Prediction Market Biases

Prices aggregate opinion, but the process isn't neutral. Here are the recurring ways prediction market prices drift from a fair read of the odds.

A price is not automatically unbiased

It's tempting to treat a prediction market price as a clean, objective probability. In practice, a price is the output of whoever happened to trade, weighted by however much money they put behind their view. That process can be distorted in systematic, repeatable ways — not just random noise, but biases that push in a predictable direction. Knowing the common ones makes it easier to read a price for what it is rather than what it appears to be.

Favorite-longshot bias

Across betting markets and prediction markets alike, long-shot outcomes tend to trade a bit rich relative to how often they actually occur, while heavily favored outcomes trade a bit cheap. A plausible driver is that a portion of traders like the asymmetric appeal of a cheap contract with a big payoff and are willing to overpay for that chance, and there isn't always enough opposing pressure to correct it fully. The result is a consistent, mild skew rather than a one-off mistake.

Herd behavior

When traders start reacting to the price itself — buying because it's rising, selling because it's falling — rather than to their own independent read of the situation, the market stops aggregating diverse information and starts amplifying whatever direction it's already moving in. This is the mirror image of the conditions that make the wisdom of crowds work: once independence breaks down, a price move can become self-reinforcing regardless of whether it's justified.

Liquidity bias

Thinly traded markets are especially exposed to distortion. A single large trade can shift the price meaningfully in a market with little competing volume, and that shift can look like a change in collective belief when it's really just one participant's position. This is why volume is worth checking alongside price — a 70% probability on a market with heavy trading carries more weight than the same number on a market almost nobody is trading.

• Favorite-longshot bias: long shots overpriced, favorites underpriced.
• Herding: traders following the price instead of independent information.
• Liquidity bias: thin markets swinging on a single large trade.

Participant pool bias

Whoever is actively trading on a platform is not a random sample of everyone who might have relevant knowledge. If a platform's userbase skews toward a particular demographic, region, or worldview, prices on questions where that skew matters — politically charged outcomes are the clearest example — can reflect the pool's composition rather than a neutral aggregation of all available information. This doesn't necessarily make a price wrong, but it's a reason not to treat any single platform's odds as the final word.

Reading around the bias

None of this means prediction market prices are useless — it means they should be read with the same skepticism you'd apply to any single data source. Cross-checking price against volume, treating extreme long-shot prices with some suspicion, and being wary of prices that moved sharply on thin trading are simple habits that go a long way. For the deeper question of when these distortions become severe enough that a market meaningfully misprices an outcome, see the guide on why prediction markets fail.