We’re going off the beaten track this week at Differentgame, and doing exactly what we’ve said football number bloggers should step away from. We’ve made this blog look like an economics paper in order to test SPAM’s relationship with reality.
We had to learn new skills to come up with this post and are striding into uncharted territory. It might end up being total rubbish, but we’re sure someone in the community will point out any errors (please) if we do make them. In any case, we were inspired by this by @penaltyblog. Ok. Deep breath. Here goes…
First up we tested how a team’s SPAM (expected goals based on shot position) in a single match correlates with how many goals the team actually scores in the match:
We tested it on over 1500 games and with r2=0.167, the relationship between SPAM and scoring goals in one-off games looks pretty small.
Ok, so how does a team’s SPAM over a season correlate with the number of goals it scores over a season? The relationship looks a lot healthier with r2=0.759:
One thing we’ll look at in the future (when we learn how to do it) is just how many games it takes before SPAM starts to settle into a relationship with goals scored.
For now though, we’ll move swiftly on to the defensive side and test the correlation between a team’s expected goals against over a season and how many goals they actually concede:
The relationship here (r2=0.755) looks equally healthy as it does for scoring goals.
The relationship between a team’s expected goal difference and their actual goal difference (r2=0.76) looks even better:
The relationship between a team’s expected goal difference and points at the end of the season is a also healthy r2=0.681:
All in all, the SPAM way of calculating expected goals looks to have a decent relation to the usual outcomes over time. The lack of the relationship between expected goals and actual goals for one off matches, suggests that there’s a deal of other variables at play during 90 minutes football.
Whatever those variables are, they look to have been in play during Sunderland’s games since Paolo Di Canio took over. They’ve had 9 open play shots from outside the box in the last 2 games and scored 3 of them. Under Martin O’Neill they’d only scored 4 from 138 attempts of the same kind. Is this skill going to be repeatable long term? Probably not. Sunderland are going to have to muster more than the 3 efforts from the central area inside the box they’ve managed since Di Canio took over in order to sustain their improvement.
But, like the Murphy’s, us Everton fans are not bitter.
Follow us on Twitter here.