The Model

The goal of this model is to use college football statistics to predict the fantasy football success of a college quarterback prospect to the NFL. To do this I have created two models, one which predicts fantasy points per game (PPR) and one that predicts NFL draft position using a neural network.

Previously, I had used linear regression models. Not every metric has a linear relationship to the input and output variables, which is where linear regression over or understates outcomes with multiple input variables. Neural Networks are the computerized representation of the human brain that finds complex relationships between input and output nodes to create weights that help predict the desired outcome.

The Results 

The data was trained on 87 QBs in the classes of 2016-2023, and the prediction was on 87 QBs in the classes of 2024-2026. The root mean squared error for the fantasy points per game model is 0.039, and the draft position model is 0.028, which identifies the accuracy of the model. The neural network is attempting to minimize the sum of squared error as an objective function within the model, which minimizes variation in the prediction from the actual result. The input variables are: 

  1. Power 5
  2. Cumulative Pass EPA: Increments of 100 pass attempts between 100-1000
  3. Cumulative Rush Yards Over Expectation: Increments of 10 rush attempts between 10-200
  4. QBR: Years 1-5 out of High School
  5. Games Played: Years 1-5 out of High School 
  6. Cumulative Sack EPA Lost: Increments of 100 pass attempts between 100-1000
  7. Cumulative Interception EPA Lost: Increments of 100 pass attempts between 100-1000

The Context

The most important factors for the fantasy points per game model are year 1 QBR, Rush Yards Over Expectation at 80 and 160 attempts, and Total Pass EPA at 600 and 1000 attempts. The conclusion here is that Year 1 play, running the ball at high volume and at high efficiency, and/or passing for volume and at high efficiency are important success factors for a QB to score fantasy points in the NFL.

The most important factors for the draft position model are interception EPA lost at 200 attempts, Pass EPA at 700 attempts, and Sack EPA Lost at 200 and 400 attempts. The conclusion here is that a QB who protects the ball via interceptions and sacks as well as passes for volume and at high efficiency will get drafted higher.

The Prediction

Below I have identified the Top 60 QBs in the model (Sorted by Fantasy & Draft QB Index) for the class 2017-2026 and the Top 40 QBs in the model for the class 2024-2026.

Personal NFL Cornerstone Rankings Past 4 Years: 

  1. C.J. Stroud
  2. Caleb Williams
  3. Anthony Richardson
  4. Drake Maye
  5. Justin Fields
  6. Brock Purdy
  7. Trevor Lawrence
  8. J.J. McCarthy
  9. Jayden Daniels
  10. Bryce Young
  11. Will Levis
  12. Spencer Rattler
  13. Bo Nix
  14. Michael Penix Jr.
  15. Mac Jones
  16. Zach Wilson
  17. Kenny Pickett

A few takeaways from the NFL: 

