Every now and then I hear pundits refer to teams as good away teams, and there was also a lot of discussion of United’s ‘home form’ this season. I wanted to take a look at whether being better at home or away was repeatable (and thus more of a skill) or not (and thus more random). I thought the results were worth sharing but I don’t really feel that there’s a need for a long piece here – if there’s anything you’d like me to elaborate on feel free to leave a comment or ask on twitter.
A quick note before I begin – for these plots I use the terminology ‘this year’ and ‘next year’. To give an example – in ’00-01 Arsenal scored 26 more points at home than away, in ’01-02 they scored 7 points fewer at home than away. In this example ‘this year’ refers to ’00-01, and ‘next year’ refers to ’01-02. The sample size I’m using are the 221 teams that played in consecutive seasons between the ’00-01 and ’13-14 seasons (17 non-relegated teams x 13 pairs of consecutive seasons).
First lets look at points:
Teams score an average of 11 points more at home than they do away, however there’s essentially no correlation from season to season, suggesting that the variance around this number is largely randomness.
Next lets look at goal difference:
The R^2 is a bit higher, but still low. (For the record teams on average have a home goal difference that is ~15-16 goals better than their away goal difference. Further note: 11 points / 15.5 goals = ~0.71 goals per point, which is consistent with the number I’ve found in a couple of previous posts)
Finally I’ll use the rating I’ve developed (note – see appendix):
The correlation is still very low. In short, these plots suggest that there’s very likely to be no such thing as a team that is particularly good at home compared with how good they are away. Good teams are good and bad teams are bad, but I’d suggest you’re very likely to know more about a team if you consider all of its games rather than concentrating on just those it plays at home or those it plays away.
Appendix. In this case I’ve calculated separate home and away ratings for each team, where the calculation of each considers only home/away matches. The values are then calibrated as before so that the best home team in the sample has a home rating of 10, and the best away team has an away rating of 10. The worst home team has a home rating of 0, and the worst rated away team has an away reading of 0. I know this explanation is kind of ugly, but I’m struggling to think of a better way to word it right now.