January is waning and February is approaching, which means college boosters across the country sit nervously awaiting national signing day in college football. After successfully using analytic’s to correctly pick over 70% of the College Bowl Games I decided to use some of those same numbers to research recruiting’s affect on teams ratings. It is generally understood that better recruiting ratings result in better team ratings, though the extent is unclear.

I use College Football Reference’s SRS ratings for my team rating and 247 Sports for my recruiting ratings. My time frame for his analysis was the lest five college football seasons. I took 247 Sports recruiting rankings for each team over the previous five seasons and weighted them based on the average number of starters in each class. The weightings were based on a rough analysis of the average number of starters of each class across all teams in Division 1 (Freshman, Sophomore etc.) and were as follows: Freshman-3%, Sophmore-18%, Junior-29%, Senior-33%, Fifth Year Senior-17%.

So the formula to find a teams 5 Year Weighted Recruiting Rating for 2015 would be as follows:

*2015 Weighted Recruiting Rating = (.03*2015 Recruiting Rating)+(.18*2014 Recruiting Rating)+(.29*2013 Recruiting Rating)+(.33*2012 Recruiting Rating)+(.17*2011 Recruiting Rating)*

I found this number for every college team over the past five seasons and uploaded that data into R. Using R’s plot function I plotted all the data points and added a linear regression line using the lm function. The formula generated by the regression line was:

*SRS Rating = -11.89274 + 0.08631(5 Weighted Year Recruiting Rating)*

The graph can be seen in the figure below.

However I noticed an unusual occurrence in the graph. There was a large clumping of teams with low recruiting ratings and an SRS rating of zero, as you can see in the graph below. This occurred because College Sports Reference does not provide ratings for non-D1 teams and schools like Charlotte, Old Dominion and others recently made the move from 1AA to 1. These teams had no SRS rating and were therefore given a rating of 0. However, College Football Reference sets it SRS rating so zero would be the average team. Therefore 1AA teams were being given the ratings of average D1 teams.

This error causes the regression line to flatten because teams with low recruiting ratings were given higher SRS ratings than expected. Therefore I removed all instances of this occurrence from the data set and replotted which produced the following graph.

The new regression line formula is a follow:

*SRS Rating =-15.5987 + 0.1066(5 Year Recruiting Rating)*

There are many other factors that affect a teams SRS rating (injuries, coaching etc.) however this equation can be used to determine a teams *expected *rating. I am currently working on a prediction model that will use this equation among other factors to predict a teams win total for 2016.

## One comment