CROWD WISDOM VS. ACTUAL PROBABILITY
Plain English
Prediction market prices aggregate information from many participants — making them better than any individual expert in many cases. But markets can also be systematically biased. Longshot bias (overpricing rare events), recency bias, and political/emotional trading all create predictable mispricings that disciplined traders can exploit.
Going deeper
Research on prediction market accuracy (Wolfers, Zitzewitz; Tetlock's Superforecasters): prediction markets outperform expert panel forecasts and polls in most domains. However, systematic biases exist: (1) Longshot Bias — low-probability events are consistently overpriced (the market prices 10% events at ~12-15%). Selling No on 'unlikely' events can have positive expected value. (2) Recency Bias — recent outcomes anchor prices too strongly. (3) Political Bias — participants have emotional stakes in political outcomes, distorting prices toward their preferred candidate. (4) Thin Market Manipulation — in low-liquidity contracts, a single large participant can move prices significantly. (5) Early Trading Drift — contracts often drift toward extremes in the hours just before resolution as informed traders dominate over uninformed traders. Being aware of these biases lets you position against them.
Examples
Longshot Bias in Practice
Historical Kalshi data shows that contracts priced at 5% Yes resolve Yes approximately 3.5% of the time — meaning they're systematically overpriced. Selling No on very low-probability events (5-10% Yes contracts) in liquid markets can have positive expected value, especially for events with hard constraints.
Political Emotion Distortion
In major US elections, research shows that markets participants who 'want' their candidate to win bid up that candidate's contract beyond what polling data supports. This creates opportunities to fade the emotionally-inflated side after markets update to post-debate/post-event data.