Not related to this post but do you plan on doing updates to the all time QB rankings ever? I'm sure this wouldn't be easy but would be cool to see it as a live standings page to see how active players move up and down.
Is EPA adjusted across different websites? When you look at total EPA on NFL Pro, FTN, RBSDM, Elo and your site, all seem to provide different calculations because the EPA doesn't match. Some sites have Lamar leading and others have Josh Allen. It's curious, because I thought unadjusted EPA followed one formula. Am I mistaken?
It's a model calculating the pre- and post-play expected points to come up with those added on each play. Different places have different models. I use the same one was RBSDM, with some additional adjustments, like decreasing the loss on aborted fumbles.
You might also be seeing EPA per dropback in some places versus per play, the latter including designed runs. Per dropback, Jackson is leading.
Not related to this post but do you plan on doing updates to the all time QB rankings ever? I'm sure this wouldn't be easy but would be cool to see it as a live standings page to see how active players move up and down.
Plan is to tackle that after the draft this summer
Do you know what the primary differences between your adjustments and Ben’s are? Picking two qbs with some of the larger adjustments an example.
In his model Goff and Lawrence get adjusted to basically the same EPA per play but here they are still pretty far apart.
Directionally you both adjusted them the same but was much larger in magnitude.
Just curious since it seems like you are both adjusting for similar things.
Is EPA adjusted across different websites? When you look at total EPA on NFL Pro, FTN, RBSDM, Elo and your site, all seem to provide different calculations because the EPA doesn't match. Some sites have Lamar leading and others have Josh Allen. It's curious, because I thought unadjusted EPA followed one formula. Am I mistaken?
It's a model calculating the pre- and post-play expected points to come up with those added on each play. Different places have different models. I use the same one was RBSDM, with some additional adjustments, like decreasing the loss on aborted fumbles.
You might also be seeing EPA per dropback in some places versus per play, the latter including designed runs. Per dropback, Jackson is leading.