Favourite-Longshot Bias: Why Longshots Lose More

Favourite-longshot bias is the pattern that, at the same bookmaker, backing longshots loses more than backing favourites. It is a statement about returns by price, not a claim that favourites are a value bet. On Pinnacle 1X2 closes for five European leagues, SteamWatch's live table (read 7 November 2026) shows +0.26% from 9,375 favourite bets, −10.66% from the underdogs in the same matches, and −2.36% from the draw.

This page defines the bias, shows those football numbers and the 8,867-match study behind them, and explains why a sharp book can still load more margin onto outsiders. It is not a system. For how a two-way price is built from a 1X2, see draw no bet. For closing-line tests, see is Pinnacle's closing line accurate.

Definition. Favourite-longshot bias (FLB): in a betting market, the expected return of a level-stakes bet gets worse as the decimal odds get longer. Favourites (short prices) sit closer to fair; longshots (long prices) carry more of the overround.

Key takeaways

• On SteamWatch's Pinnacle 1X2 sample through 7 November 2026, blind favourites returned +0.26% and blind underdogs −10.66% over 9,375 matches.
• The dated study through August 2026 (8,867 matches) was +0.2% / −10.5% / −2.0% for favourite / underdog / draw. Super-dogs (5.30–36.00) lost 14.0%.
• A small plus on favourites is the vig handed back, not an edge you can bank. The useful fact is the gap: the dog needs a much larger misprice before the bet is even level.
• Two explanations coexist: bettors over-weight longshots (Snowberg & Wolfers, 2010), and books choose a heavier margin on long prices to equalise risk (Abrams / Buchdahl).
• Judge a price against the fair close, not against “it is the favourite”.

What is favourite-longshot bias?

Take every settled match. Back the shorter 1X2 side at the close, one unit. Then do the same for the longer side, and for the draw. If the short side loses a little or breaks even and the long side loses a lot, that is the bias.

It is not “favourites always win”. A 1.40 favourite still loses often. It is not “always back favourites”. A favourite at a soft book can be worse than a longshot that is too long relative to Pinnacle's fair price. The bias is a shape in returns by odds band.

Snowberg and Wolfers (NBER working paper 15923, April 2010, published Journal of Political Economy 2010) describe the same regularity in horse-race win bets: longshots are overbet, favourites underbet, and the pattern fits probability misperception better than risk-love. Football 1X2 is a three-way market, so the draw is a third band of its own.

What do Pinnacle football closes actually return?

SteamWatch publishes a live table of level-stakes 1X2 results at the recorded closing odds, five leagues combined from 2021/22, with league and season filters. Read 7 November 2026:

The favourite is whichever side closed shorter. A draw loses both the favourite bet and the underdog bet. Prices are football-data.co.uk's Pinnacle closes to January 2026 and SteamWatch's own Pinnacle close from February 2026; a gap of about 200 matches in between uses the Betfair Exchange close (quoted before commission, so a shade better than Pinnacle).

The original study on the same page, published 2 September 2026, stopped at the end of August 2026 with 8,867 matches: favourites +0.2%, underdogs −10.5%, draw −2.0%. The live table has grown by 508 matches and the shape has not changed.

Odds bands (study through August 2026)

Terciles, 2,956 matches each, boundaries from the data, not chosen after the fact:

Favourite and dog bands overlap on the decimal axis because home/away and the draw split the 1X2. The direction is the point: the longest prices are the worst returns.

Away favourites were the one blind cut that paid in that sample: +3.2% over 3,066 matches, against −1.4% for 5,801 home favourites. Slight favourites away were +6.8%; super-dogs at home against an away favourite lost 18.6%. Serie A underdogs were −15.6% overall and −23.9% on super-dogs. Premier League: favourites −0.5%, dogs −9.3%, super-dogs −17.6%. Ligue 1 mid-dogs showed +8.4% on 553 matches — SteamWatch flags that as noise.

None of those cuts is a betting instruction. They describe where the closing margin sat in this sample.

Is the bias a mistake, or how a book is built?

Two stories, both with sources, and they can both be true in different markets.

Demand-side. Bettors like longshots. They over-weight a 3% chance. The book takes the other side. Snowberg & Wolfers (2010) argue for misperception over risk-love in horse-race pools.

Supply-side. A bookmaker who puts the same percentage margin on 1.20 and on 6.00 takes more short-run risk on the 6.00. Dan Abrams' Pinnacle article (April 2022, “Why the favourite-longshot bias is not a bias”) treats equal maximum expected growth (Kelly) on both sides of a two-way market as the target; that produces a lighter margin on the favourite and a heavier one on the longshot. Joseph Buchdahl recast the same idea as equalising the chance a customer is ahead after 1,000 bets (football-data.co.uk, 14 November 2023, page updated 6 November 2026, read 7 November 2026). On Pinnacle ATP/WTA closing odds, 2014–2023, 47,269 matches and 94,538 prices, actual unit-stake losses tracked the losses implied by that margin split (t-test p = 0.50).

