Jump to content

Advanced Metrics and Actual Points

Josh Hancher

If you are reading this, then I know that at the very least you have a basic understanding of the analytics and metrics we discuss on the show and on this blog.  

Success rate is percentage of plays which gain a certain number of yards based on down and distance.  50% of yards to gain on 1st down. 70% on 2nd down, and 100% of yards gained on 3rd and 4th downs.  Offense wants a high percentage and Defense wants a low percentage.  It is binary and does't offer any context.  A contextualized metric is the EPA (Expected Points Added) which factors in down, distance, and field position.  But, for this post I am going to use the success rate and actual points scored and attempt to use as a predictor.  I did this a little last year (came up with a 45-0 score v Tech, wildly off for SECCG, and if Ringo takes a knee - my CFP final score prediction was pretty good 28-23 and nailed the total)

I based those on the how many successful plays each team would have and averaged points scored each team had based on those successful  plays then adjusted for opponent.  Georgia ran 31.4 successful plays per game in 2021, and they averaged 1.2 points per successful play. I adjusted than 1.2 down to .8 (Alabama allowed .7 points per successful play on defense) for the Alabama defense.  Alabama average 35.3 successful plays on offense per game and averaged 1.3 points per successful play.  Made that adjustment .7 for the Georgia defense and also adjusted that total play to 24.

Georgia ran fewer offensive plays than season average and scored 1.1 points per successful play (not including the Chip 6) and Bama scored .6 per successful play.   I was on the right team and had one team really close (.7 predicted to .6 actual for Bama). 28-23 was my prediction. Look at that total ! 

Here is a chart of 11 win teams dating back  to 2016.  Very interesting to see how closely pointers per successful play mirrors the offensive explosiveness but the defensive explosiveness is very noisy.  Explosiveness is the EPA of successful plays.  

Screen Shot 2022-07-24 at 2.07.39 PM.png

What is clear to me that while Net YPP doesn't immediately jump out as correlated to Points per and Points Allowed per successful play, teams with 1.7 Net YPP is a clear indicator of good to playoff quality and 2.0 is getting to elite.  

Here is what I am working on next.  Using Net YPP to predict the better team.  Then adjusting the Points per offensive successful play for each team with the "better" team getting 1.0 to 1.2 per play and the weaker team getting .5 to .9 depending on the quality of defensive success rate for the better team and the difference in Net YPP between the two teams.  

Example.  Tech and Ole Miss play week 3.  Ole Miss has a 3 year average of 45% offensive success rate and 33 successful plays a game and 1.0 offensive points and allowing .7 per successful play on defense.   I would leave that 1.0 per offensive play and estimate 29 to 33 successful plays for Ole Miss on offense.  29-33 points.  I would downgrade the Tech offense to .6 and based on a 3 year average success rate of 41% on offense and 22 successful plays per game give them .6 points on 19 to 24 successful plays for a estimated score of  11 tp 14 points.


Ole Miss by 15 to 22 points. 

Edited by Josh Hancher



Recommended Comments

There are no comments to display.

  • Create New...