My last post outlined the origins of PDO and why it is useful in hockey. Now I’m going to apply it for football in the same way I have looked at shooting and save percentage. So firstly here’s a plot of PDO against a teams shots on target ratio.
This mirrors the lack of a pattern in shooting and save percentage ie better teams do have a marginally better PDO, but the effect is small (best fit line has a range of 970-1030) and almost entirely random (low r2 value). In terms of the distribution of PDO values the Bell curve is below.
The standard deviation here is about 60 (which makes sense; a team one SD above the mean in both sh% and sv% would have a PDO of ~1080 so we’d expect PDO to have a SD<80). It's important to determine whether PDO is a number that teams are capable of repeating or whether it's essentially random and down to luck so next I'll look at how strongly sh%, sv% and PDO regress to the mean on a season by season basis.