The pattern is most pronounced in lower-liquidity markets where a few large traders can move prices.

Sports and event markets

Tetlock (2004) analyzed TradeSports contracts around Iraq War events and found evidence of short-term overreaction to breaking news. Prices moved sharply on initial reports and partially reverted as the full picture emerged. The reversions were larger for events where initial information was ambiguous or incomplete.

Thin market effects

Overreaction is amplified in thin markets. When a single large order moves the price 8 cents in a low-volume market, subsequent trades by other participants often push it back. This is partly mechanical (the price was moved by a liquidity shock, not information) and partly behavioral (the initial move attracts contrarian traders).

When overreaction is most likely

Research and market observation suggest overreaction is more common in certain conditions:

• Low liquidity — Thin order books amplify the price impact of any single trade. A market with $500 in volume is more susceptible to overshoot than one with $50,000.
• Ambiguous information — When the news is unclear or its implications are debatable, initial reactions are more likely to be wrong. Clear, unambiguous information (a definitive election result) is priced correctly quickly.
• Emotional topics — Political markets are more prone to overreaction than weather markets because participants have emotional attachments to outcomes (partisan bias). Mellers et al. (2014) documented how "superforecasters" outperform by resisting emotional anchoring.
• After hours — News that breaks when trading activity is low can cause outsized moves that revert when broader participation resumes.

Implementing mean reversion

A mean reversion approach in prediction markets typically involves:

1. Monitoring for sharp moves — Identify contracts that have moved significantly (e.g., 10+ cents) in a short period
2. Assessing the catalyst — Determine whether the move is justified by genuinely new, definitive information or is likely an overreaction to ambiguous news
3. Fading the move — If assessed as an overreaction, take the opposite side with the expectation that the price will partially revert
4. Managing risk — Set a stop-loss or maximum holding period, because some "overreactions" turn out to be correct repricing

Key considerations

• Not all sharp moves are overreactions — Sometimes a 15-cent move is the correct repricing and the price continues moving. Distinguishing between overreaction and rational repricing is the central challenge, and getting it wrong means buying into a trend rather than fading one.
• Binary settlement risk — If a prediction contract settles to 0 or 100 before a reversion occurs, the trader takes the full loss. There is no gradual unwinding — the binary payoff structure makes "waiting for mean reversion" riskier than in equity markets.
• Survivorship bias in backtests — Documented examples of prediction market overreaction tend to focus on the reversions that happened, not the times when the initial move was correct. The base rate of "sharp moves that revert" vs. "sharp moves that continue" matters enormously.
• Requires real-time monitoring — Overreaction opportunities are fleeting. By the time a daily scanner picks up the move, the reversion may have already occurred.
• Behavioral edge erodes — As prediction markets mature and attract more sophisticated participants, overreaction frequency may decrease. The easy overreaction trades may have already been competed away.