Favorite-Longshot Bias: Why Cheap Contracts Lose Money
Cheap prediction market contracts win less often than their price implies. What the favorite-longshot bias is, what 300,000+ prices show, and what it means.
By Lenium · Published · 3 min read
A contract priced at 5¢ should, if prices were fair, win about 5% of the time. On prediction markets, cheap contracts win less often than that, and expensive ones win slightly more often. This pattern is called the favorite-longshot bias, and it is one of the most reliable findings about how people bet.
What the favorite-longshot bias is
The bias was first documented at horse races in the 1940s: bettors paid too much for longshots and too little for favorites, so betting on longshots lost money much faster than betting on favorites. It has since been found in sports betting and in options markets.
In a prediction market, the price of a yes/no contract is often read as a probability: 30¢ means a 30% chance. The bias means that reading is off at the extremes. Low prices overstate the chance of a hit, and high prices understate it.
What the exchange data shows
In a 2025 working paper by economists Constantin Bürgi, Wanying Deng and Karl Whelan of University College Dublin, the authors studied 313,972 contract prices from a major US-regulated prediction market exchange, from its launch in 2021 to April 2025. Their main findings:
- Prices are informative, but biased at the edges. A contract at 50¢ wins roughly half the time, and prices get more accurate as a market nears its close. But contracts priced low win less often than their price implies, and contracts priced high win a little more often.
- Cheap contracts lose heavily. Contracts bought at 10¢ or less lost more than 60% of the money put into them, on average.
- Much of the trading happens at extreme prices. About a third of all contracts traded at 1¢ to 10¢ and another third at 90¢ to 99¢.
- People taking prices lose the most. After fees, buyers who took the best available price (takers) lost about 31% on average; those whose resting orders were filled (makers) lost about 10%. Makers buying contracts priced at 50¢ or more made about 2.6%.
- The bias holds across categories, including politics, entertainment and economic data releases.
The authors' explanation is that takers are too optimistic about the outcomes they buy and that people overweight small chances, treating a true 20% chance as something closer to 23%.
Fees make longshots worse
The exchange's standard taker fee, as described in the paper, is 0.07 × P × (1 − P) dollars per contract, where P is the price in dollars, rounded up to the next cent. Because of the rounding, the fee is a much larger share of the cost of a cheap contract:
| Price | Cost of 100 contracts | Fee | Fee as % of cost |
|---|---|---|---|
| 5¢ | $5.00 | $0.34 | 6.8% |
| 20¢ | $20.00 | $1.12 | 5.6% |
| 50¢ | $50.00 | $1.75 | 3.5% |
| 80¢ | $80.00 | $1.12 | 1.4% |
| 95¢ | $95.00 | $0.34 | 0.4% |
A single 5¢ contract pays a 1¢ fee, 20% of its price. Check the exchange's current fee schedule before trading; fee rules change and vary by market.
What it means for your trading
- Be skeptical of cheap contracts. A 5¢ price is not a 5% chance. To make money buying longshots, you need to be right much more often than the price says, and then beat the fee.
- A big hit is not a strategy. One 5¢ contract that wins pays 20 times its cost, which makes a lucky month look like skill.
- Your entry price matters as much as your view. The study's gap between makers and takers shows that crossing the spread on every trade is expensive.
- Keep a record by price range. If your results on contracts under 20¢ are much worse than elsewhere, the bias is probably part of the reason.
Why Lenium limits trades to 20¢ to 80¢
Lenium evaluations only allow positions in contracts priced between 20¢ and 80¢ (a NO at 85¢ counts as a YES at 15¢, so it is outside the band). The reason is the same bias. At the extremes, prices are least reliable and results are most driven by luck: on a $25,000 account, $1,250 of 5¢ contracts that hit would make $23,750, nearly ten times the profit target, in a single trade. An evaluation that can be passed that way measures luck, not skill.
The evaluation also charges simulated exchange fees on the same formula, so results reflect what trading would actually cost. See all of Lenium's rules, or read how many trades it takes to prove an edge.