Football and statistics

I always wanted to know how football odds are calculated and this is why I started using some of my free time for football statistic analysis in this summer.

One interesting bet is the “do both teams score?”-bet as it is binary (and not a three item bet); so one of the things I wanted to look at is the distribution of the goals scored and conceded. I wanted to use the (normalised) Gini-Index to do this (it is one way, there are others as well). The results are present in the following table:


The interpretation is as following: A high Gini-Value indicates a highly unequal distribution. For example: Bayern has a Gini-Value in conceding of 0.75, which can be interpretated as “The overall goals they conceded were highly unequally distributed to just some very few games”

Looking at Wolfsburg: “The overall goals they scored were quite equally distributed over all games”.

Looking at Dortmund and Frankfurt, these were good teams for the explained “do both teams score?”-bet as they conceded their goals quite equally over all games. Please note that this is not about the total amount of goals or conceded goals. That is why Paderborn and Hamburg for example do have quite high scoring values for the Gini in the first column.

Another statistic I wanted to look at were the ratios of “Shots on Target/Total Shots” (as a measure how well a team might get into promising scoring positions) and the “conceded shots on target/total shots” (as a proxy for the defence). Afterwards I just took the differnces in those numbers to gain a nice plot that you can see in the following figure.


Gladbach did really well, nearly as good as Bayern. Also: Dortmund ist a quite interesting team in this kind of analysis. Their values in this plot as quite close to Paderborn, but they still manage to come up 7th.

Just looking at some plain numbers, they had 17 shots per game (median) which is 1st in the league (with Bayern). Also their median of shots conceded is 7.5 (2nd best in the league, only with Bayern on 1 with a median of 7 shots conceded per game). I think having the poor conversion rates in mind, they just had a lot of bad luck in the recent season. So probably they do not need to change that much after all,while the players still manage to win a good amount of games (maybe more due to individual skill).