How to Use Poisson Distribution to Predict Football Results
The Poisson distribution is one of the most widely used statistical tools in football modelling. It allows analysts to estimate the probability of different scorelines — and therefore match outcomes — based on a team's expected goal-scoring rate. While no model perfectly captures the complexity of a football match, the Poisson approach provides a structured, mathematically grounded framework for thinking about results.
This article explains what the Poisson distribution is, why it is well-suited to football, and how to apply it step by step using a real match scenario.
What Is the Poisson Distribution?
The Poisson distribution is a probability model that describes the likelihood of a given number of events occurring in a fixed interval, when those events happen independently of each other at a known average rate.
In football, the "events" are goals, and the "interval" is the match. The model asks: if a team is expected to score an average of, say, 1.6 goals per match, what is the probability they score exactly 0, 1, 2, 3, or more goals in a specific game?
The Poisson formula is:
P(k) = (λ^k × e^−λ) / k!
Where:
- P(k) is the probability of exactly k goals being scored
- λ (lambda) is the expected number of goals (the average goal rate)
- e is Euler's number, approximately 2.71828
- k! is the factorial of k (e.g. 3! = 3 × 2 × 1 = 6)
This looks more complex than it is in practice. Let's work through an example.
Why Football and Poisson Are a Good Match
Football goals are relatively rare events that occur more or less independently of each other during a match. One goal does not substantially change the probability of another goal in a way that the basic Poisson model cannot accommodate reasonably well.
Research has consistently shown that the Poisson distribution approximates real-world goal distributions in football well enough to be useful — particularly for matches in the 60–90 minute range in top-level professional football.
Step 1: Establish Expected Goals for Each Team
The first step is to estimate how many goals each team is likely to score. This is not simply a matter of looking at a team's season average — it requires adjusting for the quality of the opposition.
A standard approach:
- Calculate each team's average goals scored per home match (for the home side) or away match (for the away side) over the season
- Calculate each team's average goals conceded per home match (for the home side) or away match (for the away side)
- Calculate the league average for home goals scored and away goals scored
- Apply an attack strength and defence strength calculation
The Attack and Defence Strength Method
Attack strength = Team's average goals scored ÷ League average goals scored
Defence strength = Team's average goals conceded ÷ League average goals conceded
Expected goals for home team = Home attack strength × Away defence strength × League average home goals
Expected goals for away team = Away attack strength × Home defence strength × League average away goals
A Worked Example
Suppose we are modelling a match between Atlético Madrid (home) and Sevilla (away) in La Liga.
Over the season to date:
- Atlético Madrid home: 1.8 goals scored per game, 0.9 goals conceded per game
- Sevilla away: 1.0 goals scored per game, 1.6 goals conceded per game
- La Liga average: 1.5 home goals per game, 1.2 away goals per game
Atlético home attack strength: 1.8 ÷ 1.5 = 1.20
Sevilla away defence strength: 1.6 ÷ 1.2 = 1.33
Expected goals for Atlético (home): 1.20 × 1.33 × 1.5 = 2.39
Sevilla away attack strength: 1.0 ÷ 1.2 = 0.83
Atlético home defence strength: 0.9 ÷ 1.5 = 0.60
Expected goals for Sevilla (away): 0.83 × 0.60 × 1.2 = 0.60
So our model estimates Atlético is expected to score approximately 2.39 goals, and Sevilla approximately 0.60 goals.
Step 2: Calculate Scoreline Probabilities
Now we apply the Poisson formula to each expected goal rate to find the probability of each possible scoreline.
Using λ = 2.39 for Atlético:
| Goals | Probability | |-------|-------------| | 0 | 9.1% | | 1 | 21.8% | | 2 | 26.0% | | 3 | 20.7% | | 4 | 12.4% |
Using λ = 0.60 for Sevilla:
| Goals | Probability | |-------|-------------| | 0 | 54.9% | | 1 | 32.9% | | 2 | 9.9% | | 3 | 2.0% |
To find the probability of any specific scoreline, multiply the independent probabilities together:
P(Atlético 2–0 Sevilla) = 26.0% × 54.9% ≈ 14.3%
P(Atlético 1–0 Sevilla) = 21.8% × 54.9% ≈ 12.0%
P(Atlético 3–1 Sevilla) = 20.7% × 32.9% ≈ 6.8%
Step 3: Convert to Match Outcome Probabilities
To find the probability of each match result, sum the scoreline probabilities:
- Home win: Sum of all P(Home goals > Away goals)
- Draw: Sum of all P(Home goals = Away goals)
- Away win: Sum of all P(Away goals > Home goals)
In our example (summing across all meaningful scorelines up to 6 goals per side):
- Atlético win: approximately 76%
- Draw: approximately 16%
- Sevilla win: approximately 8%
These probabilities can then be compared to bookmaker odds to assess whether there is potential value in any outcome.
Limitations and Important Caveats
The Poisson model in its basic form carries several important limitations:
It assumes goal independence. In practice, the score affects behaviour — teams that fall behind often push forward and concede more, creating a feedback loop the model does not account for.
It relies on averages over the season. Recent form, injuries, tactical changes, and fatigue may make the season average an imprecise guide to the upcoming match.
It doesn't account for match context. Motivation, cup competition versus league, and whether a team has something to play for all influence outcomes in ways the model cannot capture.
Estimates compound errors. The attack and defence strength calculations each contain uncertainty; multiplying them compounds that uncertainty.
The Poisson model is best treated as one input in a broader research process, not as a standalone system.
Building on the Model
FCResults uses predictive modelling — including Poisson-based frameworks — as part of the analytical foundation for its daily research insights. Subscribers can also access simulation tools and export data for their own analysis. If you want to explore structured football research with full transparency into historical performance, a free trial is available at fcresults.com.
Summary
The Poisson distribution gives analysts a principled way to convert expected goal rates into scoreline and match outcome probabilities. While the model has limitations and should always be supplemented with contextual research, it represents a solid mathematical foundation for anyone serious about data-driven football analysis. The key steps are: estimate expected goal rates for each team using attack and defence strengths, apply the Poisson formula to generate scoreline probabilities, and sum those probabilities to produce outcome estimates.