Football 1X2 has a draw, so Abrams' two-way formula does not drop in unchanged. The SteamWatch table is the empirical picture for football; Buchdahl's tennis set is the clean two-way test.

Power and Shin devig methods exist partly because a multiplicative (proportional) strip of the overround still leaves longshots looking too short. If you treat every price as equally far from fair, you will see “value” on dogs that is only the bias.

What this means for a dropping-odds bet

A Pinnacle drop on a longshot is a change in probability. It is not a reason to ignore the starting margin. If the close on a 6.00 dog is still 10% worse than fair in the long run, a 5% drop from 6.30 to 6.00 has not created an edge; it has reduced a large minus.

The check is the same as on any other market:

1. Remove the margin (devig calculator).
2. Compare your book's price with Pinnacle's fair price.
3. Record CLV against that fair close, with a sample large enough to mean something (CLV sample size).

Odds-on is the favourite's usual neighbourhood; see odds-on vs odds-against. How football 1X2, handicaps and totals move together is in football dropping odds.

How Pinnacle Terminal handles this

Pinnacle Terminal does not tag “favourite” or “longshot”. It shows Pinnacle's live price, the fair price (Power by default), the limit, and a 0–10 grade on every pre-match move (Big move, Sharp drop of about 20% or more from the recent high, Limit-backed, Downtrend, Rebound, Standard). Live in-play football drops are listed and filtered, without a grade. You still have to decide whether the fair number is short or long of your book's price. The terminal does not accept bets and is not affiliated with Pinnacle. Product: Pinnacle dropping odds. Plans: pricing.

FAQ

What is the favourite-longshot bias?

The pattern that long-priced outcomes return worse than short-priced ones at the same book, after you account for how often they win. In SteamWatch's Pinnacle 1X2 table through 7 November 2026, underdogs returned −10.66% and favourites +0.26%.

Does it mean I should always bet favourites?

No. A favourite can still be too short, and a longshot can still be too long relative to a fair price. The bias is about average returns by odds, not a rule for the next match.

Why do longshots have a bigger margin?

Books can load more juice onto long prices to keep short-run risk in line (Abrams / Buchdahl). Bettors also overbet longshots. In football 1X2 both can show up at once; the draw sits in between.

Is Pinnacle free of favourite-longshot bias?

No. The SteamWatch table is Pinnacle's close. The favourite side is close to level; the underdog side is not. Pinnacle's margin is smaller than a soft book's, so the hole on dogs is smaller than at a high-juice book, but it is still there.

How do I see it on a dropping-odds screen?

Look at the fair probability, not only the percentage drop. A 12% drop on a 1.50 favourite and a 12% drop on a 7.00 underdog are different amounts of probability, and they start from different margins.

Sources

• SteamWatch, Favourite–Longshot Bias, live table through 7 November 2026 (read 7 November 2026): 9,375 matches, five leagues from 2021/22; favourite +0.26% (5,086–4,289, +23.99u), underdog −10.66% (−999.15u), draw −2.36% (−220.84u).
• Neil Macdonald, “Favourite-Longshot Bias in Football: The Data”, SteamWatch, 2 September 2026 (read 7 November 2026): 8,867 matches through August 2026; +0.2% / −10.5% / −2.0%; tercile bands and home/away splits as quoted above.
• Joseph Buchdahl, “Favourite-Longshot Bias: A Bias or Not a Bias?”, 14 November 2023, page updated 6 November 2026 (read 7 November 2026): Pinnacle ATP/WTA close 2014–2023, 47,269 matches, 94,538 odds; t-test p = 0.50 vs expected losses from Abrams' split.
• Dan Abrams, “Why the favourite-longshot bias is not a bias”, Pinnacle Betting Resources, April 2022 (URL confirmed 7 November 2026).
• Erik Snowberg and Justin Wolfers, “Explaining the Favorite-Longshot Bias: Is it Risk-Love or Misperceptions?”, NBER Working Paper 15923, April 2010 (abstract read 7 November 2026): longshots overbet, favourites underbet; evidence for misperceptions.

Try it

Compare each Pinnacle price with its fair number, and grade the pre-match move, before you decide the longshot is “too big a drop to ignore”: start the Pinnacle Terminal 7-day free trial. Nothing is charged before day 8. Plans: pricing.

18+ · Bet responsibly. Only bet money you can afford to lose.