Favorite-Longshot Bias in Prediction Markets: Why Longshots Are Overpriced

TL;DR

• Favorite-longshot bias means low-probability contracts systematically trade above their true odds, and favorites trade slightly below theirs, on every prediction market that prices through open trading, including Polymarket, Kalshi, Limitless, Predict.Fun, and Opinion.
• The distortion concentrates at the extremes. Contracts priced under 10% tend to resolve YES less often than their price implies; contracts over 90% resolve YES more often. Between roughly 30% and 70%, prices track outcomes closely.
• The cause is thin liquidity, not manipulation: a $500 order can move a thinly traded 5-cent contract far more than the same $500 moves a heavily traded 70-cent favorite.
• Discount extreme prices before trusting them. A 25% contract needs little adjustment; a 3-5% contract on a thin market often overstates the real probability by a meaningful margin.

Ever looked at a contract trading at 3 or 4 cents and thought that's basically free money if you're right? You're not the only trader reaching for it, and that crowding is exactly what makes the price wrong.

That tendency, longshot contracts trading rich and favorites trading slightly cheap, is called favorite-longshot bias. It shows up on every platform that lets traders set prices through open buying and selling, including Polymarket, Kalshi, Limitless, Predict.Fun, and Opinion.

It is not a glitch. It is a structural side effect of how thin markets absorb demand. Once you understand the mechanism, you can adjust for it instead of taking extreme prices at face value.

What Is Favorite-Longshot Bias?

The pattern was first documented in horse racing betting markets, then confirmed across sports books, options markets, and now prediction markets. Contracts priced under 10% tend to resolve YES less often than 10% of the time. Contracts priced above 90% tend to resolve YES more often than 90% of the time.

The distortion is not symmetric. It concentrates at the extremes and mostly disappears in the 30% to 70% range, where prices track outcomes closely.

For a reader using PredictionHero to compare markets, this matters most when a contract looks cheap. A 4% price feels like free money if you think the real chance is 8%. Sometimes it is. Often the 4% itself is the distortion.

This is a different quirk from prediction market overround, where a multi-outcome market's prices sum above 100% because of market-maker cushion. Favorite-longshot bias is about a single contract's price drifting from its true probability at the extremes, not about the sum across outcomes.

Why Does Favorite-Longshot Bias Happen?

Prediction markets set prices through supply and demand, not a bookmaker's judgment. That is the whole appeal. It is also the source of the bias.

A favorite with a 70% implied probability usually has deep interest on both sides. Plenty of contracts are listed, and plenty of capital is willing to take the NO side. A single trader placing a large order barely moves the price because there is enough depth to absorb it.

A longshot with a 3% implied probability lives in a much thinner market. Fewer people bother trading a contract that costs three cents and will probably expire worthless. That thinness is really a liquidity problem: the order book on both sides is shallow.

When a handful of buyers get excited about a specific outcome, an upset, a surprise announcement, a low-probability event with a big story attached, their combined buying pressure has nowhere near as much liquidity to push against. The price jumps further per dollar of demand than it would on a well-traded favorite.

There is also an asymmetry in what pulls people toward longshots in the first place. A contract priced at 3 cents offers a payout of roughly 33 to 1 if it hits. That ratio is exciting even to traders who know the true odds are worse than the market says. A small amount of capital chasing a large multiple is enough to move a thin book meaningfully, and it keeps happening because the payout story is more persuasive than the math.

How Much Can One Order Move a Longshot's Price?

Say a market has a longshot contract trading at $0.05, implying a 5% chance. The order book is thin: only $2,000 in total resting liquidity across both sides, because almost nobody thinks this outcome matters enough to trade.

A trader convinced this outcome is undervalued puts in a $500 buy order. On a thin book, $500 is enough to walk the price up through several price levels, and the contract now trades at $0.08. That is a 60% jump in implied probability from a single order that is a fraction of the size that would move a heavily traded favorite by even a cent.

Compare that to a favorite trading at $0.70 on the same platform, sitting on $200,000 of resting liquidity because it is one of the most talked-about markets that week. A $500 order there barely registers. The price might tick from $0.70 to $0.701.

Same dollar amount. Wildly different price impact. That gap is the mechanism behind favorite-longshot bias in one example. It has nothing to do with the outcome itself and everything to do with how much capital is sitting on either side of the trade before the order arrives.

How Should You Discount a Longshot's Price?

Treat extreme prices as a starting point, not a probability estimate, using three checks.

• Check the liquidity, not just the price. A 4% contract backed by six figures of resting orders on both sides has earned some trust. A 4% contract sitting on a few hundred dollars of total interest has not. Thin books produce noisy prices.
• Compare the number across platforms. This is where cross-platform aggregation earns its keep. If a longshot prices at 6% on one exchange and 3% on another, the spread itself is informative. That is often the same signal behind why prediction market odds disagree across platforms: at least one platform's price was set by a small number of trades, not broad consensus.
• Discount harder as the number gets smaller. The bias is not linear. A contract at 25% is usually close to fair. A contract at 8% deserves real skepticism. A contract at 2% is often several multiples too high relative to its true chance, because it takes so little capital to hold a small market at an inflated level.

None of this means longshots are always wrong. Some 5% contracts do happen 5% of the time. The framework is about weighting: treat a favorite's price as close to its real probability, and treat a deep longshot's price as an upper bound that needs to be pulled down before you use it for anything.

Which Platforms Show Favorite-Longshot Bias Most?

The size of the bias tracks liquidity depth, which is why it looks different platform to platform.

Polymarket carries the deepest books in the industry on its biggest events, which compresses the bias on high-attention markets but does not eliminate it on the long tail of niche contracts. Kalshi, as a CFTC-regulated exchange, draws a domestic retail base that concentrates volume on a narrower set of markets, which can leave secondary contracts thinner than they look.

Limitless and Predict.Fun run on-chain with smaller overall volume than the two majors, so their longshot contracts are more exposed to single-trade price swings. Opinion's trader base leans toward macro and economic events, and pricing on lower-probability tail scenarios there follows the same pattern: thinner interest, bigger swings per dollar traded.

PredictionHero currently tracks 342,448 markets across 28,371 events and matches 2,806 of them across all five platforms, which is enough live coverage that pulling a current order book instead of a stale one is always the better move before trusting a longshot price.

If you are comparing a longshot contract across all five platforms in one place, the spread between platforms is often the fastest tell that a price has drifted from fair value rather than reflecting new information.

Ready to see this in action? Compare current longshot pricing across Polymarket, Kalshi, Limitless, Predict.Fun, and Opinion on PredictionHero.

Sources

• NBER Working Paper No. 15923, "Explaining the Favorite-Longshot Bias." nber.org

PredictionHero aggregates publicly available prediction market data for informational purposes only. This is not financial advice. Prediction markets may not be available in all jurisdictions.