Story file
What are good odds in betting? Priced against probability
Good odds are not long odds. A price is good only when it beats the true chance. Five prices, a 107.14% book and the longshot evidence, in pounds.
Good odds in betting are odds longer than the outcome’s true chance deserves: a price whose implied probability is lower than the real probability, so the bet has positive expected value. A long price is not automatically a good one. 10/1 is good only if the outcome wins more often than about one time in eleven.
This is an evergreen explainer with no news event behind it. Everything below is arithmetic on published prices, plus the published abstracts of two academic papers this desk read on 17 October 2026.
What are good odds in betting, as a number?
Every price carries an implied probability. For fractional odds it is the denominator divided by the sum of both numbers; for decimal odds it is 1 divided by the price. That figure is also the break-even strike rate: the share of bets that must win for level stakes to come out square.
| Fractional | Decimal | Implied probability | £10 returns | Wins per 100 £10 bets to show a profit |
|---|---|---|---|---|
| 1/2 | 1.50 | 66.67% | £15.00 | 67 |
| Evens | 2.00 | 50.00% | £20.00 | 51 |
| 2/1 | 3.00 | 33.33% | £30.00 | 34 |
| 5/1 | 6.00 | 16.67% | £60.00 | 17 |
| 10/1 | 11.00 | 9.09% | £110.00 | 10 |
Returns include the stake. These are illustrative prices, not any bookmaker’s quote. The question is never “is 5/1 big?” but “does this happen at least 17 times in 100?” The evens row is worked in more detail in what EVS means in betting.
How do you tell if odds are good value?
Expected value is the average result of a bet over many identical repeats. £10 at 2/1 on an outcome with a true 40% chance has an EV of +£2 (0.40 × £30 − £10). At a true 30%, it is −£1. The price is the same in both cases; only the probability changed.
That is the catch. Nobody can observe the true probability. A bettor has an estimate, and an estimate can be wrong for a long run of results, while a real edge can be buried by variance for months. Good odds are a judgement about a number you cannot see, checked against one you can.
How much does the bookmaker’s margin cost, in pounds?
Outplayed’s explainer, dated 16 February 2026 and read by this desk on 17 October 2026, uses a Cardiff v AFC Wimbledon market priced Cardiff 1.5, draw 4.2, Wimbledon 6. It says bookmakers add a margin to all their odds; in the text read, it does not work that margin out. The desk’s arithmetic does:
- Cardiff: 1 ÷ 1.5 = 66.67%
- Draw: 1 ÷ 4.2 = 23.81%
- Wimbledon: 1 ÷ 6 = 16.67%
The book adds up to 107.14%, an overround of 7.14 points. To get £100 back whichever result lands, stakes of £66.67, £23.81 and £16.67 cost £107.14 in total (£107.15 once each stake is rounded to the penny). Backing everything loses £7.14, or 6.67% of the money staked.
Strip the margin out in proportion and the fair prices are about 1.61, 4.50 and 6.43. If the bookmaker’s own view were exactly right and the margin spread that way, every £10 at these prices would return £9.33 on average. The margin multiplies through multiples, as the Lucky 31 vs Lucky 15 settlement shows.
Are long odds good value? The favourite-longshot bias
The research says, if anything, the reverse. Erik Snowberg and Justin Wolfers’ NBER Working Paper 15923, issue date April 2010, describes the favourite-longshot bias as a longstanding regularity in which betting odds misestimate a horse’s chance: longshots are overbet and favourites underbet. Its data was supplied by Equibase and its tests use exacta, quinella and trifecta pools, so it is North American racing, not British. The authors find the evidence favours misperception of probabilities, as in Prospect Theory, over gamblers’ love of risk.
The British evidence is Smith and Vaughan Williams in the International Journal of Forecasting (vol. 26(3), July 2010). Using bookmaker odds from ten seasons of UK Flat racing, they confirm the bias: level stakes at shorter odds returned more than level stakes at longer odds. They also found the bias declined significantly, lower in the five years after 2000 than in the five before.
Two limits. Neither paper studies football, so the finding does not carry to a football market by assumption. And a bias that shrank once can shrink again: it is a reason a 20/1 shot is not good value by default, not a system. Bookmaker promotions get the same audit in price boosts, audited.
Frequently asked questions
What are good betting odds? Odds that pay more than the outcome’s true chance deserves. Convert the price to an implied probability (1 divided by the decimal odds) and compare it with your own estimate of the chance. If the implied figure is lower, the price is good; if it is higher, it is bad, however long it looks. The estimate is the hard part.
How are betting odds calculated? A bookmaker starts from an estimate of each outcome’s chance, then shortens every price to build in a margin, and moves prices as money arrives. That is why a market’s implied probabilities add up to more than 100%. Fractional odds convert to implied probability as denominator divided by numerator plus denominator, so 2/1 is 1 in 3.
Are long odds good value? Not by default. The favourite-longshot bias, found in North American racing by Snowberg and Wolfers and in ten seasons of UK Flat racing by Smith and Vaughan Williams, means longshots have tended to pay less per pound staked than favourites. The UK study also found the bias shrank after 2000, so it is not a standing edge.
A good price still loses often; at 2/1 it should lose two times in three. Set the week’s loss limit in pounds before the first bet, and treat reaching it as the end of the week, not a reason to find a longer price. 18+.