The favourite-longshot bias is the pattern that longshots win less often than their odds suggest, while favourites win slightly more often. So the average return from backing roughies is much worse than from backing favourites. Wikipedia's summary of the racing research puts it this way: losing 5% on favourites but 40% on longshots is not uncommon. For an Australian punter it matters most in one place: when you strip the margin out of a racing market to find fair odds, the method you use can turn a roughie from "value" into a clear loser.
This guide covers what the bias is, how big it is in racing compared with sport, why it happens, and a worked example with invented prices. No selections, just the maths.
What the favourite-longshot bias is
The pattern was first written up by R. M. Griffith in 1949, in a paper on American horse racing bettors in The American Journal of Psychology (JSTOR Daily). He found favourites were undervalued by bettors and longshots overvalued. It has turned up in racing markets around the world ever since.
Wikipedia gives a simple illustration, which looks like this in decimal odds:
| Quoted price | What the price implies | "True" price in the example | True chance |
|---|---|---|---|
| $3.00 (2-1) | 33.3% | $2.50 (1.5-1) | 40.0% |
| $101.00 (100-1) | 1.0% | $301.00 (300-1) | 0.3% |
The favourite is better than its price says. The longshot is far worse. Every market carries a margin, so both bets lose on average, but the longshot loses much more per dollar.
How big is it? Racing vs sport
The size of the bias depends on the market. Here's what the public evidence shows.
| Market | What was found | Source |
|---|---|---|
| Racing, bookmaker and tote | Losses around 5% on favourites against 40% on longshots are common | Wikipedia |
| US racing pools | Longshots overbet and favourites underbet, best explained by bettors misjudging small chances | NBER |
| Betfair Exchange racing | Overrounds of 101% to 102% close to the off; the gap to bookmaker prices widens further down the market | Sporting Life |
| AFL head to head, 2013 to 2024 | No clear bias across 2,375 games | Matter of Stats |
Racing: where it's strongest
Racing has everything the bias needs: big fields, lots of long prices, and plenty of casual money on big days. In a large study of US racing, Erik Snowberg and Justin Wolfers tested whether the bias came from punters enjoying risk or from punters misjudging probabilities. Using win bets and exotic pools (exactas, quinellas and trifectas), they found the evidence pointed to misjudged probabilities (NBER). Put simply, people give very small chances more weight than they deserve.
One finding is easy to miss. A summary of the same research notes the bias is not large enough to yield profit opportunities. Favourites lose less, but they still lose.
Exchanges: smaller margins, longer outsiders
On an exchange, punters set prices for each other. Sporting Life's Betfair guide reports UK exchange overrounds of 101% to 102% near the jump, while bookmakers may build in 105% or more. It also says 46% of all money staked at Betfair Starting Price on UK and Irish racing goes on the favourite, and that the gap between exchange and bookmaker prices widens as you go down the market. In other words, bookmakers tend to cut outsiders hardest. Remember the exchange charges commission on winnings, so compare after it, as our guide to Betfair commission shows.
Two-way sport: often little or none
Head-to-head markets look different. An analysis at Matter of Stats tested 2,375 AFL games from 2013 to 2024 using closing head-to-head prices (Matter of Stats). Its calibration measure came out at 0.98, within one standard deviation of no bias at all. The only possible exception was teams at $20 or longer: 12 home teams and 33 away teams across the period, and all of them lost. That's too few to measure, and too rare to matter for most bets.
So don't assume the bias is everywhere. Check the market you're betting in.
Why it happens
Researchers don't fully agree, and it's probably a mix.
- Misjudged small chances. Snowberg and Wolfers' evidence favours this one: bettors act as if very unlikely outcomes are more likely than they are (NBER).
- Fun. Some punters happily pay extra for the small chance of a big collect.
- Bookmaker risk. A long price that wins costs a bookmaker a lot. Wikipedia quotes a UK bookmaker saying it discounts the chances of anything at 100/1 or bigger, because lengthening near-hopeless outsiders isn't worth the risk.
- The margin itself. Data Golf argues that falling returns at longer odds follow from any market with a margin and punters who disagree, even when the middle of the market is unbiased. In their example, a 20% chance priced with the same absolute margin as an 80% chance returns -4.76% against -1.23%.
