1

i've been trying to come up with a way to measure attack and defense of a team in terms of goals scored and received.

The only thing that has crossed my mind is to divide the number of goals scored/recieved by the total number of goals scored/received. So if for example a team scored 10 goals in the last 5 matches and the total number of goals scored is 100 that would mean its attack is equal to 10%.

However this seems to be too elementary and i am not sure if it will be of practical use... Can someone propose something better?

Sotiris
  • 23
  • 1
  • 3
  • There are *many* questions related to this on our site. I invite you to try out our search functionality. [Here is one example with some relevant "magic" words.](http://stats.stackexchange.com/search?q=Dixon+Coles) Less-magic words also work. :-) – cardinal Oct 18 '14 at 15:15
  • Hello and thank you for your comment. I did search about similar questions but the thing is that most of them answer the question how to model soccer scores in order to make a prediction about the score. I am not interested in the score but rather in defining a measure to quantify defense and offense ability based on the score. In most of the answers they propose the poisson distribution. I am not sure it can be helpful in my case. – Sotiris Oct 18 '14 at 15:47
  • 1
    Your question is still answered by those if you avert your eyes from the predictions and, instead, focus on the *interpretation* of the model parameters. For example, in the Dixon & Coles paper and simpler GLM versions, the parameters can be interpreted as offensive and defensive affinities/strengths. See also [near the end of this answer](http://stats.stackexchange.com/a/16148/2970). – cardinal Oct 18 '14 at 15:54

0 Answers0