Enter the expected goals for each side – the home lambda and the away lambda – and the calculator returns the probability of each goal count, the over/under 2.5 split, and fair odds for individual scorelines. A lambda of 1.6 is an average, so it describes the whole spread of outcomes from a blank upwards rather than a prediction of 1.6 goals.
What Poisson Does With an Expected Goals Figure
The Poisson distribution takes one input, an average rate of events, and returns the probability of seeing exactly 0, 1, 2, 3 or more of them. Goals fit that shape: infrequent, spread across 90 minutes, with no fixed ceiling.
P(k) = e^-λ × λ^k ÷ k!
λ (lambda) is the expected goals for one team. Run it for the home side, run it again for the away side, and multiplying the two counts gives a probability for every square on the scoreline grid. Totals markets come from adding the lambdas and reading the combined distribution.
A Worked Example
A home side with expected goals of 1.6, away side on 1.1.
- Home scores 0: e^-1.6 = 20.2%.
- Home scores 1: 32.3%.
- Home scores 2: 25.8%.
- Total match lambda: 1.6 + 1.1 = 2.7 goals.
- Under 2.5 goals: P(0) + P(1) + P(2) across that 2.7 total = 49.4%.
- Over 2.5 goals: 50.6%.
- 0-0: e^-1.6 × e^-1.1 = 6.7%, which prices at fair decimal odds of 14.9.
The goals market sits close to even at 50.6% against 49.4%, so the price on either side of 2.5 leaves little room before the bet turns negative. And the goalless draw, which feels like a freak result, lands about once in fifteen matches at these lambdas.
Run the bookmaker’s over/under pair through the No-Vig Calculator before comparing, otherwise you are measuring your model against a price with the house cut still in it.
How to Read the Results
Read one team’s column first. At a lambda of 1.6 the likeliest home score is a single goal at 32.3%, and a 1.6 lambda does not make 2-1 the default expectation.
The totals line is what most bettors come for: adding the lambdas gives 2.7 and splits the market at 49.4% under and 50.6% over 2.5 goals. Drop either figure into the Expected Value Calculator with the price you have been offered to see whether the bet clears.
Correct-score probabilities are products of two goal counts: the 6.7% for 0-0 is the home side’s 20.2% chance of a blank multiplied by the away side’s own zero-goal probability. Divide 1 by 6.7% and fair odds come out at 14.9, the number to hold against the 12.0 on the correct-score list. The Implied Probability Calculator converts any other scoreline the same way.
Where Plain Poisson Gets Football Wrong
Poisson treats the two goal counts as independent, and football does not behave that way at the bottom of the grid. Matches finish 0-0 and 1-1 more often than independent counts predict, because a side that goes behind late chases and a side that leads late defends.
The miss looks small in probability terms and large in price terms. A correct-score price is the reciprocal of a small number, so a one-point error moves the fair odds by more than a full point on a line like 0-0.
The Dixon-Coles correction is the standard fix. It adjusts the probabilities on the low-scoring outcomes where plain Poisson is known to be wrong, and Gecko Edge applies it as part of the pipeline rather than shipping raw Poisson numbers.
The second limitation belongs to you. This page prices whatever lambdas you type in, and building 1.6 and 1.1 for a fixture is the real work: attack and defence strength, home advantage, a striker missing, a manager resting players before a cup tie. Enter a lambda 0.3 too high and every line moves with it.
Where This Sits in the Gecko Edge Pipeline
Gecko Edge starts where this page stops. The platform builds a lambda for each side, runs the Poisson grid, applies the Dixon-Coles correction, blends in market and league priors, then measures the finished probabilities against live prices across 130+ leagues. The maths runs first; the AI just puts it in plain English.
Further Reading
- All betting calculators: the full library of 34 free tools
- 7 mistakes you are making with xG football analysis
What expected goals figures should I enter into a Poisson calculator?
Build each lambda from the team’s scoring and conceding rates against comparable opposition, adjusted for home advantage, or start from published xG averages and adjust for the opponent. Ten to twenty matches is a workable sample, and early in a season the numbers are noisy, so blend in last season’s rates.
Why does the Poisson distribution underrate the draw?
It assumes the home and away goal counts are independent, and they interact. Teams protect a lead, chase a deficit, and settle for a point late on, which clusters real results on 0-0, 1-0, 0-1 and 1-1 more tightly than the maths allows. Dixon-Coles adjusts those low scorelines, and Gecko Edge runs it on every fixture.
Can a Poisson calculator predict the correct score?
No. It returns a probability for each scoreline, and even the likeliest one carries a small share of the total. At lambdas of 1.6 and 1.1 the 0-0 line holds 6.7%, so its fair price is 14.9 and the rest is spread across dozens of other results. Use the grid to find prices above fair value.
How do I turn a Poisson probability into odds?
Divide 1 by the probability. The 6.7% chance of 0-0 becomes 1 ÷ 0.067 = 14.9 in decimal odds, and the 50.6% for over 2.5 goals converts the same way. Compare the result with the market price, remembering that the bookmaker’s number carries margin and yours does not.