The last point matters for you. Whatever the cause, the result is the same: in many racing markets, more of the margin sits on the outsiders than a simple calculation assumes.
Why it matters: a worked example
Here's an invented six-runner race. The prices add up to a market of 114.9%. Our guide to market percentage in horse racing shows how to add it up.
There are two common ways to remove that 14.9% margin:
- Proportional (multiplicative): divide every runner's implied chance by 1.149. Every runner gives up the same share of its price.
- Power: raise every implied chance to the same power so they add to 100%. Here the power is 1.10. This takes relatively more off the longshots.
A study comparing these methods across three large bookmaker datasets found the power method allows for the favourite-longshot bias and outperformed the proportional method every time (Clarke, Kovalchik and Ingram).
| Runner | Price | Implied | Proportional fair | Power fair |
|---|---|---|---|---|
| A | $2.20 | 45.5% | 39.6% ($2.53) | 42.0% ($2.38) |
| B | $3.80 | 26.3% | 22.9% ($4.37) | 23.0% ($4.34) |
| C | $5.50 | 18.2% | 15.8% ($6.32) | 15.3% ($6.52) |
| D | $8.00 | 12.5% | 10.9% ($9.19) | 10.2% ($9.85) |
| E | $13.00 | 7.7% | 6.7% ($14.94) | 6.0% ($16.80) |
| F | $21.00 | 4.8% | 4.1% ($24.13) | 3.5% ($28.46) |
Under the proportional method, every runner carries 14.9% of margin on its price. Under the power method, the favourite carries about 8% and the $21 runner about 35%.
The value call flips
Now say another bookmaker has the favourite at $2.45 and the roughie at $26.00. Expected value is the fair chance times the price, minus 1.
| Bet | Price | EV, proportional | EV, power |
|---|---|---|---|
| Runner A | $2.45 | -3.1% | +2.9% |
| Runner F | $26.00 | +7.8% | -8.7% |
Same race, same prices. The proportional method says the roughie is the value bet and the favourite is a loser. The power method says the opposite. A finder that only uses the proportional method will flag longshots as value far more often than the evidence supports. Our guide to positive EV bet finders covers what to check before trusting a flag, and the implied probability guide compares the methods on two-way and three-way markets.
Neither method is right every time. Fair odds are an estimate. The point is to know how much your answer depends on the method, especially at the long end.
What it does and doesn't mean for you
It doesn't mean "just back favourites". Favourites lose less per dollar on average, but the research says the bias isn't big enough to beat the margin by itself (Overcoming Bias).
It does mean:
- Ask more of a roughie. If a long price only looks good under the proportional method, treat it as unproven, and set a higher minimum edge for long prices than for short ones.
- Check the exchange. Exchange prices on outsiders are often longer than bookmaker prices (Sporting Life), which makes them a useful reference for fair odds at the long end.
- Know your pool. Tote dividends come from a shared pool, so the crowd's love of roughies flows straight into the price, and pools are where much of the bias research comes from (NBER). Our tote vs fixed odds guide explains how each price is set.
- Track by price band. Split your results and your closing line value into bands (under $3, $3 to $10, over $10). Long-price bets swing hard, so a few wins can hide a poor long-run edge. Our guides to closing line value and a bet tracking spreadsheet show how.
- Don't import it everywhere. In two-way markets like AFL head to head, the evidence for the bias is weak (Matter of Stats).
Quick checklist
- Add up the market percentage for the race.
- Work out fair odds with the proportional method and with a method that allows for the bias, such as power. The BetBank calculators do the conversions.
- Compare the best available price with both fair prices.
- If a bet only looks like value under one method, treat it as noise.
- Check the price against the exchange, after commission.
- Record the price you took and the closing price, and review results by price band.
See the whole market at once
The favourite-longshot bias only shows up when you look at a whole field, not one runner. BetBank's Racing Odds Screen shows live fixed odds for every runner across the major Australian bookmakers, with the best price highlighted and price moves from market open to the jump, so you can see where the margin sits before you work out fair odds. It's part of the Gold plan, with a 7-day free trial.
BetBank is for adults 18+. Betting involves risk, and no method or tool removes it. Set a limit, take breaks, and gamble responsibly. If it stops being fun, our responsible gambling page lists free, confidential support, and BetStop lets you block yourself from all licensed Australian online and phone betting.