## Which metrics are the most repeatable between the first 9 and final 29 games of the season?

A quick jargon buster

TS = Total shots
ST = Shots on target
G = Goals

R = ratio
D = difference
F = for
A = against

Sh% = shooting percentage (goals/shots on target for)
Sv% = save percentage (saves/shots on target against)
PDO = 1000*(Sh% + Sv%)

Shots tell you more about the succeeding 29 games than shots on target, which in turn are more informative than goals (though this could have been inferred from my previous work). Attacking metrics are more repeatable than defensive ones (that’s an interesting nugget of information!). The points scored by a given team are pretty randomly distributed throughout a season – Arsenal won’t continue to score 2.44 points per game, and nor will Palace continue to score 0.33. Read very little into a teams Sh%, Sv%, or PDO thus far, they’re all likely to regress heavily towards their individual means over the remainder of the season.

So, are the numbers above reflected well when we look at how well a given metric over the first nine games does at predicting the points a team will score over the succeeding 29 games?

The order at the top looks similar – TSR and STR rule the day, and attacking metrics are still more important than defensive ones. Goals jump up a bit, but not to the point that they’re a better predictor than shots. The luck based metrics are still terrible.

In short, if you want the best indicator out of this group as to how well a team will do over the remainder of this Premiership season then you should be looking at its TSR to this point – a copy of the table I’ve posted below. To see how the Premiership teams stack up by a few different metrics, including TSR that has been adjusted to take the strength of a teams opponents into account, then see here.