02-23-2023, 01:14 AM
(This post was last modified: 02-27-2023, 04:20 PM by xenosthelegend. Edited 3 times in total.)
Hi everyone!
Recently I put out a spreadsheet gizmo you can find in this post.
This allows you to look up individual statistics for a player in a given season, by automatically scanning play-by-play data for that player's name. I was very excited after I figured out how to make it and really could not wait to release it. This was a little irresponsible of me, as I really did not know the parameters of what might constitute a strong individual performance for a certain position group, and thus I could not adequately communicate whether an individual's EPA/play were good, or bad, or even meaningful. This became apparent when I was asked:
My initial response was that a normal EPA/play range would generally settle within ±0.3 EPA/play. This is absolutely true, if you are considering the entire offense. My response was followed up with a more specific question about Kaepercolin's S39 EPA/play which sits at 0.5996439976. At first glance I waved it off as a statistical anomaly, and I made some significant changes to my calculations in order to reduce statistical anomalies, only to go back and find that Kaepercolin's EPA/play was still way larger than what I thought was possible. When I went through some more wide receivers, I noticed that many of them had a very high EPA/play, similar to that of Kaepercolin's.
I did not seriously consider that individual players could significantly break those ±0.3 EPA/play bounds, and I certainly did not expect the majority of a position group to. I should have looked into this before releasing anything and for that I am really sorry. I am particularly sorry to @br0_0ker for giving incomplete information in my response.
To make up for it I went through the EPA data for all players to make at least 30 receptions during S39, and found some helpful information. As it is, I can't really look up individual player statistics en masse, I have to look it up one by one and manually. If I could scan multiple player's EPA/play at a time, I would probably extend my data to include more seasons, but honestly I'm fairly happy with what I've found. Please enjoy!
Wide Receivers:
As you can see, many wide receivers had an excellent EPA/target. I do not know why exactly this is. It's possible that wide receivers are targeted in favorable situations, unlike tight ends and running backs, who are checked down to when the quarterback is pressured. It also looks like there is an abundance of high-TPE wide receivers, while there are only a few high TPE tight ends and receiving backs. Either way, wide receivers are very very efficient when targeted. Unfortunately I was not able to find similar data to compare to the NFL (during a five-minute search, it is very likely that the information is out there), so I do not know if this is ISFL specific or just the nature of football.
Tight Ends:
Tight ends really just don't have it quite as good as wide receivers, I guess, because while most wide receivers have an EPA/target well above 0.2, most tight ends have a negative EPA/target. There are a few players that have found success, and my guess is that if there were more available high-quality tight ends we would see better numbers from the group. Sadly there are just not so many!
Running Backs and Fullbacks:
Something had must have been going on in OCO to cause them to have two backs with high completion totals that combine to make up the #1 and #2 least efficient receiving backs. Either way, there seems to be more of a straightforward divide between the inefficient backs and the efficient ones (shout out to Harbinger-Marjin, who manages to exist almost exactly at zero), which leads me believe that many of the top backs in this category are of the Receiving Back Archetype. I haven't bothered to look, though.
That's all I've got, so thanks for reading!
~Jenni
EDIT: I should mention that despite all of this, I still do not believe that EPA/target is a particularly meaningful stat. Ultimately, more than some other stats, EPA is a team stat and should be considered as such. Additionally, many thanks to @jeffie43 for teaching me how to make plots in R.
Recently I put out a spreadsheet gizmo you can find in this post.
This allows you to look up individual statistics for a player in a given season, by automatically scanning play-by-play data for that player's name. I was very excited after I figured out how to make it and really could not wait to release it. This was a little irresponsible of me, as I really did not know the parameters of what might constitute a strong individual performance for a certain position group, and thus I could not adequately communicate whether an individual's EPA/play were good, or bad, or even meaningful. This became apparent when I was asked:
Quote:br0_0ker Wrote:
what are the ranges for EPA?
My initial response was that a normal EPA/play range would generally settle within ±0.3 EPA/play. This is absolutely true, if you are considering the entire offense. My response was followed up with a more specific question about Kaepercolin's S39 EPA/play which sits at 0.5996439976. At first glance I waved it off as a statistical anomaly, and I made some significant changes to my calculations in order to reduce statistical anomalies, only to go back and find that Kaepercolin's EPA/play was still way larger than what I thought was possible. When I went through some more wide receivers, I noticed that many of them had a very high EPA/play, similar to that of Kaepercolin's.
I did not seriously consider that individual players could significantly break those ±0.3 EPA/play bounds, and I certainly did not expect the majority of a position group to. I should have looked into this before releasing anything and for that I am really sorry. I am particularly sorry to @br0_0ker for giving incomplete information in my response.
To make up for it I went through the EPA data for all players to make at least 30 receptions during S39, and found some helpful information. As it is, I can't really look up individual player statistics en masse, I have to look it up one by one and manually. If I could scan multiple player's EPA/play at a time, I would probably extend my data to include more seasons, but honestly I'm fairly happy with what I've found. Please enjoy!
Wide Receivers:
As you can see, many wide receivers had an excellent EPA/target. I do not know why exactly this is. It's possible that wide receivers are targeted in favorable situations, unlike tight ends and running backs, who are checked down to when the quarterback is pressured. It also looks like there is an abundance of high-TPE wide receivers, while there are only a few high TPE tight ends and receiving backs. Either way, wide receivers are very very efficient when targeted. Unfortunately I was not able to find similar data to compare to the NFL (during a five-minute search, it is very likely that the information is out there), so I do not know if this is ISFL specific or just the nature of football.
Tight Ends:
Tight ends really just don't have it quite as good as wide receivers, I guess, because while most wide receivers have an EPA/target well above 0.2, most tight ends have a negative EPA/target. There are a few players that have found success, and my guess is that if there were more available high-quality tight ends we would see better numbers from the group. Sadly there are just not so many!
Running Backs and Fullbacks:
Something had must have been going on in OCO to cause them to have two backs with high completion totals that combine to make up the #1 and #2 least efficient receiving backs. Either way, there seems to be more of a straightforward divide between the inefficient backs and the efficient ones (shout out to Harbinger-Marjin, who manages to exist almost exactly at zero), which leads me believe that many of the top backs in this category are of the Receiving Back Archetype. I haven't bothered to look, though.
That's all I've got, so thanks for reading!
~Jenni
EDIT: I should mention that despite all of this, I still do not believe that EPA/target is a particularly meaningful stat. Ultimately, more than some other stats, EPA is a team stat and should be considered as such. Additionally, many thanks to @jeffie43 for teaching me how to make plots in R.