Skip to content
Omnikalc

Bradley–Terry model calculator: probability that A beats B, odds, Elo difference and best-of series probability

Enter the strengths of two teams, a home advantage and the series format to get the win probability, the odds, the Elo difference and the series win probability.

Updated on
Your inputs

Value out of range

Value out of range

Value out of range

Results

Probability that team A wins one game

62.3 %

Probability that team B wins one game
37.7 %
Odds for team A
1.65 to 1
Equivalent Elo rating difference
87.0 points
Probability that team A wins the series
72.1 %

The Bradley–Terry model turns two strengths into a win probability. The Bradley–Terry model calculator gives the probability that team A beats team B, the odds, the equivalent Elo difference and the probability of winning a best-of series, with an optional home advantage.

How the calculation works

  • Enter the strength of team A and of team B. These are positive numbers on any common scale, for example the wins of each team or the parameters of a fitted model.
  • Enter the advantage multiplier of team A, such as 1.10 for a home field, or 1 for a neutral site.
  • The calculator applies the Bradley–Terry model: P(A beats B) = a / (a + b), where a is the strength of A times the multiplier.
  • It also gives the odds, the Elo difference that would give the same probability and, for a series, the probability of winning a best-of-3, best-of-5 or best-of-7.

Example

Team A has a strength of 1.8, team B has 1.2, and A has a home advantage of 1.10. The adjusted strength of A is 1.98, so P(A wins one game) = 1.98 / (1.98 + 1.2) = 62.3 %, and B wins 37.7 %. The odds for A are 1.65 to 1. The same probability on the Elo scale corresponds to a difference of 400 × log₁₀(1.65) = 87.0 points. In a best-of-5 series, team A wins with a probability of 72.1 %.

Why it matters

The calculator helps sports fans, bettors, analysts, game designers and students of statistics to turn ratings into probabilities and to see how much a series favors the stronger team. It also converts a ratio of strengths into Elo points, which makes comparisons with chess and sports rating systems easy.

Good to know

The model assumes that the strengths are fixed and that the games of a series are independent with the same probability, which is a simplification: a home advantage changes from game to game in a series, and teams adjust between games. Ties are not included. A probability from a model is an estimate, so compare it with the market or with real results before you use it to make a decision with money.

The Elo rating calculator updates ratings after a match, the Pythagorean expectation calculator estimates a win rate from points, the binomial calculator gives the chance of a number of wins, and the Poisson distribution calculator models the number of goals or events.

Frequently asked questions

What is the Bradley–Terry model?

The Bradley–Terry model gives the probability that one competitor beats another from a strength parameter for each. The probability that A beats B is a divided by a plus b. It is used for sports, chess, games, rankings of products and any paired comparison. The strengths are estimated from the results of past matches.

How is it linked to the Elo rating?

The Elo system is a form of the same model. If the strength is written as 10 to the power of the rating divided by 400, the win probability a / (a + b) becomes the Elo expected score. A ratio of strengths of 1.65 equals a difference of 400 times the log in base 10 of 1.65, which is about 87 Elo points.

How does the series probability work?

If the games are independent and the probability p of winning a game is the same for every game, the chance to win a best-of-n series is the chance to win a majority of the games. For a best-of-3 it is p squared times (3 − 2p), and for a best-of-5 it is p cubed times (10 − 15p + 6p squared). A series makes the better team more likely to win than a single game.

How should the strengths be chosen?

Use the same scale for both teams. You can take points scored, wins, a power rating or the result of a fitted model. If you only have a win percentage for each team, the strength can be taken as the odds p / (1 − p). The model assumes a stable strength and does not include injuries, rest or the form of the day.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.