Poisson Football Statistical Modelling with Independence

163 Views Asked by At

I am familiar with this basic Poisson model for modelling the outccome of a football match: https://help.smarkets.com/hc/en-gb/articles/115001457989-How-to-calculate-Poisson-distribution-for-football-betting

This model assumes that the goals scored by team A (say the home team) is independent to the number of goals scored by team B (the away team), and we multiply the Poisson PDFs together.

Is there any way to incorporate dependence into this, it would seem natural that there is a dependence between the two, are there any alternatives to just multiplying the models together?

1

There are 1 best solutions below

0
On BEST ANSWER

The model incorporates interdependence of the two teams in the calculation of average goals. What it does not incorporate is a claim that if A scores a lot of goals you should expect that B scores more or less goals than average. Maybe A gets tired scoring all those goals and is so far ahead it allows extra goals to B. Maybe B gets discouraged and doesn't try very hard. If you believe close games are more likely, you could multiply all the entries in the table by some factor, say $\frac {1.2}{1+\text { goal differential}/10}$, then add up all the values and divide by the total to normalize the probability. Whether that improves the prediction is left as an exercise.