New Metric: WPC+

(article repurposed from July 1, 2021 Patreon Post)

One of the toughest things to do in college football is to compare players from team to team, conference to conference and across the country. Of course, you could look at a metric like yards per rush to compare a running back, but that leaves out a significant component of that players evaluation.

Things such as down and distance and game situation are completely ignored when looking at a gross stat, such as yards per rush. If you average 5 YPR, when did you accumulate those yards? Was it 12 yards on 1 run on 3rd and 25, 2 on 2nd and 10 and 1 on 1st and 10? Or was it 12 yards on 2nd and 8, 2 on 2nd and 1 and 1 on 4th and 1? Those runs should be graded completely different, and with a stat called PPA from the gents at collegefootballdata.com, they now are.

What is PPA?

Essentially, this factors in cfbdata's win probability before a play and after in determining the value that a player added to the win chance for their team. For example, if on 3rd and 12 in the 3rd quarter from your own 20 yard line you are down 4 points, lets assume your win probability is 42%. If you rip off an 80 yard TD and are now up 3, your win probability is now 49%. That run added 7 points. You are able to utilize this statistic over a long time horizon to consistently judge a plyers value vs that of the expectation.

Why WPC+?

Once again, you can look at gross PPA and get a pretty solid idea of how good a player is - the best players normally have a much higher value than players that aren't so good! But PPA is a gross stat.

If you know baseball, there is WAR, which is Wins Above Replacement. Sound similar It's the same thing as PPA. The problem with it, is the more plays you play, the better your WAR would be, all else being equal.

In baseball, WRC+ was created to calculate how much better a player is than your average hitter. A player with a WRC+ of 125 is graded to be 25% better than the average MLB hitter and a player with a WRC+ of 60 is 40% worse.

WPC+ is a mimic of WRC+, but for college football - obviously with different inputs and variables.

How to Calculate WPC+?

We have 3 data sets that we will use as the core of the formula: Total PPA, Total Countable Plays and in turn, PPA Per Play.

We will adjust both average PA and Total PPA by the SOS that each player faced, as it's unfair to compare the schedule of Alabama's to that of Toledo.

Once we adjust the gross PPA and PPA/Play by the SOS faced, we then take the difference to average by position and use 35% of the per play PPA and 65% of the gross PPA. That ultimately creates the WPC+ and tells us how much better than average, via PPA, a player was - adjusted for schedule faced.

How to use WRC+?

Outside of obvious uses in DFS, this metric can be used to assess a players value to their team. We frequently hear stats like yards per game or completion % that are utilized in broadcast or analysis and while every stat has their value, this will help us put the stats into context of the gameflow.