The favourite-longshot bias explained
Long shots lose bettors more per pound staked than favourites do. What the favourite-longshot bias is, how big it is, and what it means for a price.
The favourite-longshot bias is the long-running finding that bets on long shots lose more per pound staked than bets on favourites. A horse or team at long odds tends to win less often than its price implies; a strong favourite wins about as often as its price implies. The gap can be large.
How big the gap is
The clearest numbers come from horse racing, where the data run to millions of bets. Economists Erik Snowberg and Justin Wolfers studied all 6.4 million US horse race starts from 1992 to 2001:
| Betting strategy | Return per unit staked |
|---|---|
| Back the favourite in every race | −5.5% |
| Back horses between 4/1 and 9/1 | about −18% |
| Back a horse at random | −23% |
| Back horses at 100/1 or longer | about −61% |
Every row is negative, because the track takes its cut first. The point is how much worse it gets as the odds get longer. A study of football betting by Daunhawer, Schoch and Kosub found the same pattern in football prices.
Why it happens
Snowberg and Wolfers tested two explanations. One is that bettors enjoy risk and will pay extra for a small chance at a big payout. The other is that people misjudge small chances, treating 1% as if it were more. Using the prices of combination bets, they found the second explanation fits the data better. A related reason in fixed-odds betting is that bookmakers face less pressure on long-shot prices, so they can take more margin there. How margin hides in a price is covered in what the bookmaker's margin really costs you.
What it means for a price
When you turn odds into a chance, as in implied probability in one minute, the bias says the long-shot number is usually too generous to the long shot. So a long shot needs a bigger gap between your chance and the implied chance before it is worth anything.
That is relevant to our own board. Today's Desk picks are mostly underdogs at modest prices, not long shots:
| Pick (Desk, 23 Oct) | Our chance | Best price | Edge |
|---|---|---|---|
| Arizona | 70.0% | 1.95 | +18.7 pts |
| Tennessee | 58.0% | 2.16 | +11.7 pts |
| Ole Miss | 54.5% | 2.46 | +13.9 pts |
| Northern Illinois | 55.5% | 2.68 | +18.2 pts |
| Virtus Bologna | 58.2% | 2.60 | +19.7 pts |
None of these is a long shot by the study's standard; the longest is 2.68. But edges this wide are exactly where we check our own work, as we explain in why a very wide edge makes us look twice.
What this does not tell you
The main figures come from US horse racing from 1992 to 2001, a pool betting market, not today's fixed-odds football or basketball markets. The size of the bias differs by sport, market and bookmaker. It does not make favourites a winning bet: they lost 5.5% in that data. And it does not tell you whether any one price is wrong; only a chance you trust, compared with the price, can do that.
Common questions
What is the favourite-longshot bias?
The pattern that bets on long shots return less per unit staked than bets on favourites. Long shots win less often than their odds imply, while favourites win about as often as theirs imply, or slightly more.
How big is the favourite-longshot bias?
In 6.4 million US horse race starts from 1992 to 2001, Snowberg and Wolfers found betting horses at 100/1 or longer returned about −61%, betting at random −23%, and backing the favourite in every race −5.5%.
Why does the favourite-longshot bias happen?
Snowberg and Wolfers found the best explanation is that bettors overweight small chances, rather than simply enjoying risk. Bookmakers can also load more margin onto prices that people do not check closely.
Does it mean you should always back favourites?
No. Losing less is still losing. Favourites in their data still returned −5.5%, so the bias only tells you where the price is usually worst, not where it is good.