Poisson Football Calculator

λPoisson calculatorUpdates as you type

Home expected goals (λ)Away expected goals (λ)

Home win-fair -

Draw-fair -

Away win-fair -

Over 1.5-fair -

Over 2.5-fair -

Over 3.5-fair -

Both teams score-fair -

Most likely scores

Score grid (%)

This Poisson distribution calculator turns two numbers, the expected goals for the home team and for the away team, into a full set of football probabilities. You get the home, draw and away chances, over 1.5, 2.5 and 3.5 goals, both teams to score, the five most likely scorelines, fair odds for each, and a 0–5 by 0–5 score grid.

What the Poisson model does

Goals are rare, roughly independent events in a 90-minute match, and the Poisson distribution describes how often rare events happen when you know their average rate. If a team is expected to score 1.6 goals, Poisson tells you the chance it scores 0, 1, 2 or more. Do the same for the other team, multiply the two, and you have a probability for every score. Statisticians have used this approach for football since at least the early 1980s, when M.J. Maher published a well-known model.

How to use the Poisson distribution calculator

1. Estimate home expected goals (λ home). The average number of goals you expect the home side to score in this match.
2. Estimate away expected goals (λ away). The same for the away side.
3. Enter both values. The tool calculates 0–9 goals for each team and adds up every score.
4. Read the outputs. Compare the fair odds with the bookmaker’s prices to spot possible value.

The formula and how scores are combined

The chance that a team scores exactly k goals is λ^k × e^−λ ÷ k!, where e ≈ 2.718 and k! means k × (k − 1) × … × 1. With λ = 1.60, the home side scores 0 goals 20.2% of the time, 1 goal 32.3%, 2 goals 25.8% and 3 goals 13.8%.

Because the model treats the two teams as independent, the chance of a scoreline is the product of the two. If the away side has λ = 1.10, it scores exactly once 36.6% of the time, so 1–1 has a chance of 32.3% × 36.6% = 11.8%. Adding every score where the home team leads gives the home win chance; adding the diagonal gives the draw.

How to estimate expected goals

The calculator is only as good as your λ values. A common method uses attack and defence strength relative to the league average:

• Home attack strength = home team’s goals scored per home game ÷ the league average home goals per game.
• Away defence strength = away team’s goals conceded per away game ÷ the league average away goals conceded.
• λ home = league average home goals × home attack strength × away defence strength. Do the mirror calculation for λ away.

Expected goals (xG) data from a full season is usually steadier than actual goals, because it is less affected by lucky finishing. Use at least 10 to 15 matches, and weight recent games more if a team has changed coach or key players.

Reading the results: a full example

With λ home 1.60 and λ away 1.10, the calculator shows:

Over 1.5 goals comes out at 75.1% and over 3.5 at 28.6%. The five most likely scores are 1–1 (11.8%, fair odds 8.45), 1–0 (10.8%, 9.30), 2–1 (9.5%, 10.57), 2–0 (8.6%, 11.62) and 0–1 (7.4%, 13.53). Notice that even the likeliest score happens less than one time in eight.

Turning probabilities into betting decisions

Fair odds are the prices with no bookmaker margin. If a bookmaker offers 2.20 on the home win and the model says 49.0%, the value bet calculator gives EV = 0.49 × 2.20 − 1 ≈ +0.08. If the offer is 1.95, there is no value. Use the same check for the goals markets on our over/under 2.5 page and both teams to score picks.

The score grid is also useful for scorelines. Our correct score tips page carries daily picks, and the dutching calculator splits a stake across two or three of the likeliest scores from the grid.

Limits of a Poisson football model

• Draws are slightly under-estimated. Real matches end level, especially 0–0 and 1–1, a little more often than independent Poisson predicts. Dixon and Coles proposed a correction for this in 1997.
• Game state is ignored. A team that goes 1–0 up often sits deeper, and a trailing team pushes; the model assumes the scoring rate never changes.
• Red cards, injuries and weather are not in the inputs. Adjust λ yourself when team news changes.
• Small samples mislead. Five games of data can make an average side look elite. Early-season numbers need blending with last season’s.
• Motivation and rotation (cup weeks, dead rubbers) are invisible to the formula.

To compare the model with the market fairly, remove the margin from the bookmaker’s 1X2 prices with the no-vig calculator first.