Poisson Distribution: What It Means in Betting
The Poisson distribution is a probability model frequently used in football betting to estimate the likelihood of different scorelines. It is one of the most accessible statistical tools for bettors who want to move beyond gut feeling and into data-driven analysis.
Named after the French mathematician Simeon Denis Poisson, the distribution calculates the probability of a given number of events occurring within a fixed interval, provided those events happen independently and at a known average rate. In football, the "events" are goals, and the "interval" is one match.
The Formula
The Poisson probability of exactly k goals is:
P(k) = (e^(-l) x l^k) / k!
Where:
- l (lambda) is the expected number of goals (the average)
- k is the specific number of goals you want the probability for
- e is Euler's number (approximately 2.718)
- k! is the factorial of k
You do not need to calculate this by hand. Spreadsheets, online calculators, and programming languages all handle it easily.
Practical Football Example
Suppose you want to predict the scoreline for Arsenal vs Wolverhampton. Based on recent form, league averages, and home/away adjustments, you estimate:
- Arsenal expected goals: 2.1
- Wolves expected goals: 0.8
Using the Poisson formula for Arsenal (l = 2.1):
| Arsenal Goals | Probability |
|---|---|
| 0 | 12.2% |
| 1 | 25.7% |
| 2 | 27.0% |
| 3 | 18.9% |
| 4 | 9.9% |
And for Wolves (l = 0.8):
| Wolves Goals | Probability |
|---|---|
| 0 | 44.9% |
| 1 | 35.9% |
| 2 | 14.4% |
| 3 | 3.8% |
To find the probability of a specific scoreline, multiply the two probabilities. For a 2-0 Arsenal win:
27.0% x 44.9% = 12.1%
This means the Poisson model estimates roughly a 12% chance of a 2-0 Arsenal victory. If a bookmaker offers correct score odds that imply a probability lower than 12%, the model suggests value in that market.
Building a Full Scoreline Grid
By calculating the Poisson probability for each combination (0-0 through to, say, 5-5), you can build a complete scoreline probability grid. Summing the relevant cells gives you probabilities for common markets:
- Match result (1X2): Sum all cells where Arsenal score more (home win), both score equally (draw), or Wolves score more (away win).
- Over/Under goals: Sum all cells where the total exceeds 2.5 for over 2.5, and all cells where the total is 2 or fewer for under 2.5.
- BTTS: Sum all cells where both teams score at least one goal.
Estimating Expected Goals
The quality of a Poisson model depends entirely on the accuracy of the expected goals input. Common approaches include:
- Using historical scoring averages adjusted for home and away performance
- Incorporating expected goals (xG) data from providers like Opta or FBref
- Weighting recent matches more heavily than older fixtures
A simple method uses league averages. If the Premier League average is 1.4 home goals and 1.1 away goals per match, adjust these based on Arsenal's attacking strength and Wolves' defensive record relative to the league average.
Limitations of the Poisson Model
The Poisson distribution is a useful starting point, but it has notable weaknesses.
Independence assumption. The model assumes each goal is independent, but in reality, a team going 1-0 down may change their approach, affecting the probability of further goals.
No in-game context. Red cards, injuries, substitutions, and tactical shifts all influence goal rates but are not captured by the model.
Underestimates draws. Research has shown that the basic Poisson model tends to slightly underestimate the frequency of draws, particularly 0-0 results.
Static expected goals. The model uses a single average, but a team's goal threat can vary significantly depending on the opponent and match context.
More sophisticated approaches, such as bivariate Poisson models or Dixon-Coles adjustments, address some of these issues but add complexity.
Past performance does not guarantee future results. The Poisson model provides probability estimates, not certainties.
18+. Gambling involves risk. Please gamble responsibly. If you need support, visit begambleaware.org.