My 2022 Fantasy Football PPR Prediction Review
With the 2022 NFL season coming to a close (Go Giants!) it is a perfect time to see how well my fantasy football player predictions went. To recap, I wanted to see if I could predict NFL players’ fantasy points for the upcoming season. I used data collected from previous years to see if that could paint an accurate picture for this year. The statistics I took into consideration were different for each position. If you want to read more about how I came up with my predictions click here.
Quarterbacks
For predicting Quarterback points I used player data from last year which includes: Passing Yards, Passing Touchdowns, Passing Interceptions, Rushing Yards, and Rushing Touchdowns.
Predicting the QB Rank was difficult for the model. Some it nailed perfectly like Josh Allen and Patrick Mahomes. While others like Carson Wentz and Aaron Rodgers not so much. It can’t predict injuries so the bad Rank for Dak, Kyler, Matt, and Ryan is forgivable.
The best bang for your buck QB from this model was Joe Burrow and the worst was Matt Stafford. I had my eye on Kirk Cousins before the draft since the model predicted a higher rank than his previous year. And he outperformed that rank… by one, but outperformed nonetheless. Now here are some QBs that ended in the top 15 that the model wasn’t able to predict.
I believe this is mostly due to the large amount of QBs in the top 15 last year that ended up hurt this year. But also since the model goes off of data from last year players who were hurt last year were given a terrible rank, like Lamar.
I think the QB model is decent but should be adjusted toward favoring mobile QBs. Like Justin Fields and Jalen Hurts, as the NFL QB style shifts.
Running Backs
Here I mainly considered: Rushing Yards, Rushing Touchdowns, and Fumbles. But receptions are very important in a PPR league. In my model, I did include receiving stats as my league is PPR.
Again this article is going to sound like a broken record. The model can’t predict injuries. If you picked Jonathan Taylor first overall as many did, you got the worst bang for your buck. While picking up Josh Jacobs was a steal.
Two players coming back from injuries that didn’t make the models cut were Saquon Barkley and Christian McCaffrey. Which ranked 5th and 2nd respectively. Unexpected players were Rhamondre Stevenson and Tony Pollard who both broke the top ten. Before the draft, I was looking to grab Derrick Henery, as I thought the model was undervaluing him. Sure enough, I was correct, would have been a good pick.
I also mentioned looking at Gibson and Cook. Gibson would have hurt your team while Cook was perfectly ranked.
Wide Reciever
With the Wide Receivers, I looked at Receptions, Receiving Yards, and Receiving Touchdowns. Here’s how the season ended up.
The model was very boom or bust with the Wide Receivers. Either it was very accurate with Davante Adams, Tyreek Hill, Ja’Marr Chase, Justin Jefferson, Stefon Diggs, and Ceedee Lamb. Or entirely incorrect with Dionate Johnson and Dj Moore. Many of these other players who have lower than predicted ranks were because of injury. It seems to me that the best fantasy draft would be picking players that are the least likely to get hurt.
In August I was looking to draft Dionate Johnson or Ceedee Lamb as they seemed undervalued. I was right with Ceedee but very wrong with Dionate. Dionate did have a new Rookie QB throwing to him so that may have affected his performance and is something to look at in the future. I also said to stay away from Deebo Samuel as his repeat production would be unlikely. I was correct.
Tight Ends
I used very similar data as the Wide Receiver model to create the Tight End model.
If you want a good consistent Tight End in Fantasy Football, you need to go with the big three. That is Travis Kelce, George Kittle, and Mark Andrews. There is something to note about adding T.J. Hockenson into that mix as he finished rank 2 among Tight Ends this year. All other tight ends seem like a grab bag, you never know what you’ll get.
I mentioned looking at either Dallas Goedert or Darren Waller. Both would have given you a worse value than the average draft pick. For Tight ends id recommend grabbing the top 4 guys or waiting till the end to grab one and hope you luck out.
Caveats
Again these models are nowhere near perfect. In fact, they can only explain nearly 50% of the variation in the rank of a player in the upcoming year. Not great.
It doesn’t account for injury. Players that switch teams. The players around them. The coaches around them. That’s a lot that is not accounted for that most definitely affects a player’s fantasy production.
Next year
Next year I plan on updating each of these models to see if I can get a more accurate model. Hopefully, I can find some more data that includes some of those caveats that these current models don’t account for.
I also would like to try and predict how well college players will perform in the NFL. See if college stats can predict production in the NFL. These kinds of models are exactly what NFL teams currently do. When you see them draft a player no one was expecting it is likely their model caught something about them that other models did not. Or they saw something about that player in person that no model could capture.
Stay tuned for the 2023 Fantasy Football Player Prediction.