The Poisson distribution, to model football goals
Definition
It was developed by a French mathematician. The Poisson distribution displays the number of times that
an event is likely to happen in a specific time. Also, the time intervals between successive events are
independent.
Formula
This example uses the data from the 2022-2023 English Premier
League to calculate a hypothetical match between Liverpool and
Chelsea. Chelsea is the home team, while Liverpool is the away team.
Goals Away Goals Home Goals Goals Aways Goals
scored at by all EPL by Chelsea allowed at Goals by allowed
home by all teams home by Liverpool away by
EPL teams Chelsea Liverpool
Goals 600 456 58 12 35 12
Matches 350 350 20 20 20 20
Goals/Match 1.714 1.303 2.9 0.6 1.75 0.6
Before calculating these, we need to know:
The total home goals scored by all EPL teams
The total away goals scored by all EPL teams
The average number of home goals and away goals per match for
the whole league
We need to calculate Chelsea’s:
Home goal average
Average goals allowed per home match
We need to calculate Liverpool’s:
Away goal average
Average goals allowed per away match
Calculating Attack Strength
Attack Strength is the team’s average number of goals, divided
by the league’s Average number of goals.
Home
Chelsea’s Attack Strength: 2.9 ÷ 1.71 = 1.69
Away
Liverpool’s Attack Strength: 1.75 ÷ 1.3 = 1.34
Calculating Defence Strength
Calculating Defence Strength is the team’s average number of
goals allowed, divived by the league’s average number of
goals allowed.
Chelsea’s Defence Strength: 0.6 ÷ 13 = 0.46
Away
Liverpool’s Defence Strength: 0.6 ÷ 1.7 = 0.35
Goal Expectancy
We have found out each team’s Attack Strength and Defence Strength,
we can calculate each team’s expected score
Chelsea goal expectancy
= Chelsea’s Attack Strength ×Liverpool’s Defence Strength×
League’s Average Home Goals
= 1.69 * 0.35 * 1.714 = 1.013
Liverpool goal expectancy
= Liverpool’s Attack Strength × Chelsea’s Defence Strength
League’s Average Away Goals
= 1.35 * 0.46 * 1.303 = 0.81
Average goals scored in the match
Chelsea: 1.013
Liverpool: 0.81
These probabilities indicate the likelihood of each team scoring a certain
number of goals, based on the given goal expectancies. For instance, there
is a 36.78% chance that Chelsea will score exactly one goal, while Liverpool
has a 44.49% chance of not scoring any goals.
Assume we have calculated the following goal expectancies:
Chelsea’s Goal Expectancy = 1.013
Liverpool’s Goal Expectancy = 0.81
We would apply the Poisson formula to calculate the probabilities of Chelsea
and Liverpool scoring 0, 1, 2, 3, etc., goals.
Then, to calculate the probability of a specific scoreline, say Chelsea 1 - 0
Liverpool, we multiply the probability of Chelsea scoring exactly 1 goal by
the probability of Liverpool scoring exactly 0 goals.
Let’s calculate the probabilities for the 0-0, 1-0, 0-1, 1-1, 2-1, and 1-2
scorelines using the given goal expectancies for Chelsea and Liverpool.
Therefore, 1- 0 scoreline is likely to happen.
The Bivariate Poisson distribution and the stacked Bayesian regression model, to model score
difference in a specific match.
Bionomial Distribution
Ronaldo is about to take 5 penalty kicks in a shootout. What is the probability that
Alex will score exactly 3 out of 5 penalty kicks? (Knowing that Ronaldo has a 75%
chance of scoring a goal on any given penalty kick.)
