College Football Stats Explained

There are a lot of advanced college football stats out there—it can be confusing and even intimidating without a glossary. Here is your definitive college football statistics guide, with definitions to some of the most common advanced college football stats.

Jump To:

Team Statistics

FEI – Fremeau Efficiency Index


  • Opponent-adjusted
  • Per-possession efficiency
  • Gives the scoring advantage per possession against an average opponent on a neutral field

FPI – Football Power Index


  • Developed by ESPN
  • Designed to measure team strength and project it going forward
  • ESPN’s stated goal is not to rank teams but to correctly predict games and seasons
  • Broken down into offense, defense, and special teams contributions
  • Represents how many points above or below average a team is

FPI is a number that can be positive or negative, and can be interpreted as the number of points a team would win (or lose) by against an average opponent on a neutral field.

Example: Ohio State’s FPI is 30. So, against the most average team in the FBS, they would be expected to win by 30 points on a neutral field (say, in a bowl game).


Elo Ratings are Staturdays’ main metric for rating teams and predicting win probabilities for single games and win totals for the season.

This simple metric looks at a few factors in the result of a game to increase or decrease a team’s rating each week.

Each Power-5 team starts at an Elo rating of 1500 at the start of the 2000 season, and then that score goes up or down based on wins and losses against other teams. At the end of the season, every team is brought back 5% closer to the league-average of 1500 again, to account for changes in the offseason.

You can read more about Elo ratings and how we calculate them in detail here.

Game Statistics

Win Probability

We use our Elo ratings to calculate the probability each team has to win their game. We add in a slight home field advantage as well that accounts for an extra 8% increase in win probability for two otherwise evenly matched teams.

Play Statistics

EPA – Expected Points Added


  • Per-play stat
  • Measure of points added (or lost) on each play compared to the expected value
  • Accounts for the expected points for the remainder of the drive, and the opponent’s next drive
  • Takes into account down, distance to go, yardline


Expected Points Added, or EPA for short, is the number of points added or lost on any given play, vs. the expected number of points that most teams score (or give up) in the current drive (if they score) or next drive (if they turn the ball over). It is based on an average of all teams in similar down and distance situations.


On 4th and 10 from your own 20 yard line, the expected points are -0.32 points, meaning that on average, teams in this position on the field will give up an average of .32 points to the opponent within the next drive.

Let’s say this team ends up going for it on this 4th and 10 (they’re desperate) and break off an 80 yard TD run. The result of the play is 6 points, vs. and expectation of -0.32 points, leading to an EPA of +6.32 on that particular play. The team scored 6.32 points more than expected.

There doesn’t need to be a score on the play to get an EPA. If that same team gains 15 yards and gets a 1st down, their new expected points from 1st and 10 at the 35 is 1.76, a 2.08 point improvement from the previous play. +2.08 is their EPA for that play.

PPA – Predicted Points Added


  • Per-play stat
  • Measure of points added (or lost) on each play compared to the predicted value
  • Accounts for the predicted points for the remainder of the drive, and the opponent’s next drive
  • Takes into account down, distance to go, yardline


This is the same as EPA with one key difference: this stat uses predicted points instead of expected points. All this means is that is uses projected points instead of the actual average points scored on a drive. If you think back to stats class, expected value is just a fancy word for the average, or mean. Meaning, take all the results on all the drives with all the same situations, and then take the average. For PPA, we use a variety of stats to predict how many points the drive will end in. It uses a neural network model instead of historic data.

Win Probability Added (WPA)


  • Using our in-game win probability model, this is the difference between a team’s win probability pre-play and post-play.
  • This can be used to see whether any individual play helped a team’s chances of winning or hurt them


You can look at any team and see how they perform in different situations. For instance, you could look only at plays where Penn State goes for it on 4th down, and then see what their average win probability added (WPA) is. If it’s positive, than that means they tend to improve their chances of winning the game by going for it – either because they execute well or because they go for it in the right situations. You could then compare that to the average of all college football teams to see if they’re better or worse than average on 4th down plays.

Line Yards


  • Way of measuring the Offensive Line’s contribution to the run-game
  • Simple weighted measurement of run plays


Line Yards is just a simple calculation that gives you an idea of how much the Offensive Line had to do with the success/failure of a run. The theory is that the first few yards of a run are mostly thanks to the gap opened up by the offensive line, but then anything beyond 5-10 yards is mostly due to the skill and speed of the running back. Therefore, the formula is…

For runs that lose yards:

Line Yards = 1.2 * Yards Gained

For all positive runs:

Yards 0-4 = 1 * Yards Gained

Yards 5-10 = 0.5 * Yards Gained Yards 11+ = 0

So you apply the calculation to pieces of the entire run.

An example is a 12 yard run by the running back. The rushing yards stat equals 12 yards. The Line Yards stat will equal 1 * 4 + 0.5 * 6 = 7, which is the max value you can get for the line yards stat on any given run play.

In R Code, it would look a bit like this:

line_Yards = 
             rush == 1 & yards_gained < 0 ~ yards_gained * 1.2,
             rush == 1 & yards_gained >= 0 & yards_gained <= 4 ~ yards_gained * 1,
             rush == 1 & yards_gained > 4 & yards_gained <= 10 ~ ((yards_gained-4) * 0.5) + 4,
             rush == 1 & yards_gained > 10 ~ 7,
             TRUE ~ 0

Explosiveness (Staturdays Definition)


  • Fairly subjective stat
  • This is the rate of rushes or passes that are in the top 10% of yards gained.
    • For pass plays, this will work out to be about 19 yards or more
    • For run plays, this is about a 12 yard gain or more
  • One of the “five factors” as described by Bill Connelly

Explosiveness (EPA)


  • Average EPA on “successful” plays (see Success Rate)
  • A measure of “how good were your good plays?”

Success Rate


  • A simple way of determining whether a play was successful or not, based on the percentage of yards to go gained.

A play is deemed “successful” when:

  • At least 50% of the yards-to-go are earned on 1st down (i.e. 5 yds. on 1st-and-10)
  • At least 70% of the yards-to-go are earned on 2nd down (i.e. 7 yds. on 2nd-and-10, or 5 yds. on 2nd-and-6)
  • 100% of the yards-to-go are earned on 3rd and 4th down (i.e. you get a first down)
  • A Touchdown would also be considered a successful play, no matter the yards-to-go