top of page

College Football Coaches,Want to Win 93% of your Games?

Every college football season, there are major upsets that send shockwaves across the nation. Who could forget the 2019 season, when the Georgia State Panthers strolled into Neyland Stadium as a massive 25.5 Point underdog and walked out with 38-30 victory against Tennessee. After games like that, coaches, players & fans often wonder how such an upset happens but one glance at the stat sheet and its pretty easy to see why. In fact, what Georgia State did on the field against Tennessee that day pretty much guarantees a victory over 90% of the time! What did they do? Well, we will get to that in a minute.

Georgia State Panthers Celebrate Win against Tennessee as a 25.5 point underdog

If you follow me on twitter @js_ace_football or are a regular visitor to my blog, you are aware that by using big data and advanced analytics, I am always looking to for ways to help coaches identify areas that can improve their team’s chances of winning. As a data scientist well versed in R Stats along with my previous experience as a college football coach, many of my projects come from my own experience, ideas and general curiosity. However, sometimes the best ideas for my analysis comes at the request of current college & professional coaches, as well as those across the sports media, betting & analytics landscape who are looking to gain insight into what contributes to winning football.

The inspiration for this analysis came from a conversation I had with former Western Michigan Linebacker & sports Betting enthusiast Jason Sylva @JasonSylva_ on twitter.

Recently, I recalled a conversation in which Jason and I spoke about many key statistical categories in football and how they correlate to one another. A few years back, Jason came across my blog and wanted to discuss the post about offensive categories that correlated to each other and wins. offensive

The correlation matrix I posted, plotted correlation coefficients between team offense variables, with the higher the number between 0 -100 telling you how closely the variables are related.

As you would expect a former linebacker to do, Jason quickly turned the focus of the conversation to the defensive side of the ball, specifically how he recalled one of his former coaches keeping a log of his teams win % when they hit two key statistical goals on defense. Over the years, his coach discovered that if his defense achieved these two main goals, they would win the game a high majority of the time. Those two main goals were:

  • Win the Turnover battle (+ turnover Margin)

  • Stop the Run (Keep opponent to 100 Yards of Less Rushing)

With this in mind, we thought it would be fun to put some hard-core data behind this and explore it at a much higher level. The idea was to look back across multiple college football seasons to see if we could filter out the number of times a team won the turnover battle & held its opponent to 100 yards or less rushing. Once armed with that data, we could see just how often a team won the game when accomplishing those two things.

Just like everything else, when I decide to do something, I’m going to do it big. I wanted to gather as much data as possible to give a clear answer of just how often teams are winning games when accomplishing those two things. As I got started, I realized that this was no easy task as we are talking about digging through thousands of games across multiple seasons. However, after a few sleepless nights and some roadblocks along the way, thankfully I was able to make it happen!

Using the API from, I was able to go through and scrape 4 seasons worth of game data from the 2016 - 2019 seasons. In those 4 seasons I had a total of 3,338 games, which I felt was plenty of data to run this analysis. The biggest challenge was getting the data & cleaning it to a usable format where I could compare the statistical categories of teams on a game-by-game basis. After reading the data into R, I was able to create a data frame consisting of 4 seasons worth of game data.

Creating data frame in R with 4 seasons of Game Data

After creating the data frame, I added columns to include winning the turnover margin, winning the game and 100 yards or less rushing for each game.

Adding columns for Turnover Margin and 100 Yards or Less Rushing Success

After completing the data frame across 4 seasons, I was able to calculate the number of times a team

  1. Won the game

  2. Won the turnover battle

  3. Held its opponent under 100 yards rushing

You can see from the table below that out of 3,338 total games there were 798 times that a team had a + turnover margin and held its opponent under 100 yards rushing. Out of those 798 games, that team won 744 times for a 93.23%-win rate!

93% Winning Percentage

Results Holding Opponent to 100 yards or Less but Not Winning the Turnover Battle

As you might have guessed, holding an opponent to 100 yards or less in itself is quite an accomplishment. You can see from the chart below that in the games where a defense limited its opponents to 100 yards or less rushing, they won 86.74% of those games. I'm sure many coaches would take that win % but when you win the turnover battle as well, you're looking at over a 93%-win rate. Thats what we call going from good to great!

You Not Only Win but You Win Big!

As evidence from a 93%-win rate, you are almost assured a victory when holding an opponent under 100 yards and winning the turnover battle. As you can see from the chart below, you not only win but you win big, with an average margin of victory of almost 25 points per game!

Go Get it Defense

Well coaches, you have the data and its as clear as day, but the rest is up to you. We know that winning the turnover battle and holding the opponent under 100 yards rushing gets you the win 93% of the time but how to go about doing that is the tough part. Stopping the run and creating turnovers should be a point of emphasis for every coach at every level and this analysis reinforces that!

I will continue to add to this analysis using data from 2021 and 2022 upon completion of this season but I don't anticipate much changing, as 4 years is plenty of data to know a trend when you see it. I hope you enjoyed this analysis. As always you can follow me on twitter @js_ace_football and can reach out to me by email to discuss guest appearances, blog contributions, consulting and collaborations.

131 views0 comments


bottom of page