Beginner Guide

Favorite-Longshot Bias in Prediction Markets

Why markets often overprice unlikely outcomes and underprice likely ones, and what that means for readers and traders.

By Top Prediction Markets EditorialReviewed September 7, 20265 min read

Answer first

The favorite–longshot bias is a pattern where markets assign relatively too much probability to longshots (low-probability outcomes) and too little to favorites (high-probability outcomes). It appears for behavioral, informational, and structural reasons and shows up most clearly in low-liquidity or fixed-odds settings. For readers and traders, the key is to compare market prices to objective or aggregate estimates, watch liquidity and fees, and avoid simple arbitrage attempts without accounting for costs and uncertainty.

A concrete scenario to carry through: a 62¢ favorite and an 8¢ longshot

Imagine a two-outcome prediction market for an event with two prominent contracts. One contract — the favorite — trades at 62¢ (implying a 62% market probability). The other contract — a longshot — trades at 8¢ (implying an 8% market probability). An analyst estimates the favorite’s true chance at 70% and the longshot’s true chance at 2%.

This setup is chosen to show the favorite–longshot bias in action: the pattern where longshots are priced as if they are more likely than they really are, while favorites are priced as if they are less likely than they really are. Keep these exact numbers in mind as we run the arithmetic and then revisit why prices might look this way.

Outcomes and quick facts

Outcome	Market price	Analyst estimate
Favorite wins	$0.62	70%
Longshot wins	$0.08	2%

Both are standard Yes contracts that cost the quoted price and pay $1 if the outcome occurs.

Running the numbers: expected value on each contract

Start from the contract payoffs. Buying one Yes contract at 62¢ for the favorite costs $0.62. If the event happens it pays $1, so the profit on a win is $0.38; on a loss you lose $0.62.

Using the analyst’s 70% (0.70) estimate, expected value (EV) per favorite contract is:

EV = 0.70 × $0.38 − 0.30 × $0.62 = $0.266 − $0.186 = $0.08.

That positive $0.08 EV shows the favorite is underpriced relative to the analyst’s estimate.

For the longshot: buying one Yes contract at 8¢ costs $0.08. If the outcome happens you gain $0.92; if it fails you lose $0.08.

Using the analyst’s 2% (0.02) estimate, expected value per longshot contract is:

EV = 0.02 × $0.92 − 0.98 × $0.08 = $0.0184 − $0.0784 = −$0.06.

That negative $0.06 EV shows the longshot is overpriced relative to the analyst’s estimate.

These calculations assume you accept the analyst’s true-chance estimates and ignore fees, bid-ask spreads, trade limits and other execution frictions.

What each outcome pays and how to think about risk

Keep the same numbers and consider the four elementary payoff scenarios:

Trade	Cost	If it wins	Net gain on win	If it loses	Net loss on loss
Buy 1 favorite	$0.62	$1.00	+$0.38	$0.00	−$0.62
Buy 1 longshot	$0.08	$1.00	+$0.92	$0.00	−$0.08

These figures are the raw building blocks for the EV calculations above. They also make a behavioral point clear: longshots offer larger upside per bet but at much lower probability, while favorites offer smaller upside with higher probability. The favorite–longshot bias is precisely about how markets and traders weigh these asymmetric payoffs relative to objective chances.

If you were to hold a portfolio of many identical trades, realized returns will depend on the true long-run frequencies. But in small samples, especially in thin markets, outcomes are noisy — and that noise interacts with trader behavior and market microstructure to produce systematic-looking biases.

Why prices deviate here: five linked causes, tied back to the scenario

Behavioral causes. People tend to overweight tiny probabilities and are often risk-seeking when faced with low-probability large payoffs. That can explain why the 8¢ longshot attracts more money than an objective 2% chance would justify, while the 62¢ favorite gets relatively less backing than a 70% chance would warrant.
Informational gaps. Favorites are often visible and well-discussed; longshots can attract speculative or emotional interest. If more informed traders concentrate on the favorite and fewer contest the longshot, the longshot can stay overpriced relative to the analyst’s estimate.
Market structure and frictions. Different payoff formats and pool structures — for example pari-mutuel pools or fixed-odds markets — change incentives. When many bettors back a longshot in such systems, payoffs for winners fall and the apparent overpricing of longshots is amplified.
Liquidity and sample size. Low trading volume makes it easy for a few bets to move prices. In our scenario, a single sizable bet on the longshot could push the price from a fair 2¢ to 8¢, producing the appearance of a bias even if outcomes are random in the long run.
House edge and overround. Commercial settings often build a margin into prices. That margin interacts with bettors’ preferences and tends to make longshots look especially poor value on average.

Each of these mechanisms can plausibly produce the exact pattern in our scenario: favorite underpriced at 62¢ vs a 70% estimate, and longshot overpriced at 8¢ vs a 2% estimate.

How the maths shifts if the price moves

Suppose trading pushes the favorite price up from 62¢ to 66¢. The win profit drops from $0.38 to $0.34 and the EV under the analyst’s 70% estimate becomes:

EV = 0.70 × $0.34 − 0.30 × $0.66 = $0.238 − $0.198 = $0.04.

The positive EV is smaller; repeating the arithmetic shows how incremental price movement changes expected profitability.

Similarly, if the longshot drifts from 8¢ down to 4¢, the upside per win becomes $0.96 and the EV with a 2% true chance becomes less negative, possibly crossing to positive at some point.

But three practical points limit straightforward arbitrage:

Fees, bid-ask spreads and trade limits often eliminate the small EVs that arithmetic produces.
Price moves require counterparties; in low-liquidity markets a trader large enough to profitably shift price may be unable to execute the needed trades at the new price.
Sample noise and uncertainty in your own probability estimates mean that apparent mispricings may be illusions — profitable only if your estimate is reliably accurate.

These constraints are why a calculation that looks attractive on paper (as with the 62¢ favorite under a 70% estimate) may not translate into an executable, low-risk strategy in practice.

Trading mistakes to avoid in this scenario

Treating market price as a perfect probability. The market gives an implied probability, but that reflects trader preferences, liquidity and fees as much as information. Don’t assume the 62¢ or 8¢ price equals the true chance.
Assuming the bias is uniform across markets. The favorite–longshot bias is stronger in some settings (horse racing, low-liquidity political submarkets) and weaker in others (high-volume, well-informed exchanges). Evaluate context before generalizing from our two-contract example.
Trying to arbitrage without accounting for costs. Small systematic mispricings can look exploitable, but fees, bid-ask spreads, trade limits, and the risk of being wrong on your probability estimate often eliminate practical profit opportunities.

Further reading related to this worked example

Frequently asked questions

What exactly is the favorite–longshot bias?

It’s a pattern where low-probability outcomes (longshots) receive relatively more money and are priced too high, while high-probability outcomes (favorites) receive relatively less and are priced too low, compared with objective or realized frequencies.

Does the bias always mean favorites are profitable to buy?

Not always. Apparent mispricing can be removed by fees, limits, execution risk, and uncertainty in your own probability estimates, so a favorite that looks cheap may still be risky after costs.

Where is the bias strongest?

It tends to be strongest in low-liquidity markets, pari-mutuel pools, and settings with asymmetric information or emotional betting. High-volume, well-informed markets usually show a smaller effect.

Can the bias be used to forecast elections or outcomes?

Markets with enough liquidity and information are useful forecasts, but the bias warns you to treat small-probability prices with caution. Compare market prices to polls, models, and aggregated information before drawing conclusions.

Are prediction markets legal?

Rules vary by location and platform. See our dedicated guide on whether prediction markets are legal in the US.