So, this is it! After a long time of basically doing nothing with it, I decide to finally post it. Why? Well it came up and I might as well make money from the two months I wasted my life on it a while back lol. This was the project that caused me to go inactive in the NSFL. This broke me. I did so much work on this that I just lost interest entirely. And when you see it, oh boy will you know why.
Alright, so what is Approximate Value? It is a stat created by Pro Football Reference I believe. It may have changed a bit in the meantime, but essentially it tries to create a magical number for each player, akin to WAR in baseball, that manages to compare players directly. It does so in a slightly....complex and to be honest a little naive way. Essentially, it's a very complicated formula, that I also had to slightly change to fit this league and our needs (I changed the way Offensive Line was graded for example). Here is a little taste of the formula:
team_offense_points = 100 * (team offensive points per drive) / (league average offensive points per drive)
where
offensive points per drive = (7*(rushTD+passTD) + 3*FG) / (rushTD + passTD + turnovers + punts + FGA)
But really, just read this: https://www.sports-reference.com/blog/appro...ue-methodology/ to understand it all. Trust me, it took me a while to even get it.
Regardless, after I figured it out, I also then had to a) slightly change formulas as mentioned and b) GET a ton of the data in the first place. A lot of the statistics needed for this weren't or aren't available in the file or in the index, so I literally had to go game by game and take down drive stats by hand. And that was just the beginning of the task.
This literally took me a month or two of my time, so yeah....this was extensive work haha.
I will say, this will not seem accurate to a lot of people. The issue with this stat is, it HEAVILY is influenced by team success. Their argument is, if a team isn't doing well (getting a lot of yards or touchdowns), then clearly the player can't be all that valuable. And it essentially gives each team a total amount of value, and then each player on that team gets a portion of that total value compared to other players on their team. So, it's definitely flawed but still interesting.
I will share the entire spreadsheet with all the math so you can see it and also one that just has a list of every player ranked by their Approximate Value. You have to scroll pretty far in the full one to see everything.
Enjoy:
Full spreadsheet with math: https://docs.google.com/spreadsheets/d/1CGj...dit?usp=sharing
Rankings: https://docs.google.com/spreadsheets/d/1StV...dit?usp=sharing
Alright, so what is Approximate Value? It is a stat created by Pro Football Reference I believe. It may have changed a bit in the meantime, but essentially it tries to create a magical number for each player, akin to WAR in baseball, that manages to compare players directly. It does so in a slightly....complex and to be honest a little naive way. Essentially, it's a very complicated formula, that I also had to slightly change to fit this league and our needs (I changed the way Offensive Line was graded for example). Here is a little taste of the formula:
team_offense_points = 100 * (team offensive points per drive) / (league average offensive points per drive)
where
offensive points per drive = (7*(rushTD+passTD) + 3*FG) / (rushTD + passTD + turnovers + punts + FGA)
But really, just read this: https://www.sports-reference.com/blog/appro...ue-methodology/ to understand it all. Trust me, it took me a while to even get it.
Regardless, after I figured it out, I also then had to a) slightly change formulas as mentioned and b) GET a ton of the data in the first place. A lot of the statistics needed for this weren't or aren't available in the file or in the index, so I literally had to go game by game and take down drive stats by hand. And that was just the beginning of the task.
This literally took me a month or two of my time, so yeah....this was extensive work haha.
I will say, this will not seem accurate to a lot of people. The issue with this stat is, it HEAVILY is influenced by team success. Their argument is, if a team isn't doing well (getting a lot of yards or touchdowns), then clearly the player can't be all that valuable. And it essentially gives each team a total amount of value, and then each player on that team gets a portion of that total value compared to other players on their team. So, it's definitely flawed but still interesting.
I will share the entire spreadsheet with all the math so you can see it and also one that just has a list of every player ranked by their Approximate Value. You have to scroll pretty far in the full one to see everything.
Enjoy:
Full spreadsheet with math: https://docs.google.com/spreadsheets/d/1CGj...dit?usp=sharing
Rankings: https://docs.google.com/spreadsheets/d/1StV...dit?usp=sharing