Announcement

Collapse
No announcement yet.

2076: A season in (too many) numbers - part II. Team EPA

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • 2076: A season in (too many) numbers - part II. Team EPA

    A lot of people will be familiar with EPA, but in case you are not, EPA stands for Expected Points Added. It is a method for attributing a value to a team's play that quantifies something that probably everyone understands: possessing the ball closer to the opponent's goal line is more valuable than possession further away. This was first codified in the 1970's in an expected points model, and later extended to account for different context in the game state. EPA is a play level analysis which simply states did this play make it more or less likely that the offense will score?

    EPA is not directly calculable from the statistics output by the FOF8 engine, so for this purpose I trained a model on c.860k FOF8 multiplayer plays in order to be able to calculate the EP & EPA analysis many people will be familiar with from the TV coverage of football, as well as the many podcasts, YouTube channels, etc. which include an element of analytics in their analysis. This thread contains some visualization and leaderboards for team level EPA analysis. Player-level EPA will be addressed in a future thread.

    This scatter graph shows the relative strengths and weaknesses of every teams' offense and defense over the 2076 regular season.

    IFL-2076-offense-vs-defense-quadrant

    It's quite difficult to be good at offense and defense, so congratulations to those teams in the upper right quarter. Kansas's defense is intriguing, as is Texas, who are the only real defense specialist in the playoffs. Overall, the NC has more offense specialists this year, while the AC has more balanced power teams, although Vegas are an exception.

  • #2
    A typical use of EPA is to plot which teams are good or poor at passing or running the ball. Here is the breakdown for the 76 regular season.

    IFL-2076-pass-vs-rush-quadrant

    The vast majority of the playoff teams are good in at least one dimension of the offense, and many of them can move the ball through the air and on the ground effectively. STC are an interesting outlier - genuinely good at running the ball, but genuinely awful when passing. WVA are just exceptional - the best offense in any phase. CAL, COL and VAN very balanced but talented. Vegas, TIJ, MUR and POR all great passing attacks.

    Comment


    • #3
      For more information about Expected Points, see here: https://www.nfeloapp.com/analysis/ex...added-epa-nfl/

      Comment


      • #4
        Cool stuff, thanks for sharing

        Comment


        • #5
          Enjoying these Boots!
          Owner since 2034

          Winner:
          2039 2053 2056 2059 2065 2068

          Hall of Famers:

          DE Shorty Cochrane (Century Club Member)
          DE Asher Hutchins (Century Club Member)
          TE Gus Graham
          OT Freddie Covington
          RB Wendell Emmons
          OT Brian Fields
          OT Roosevelt Cunningham
          C Devante Stuart
          QB Abe Campbell
          OG Ellis Trainor

          Comment


          • #6
            Thanks, guys - appreciated!

            Comment

            Working...
            X