  • Caleb Williams and Drake Maye are the top two prospects in 2024, which is what the model projected last year. Their scores both decreased from their phenomenal seasons in 2022 but still came in as the 11th and 12th player within the model. Caleb is my QB1, but Maye is a strong second option.
    • Williams checks most boxes. Elite passer who bests Joe Burrow and Trevor Lawrence in total pass EPA, Elite Rusher who is just behind Kyler Murray and ahead of Anthony Richardson and Justin Fields, but he is in the same stratosphere as Justin Fields and Bryce Young in sack EPA Lost. 
      • QBR Year 1: 100th Percentile
      • RYOE by 80 attempts: 66th Percentile
      • RYOE by 160 attempts: 82nd Percentile 
      • Pass EPA by 600 attempts: 74th Percentile 
      • Pass EPA by 1000 attempts: 88th Percentile 
      • Interception EPA Lost by 200 attempts: 95th Percentile 
      • Sack EPA Lost by 200 attempts: 35th Percentile
    • Maye checks most boxes as well. Mayes’ career started off elite, but many of his stats declined in his third season. He is a strong passer who bests Deshaun Watson in total pass EPA, mediocre Rusher, initially strong on interception EPA lost, and mediocre on Sack EPA lost.  
      • QBR Year 1: NA
      • RYOE by 80 attempts: 50th Percentile
      • RYOE by 160 attempts: 44th Percentile 
      • Pass EPA by 600 attempts: 70th Percentile 
      • Pass EPA by 1000 attempts: 70th Percentile 
      • Interception EPA Lost by 200 attempts: 100th Percentile 
      • Sack EPA Lost by 200 attempts: 52nd Percentile
  • Spencer Rattler, Bo Nix, and JJ McCarthy are in the next tier of players and come in as QB16, QB17, and QB19 in the model. Although third of this group in the Model, McCarthy is my top QB of this group. Rattler and Nix are interesting players to take a shot on in the second round of Dynasty Drafts if they are taken in the first or early second of the NFL draft. The model believes they have a strong chance to get drafted high in the NFL Draft but will most likely operate as back end QB2’s for fantasy football. 
    •  Spencer Rattler: 
      • QBR Year 1: NA
      • RYOE by 80 attempts: 22nd Percentile
      • RYOE by 160 attempts: 34th Percentile 
      • Pass EPA by 600 attempts: 60th Percentile 
      • Pass EPA by 1000 attempts: 49th Percentile 
      • Interception EPA Lost by 200 attempts: 72nd Percentile 
      • Sack EPA Lost by 200 attempts: 60th Percentile
    • Bo Nix: 
      • QBR Year 1: NA
      • RYOE by 80 attempts: 24th Percentile
      • RYOE by 160 attempts: 29th Percentile 
      • Pass EPA by 600 attempts: 34th Percentile 
      • Pass EPA by 1000 attempts: 35th Percentile 
      • Interception EPA Lost by 200 attempts: 50th Percentile 
      • Sack EPA Lost by 200 attempts: 70th Percentile
    • JJ McCarthy: 
      • QBR Year 1: NA
      • RYOE by 80 attempts: 49th Percentile
      • RYOE by 160 attempts: 14th Percentile 
      • Pass EPA by 600 attempts: 78th Percentile 
      • Pass EPA by 1000 attempts: NA
      • Interception EPA Lost by 200 attempts: 82nd Percentile 
      • Sack EPA Lost by 200 attempts: 77th Percentile
  • Jayden Daniels had a really bad start to his career, with one phenomenal season in Year 5. If you focus on just 2023 Daniels looks like a great prospect but the full body of work identifies some serious concerns. He has been a below average passer for the majority of his career and has crept up to be a slightly above average rusher. Daniels will have the same sack concerns as Justin Fields, with worse production as a passer and potentially rusher based on what Fields has shown at the NFL level.
    • QBR Year 1: 75th Percentile
    • RYOE by 80 attempts: 28th Percentile
    • RYOE by 160 attempts: 60th Percentile 
    • Pass EPA by 600 attempts: 48th Percentile 
    • Pass EPA by 1000 attempts: 41st Percentile
    • Interception EPA Lost by 200 attempts: 88th Percentile 
    • Sack EPA Lost by 200 attempts: 42nd Percentile

A few takeaways CFB: 

  • Players with minimal college starts that Campus to Canton has ranked highly who you should take over the players from the model: 
    • Nico Iamaleava
    • Jackson Arnold
    • Arch Manning
    • Dylan Raiola
  • Will Howard, you have my attention. The path that the model is following for Will Howard is Josh Allen. Howard has grown to be an above average rusher, and average passer, just below average in interception EPA lost as the attempts grow, and he has been solid in sack epa lost. I wouldn’t rank Howard over Williams, Maye, or as a top 12 QB into the NFL, but he is on the path to ascend into that and Ohio State seems to agree.
    • QBR Year 1: 58th Percentile
    • RYOE by 80 attempts: 46th Percentile
    • RYOE by 160 attempts: 53rd Percentile 
    • Pass EPA by 600 attempts: 44th Percentile 
    • Pass EPA by 1000 attempts: NA
    • Interception EPA Lost by 200 attempts: 8th Percentile 
    • Sack EPA Lost by 200 attempts: 69th Percentile
  • Jaxson Dart is a bad rusher who has demonstrated being an above average passer that can take care of the ball. If he can let loose as a passer this year he could follow a path similar to Baker Mayfield. 
    • QBR Year 1: NA
    • RYOE by 80 attempts: 30th Percentile
    • RYOE by 160 attempts: 25th Percentile 
    • Pass EPA by 600 attempts: 53rd Percentile 
    • Pass EPA by 1000 attempts: NA
    • Interception EPA Lost by 200 attempts: 79th Percentile 
    • Sack EPA Lost by 200 attempts: 92nd Percentile
  • Quinn Ewers is a younger and slightly lower-ranked prospect similar to Dart. I would attempt to trade away Quinn Ewers if possible in C2C or Devy.
    • QBR Year 1: NA
    • RYOE by 80 attempts: 22nd Percentile
    • RYOE by 160 attempts: NA
    • Pass EPA by 600 attempts: 53rd Percentile 
    • Pass EPA by 1000 attempts: NA
    • Interception EPA Lost by 200 attempts: 62nd Percentile 
    • Sack EPA Lost by 200 attempts: 91st Percentile
  • Carson Beck is older than both Jaxson Dart and Quinn Ewers but has a worse profile than both. He seems to have some hype at the moment but this is a player I would push hard to trade away in C2C or Devy. 
    • QBR Year 1: NA
    • RYOE by 80 attempts: 22nd Percentile
    • RYOE by 160 attempts: NA
    • Pass EPA by 600 attempts: 30th Percentile 
    • Pass EPA by 1000 attempts: NA
    • Interception EPA Lost by 200 attempts: 61st Percentile 
    • Sack EPA Lost by 200 attempts: 93rd Percentile

Get Our Newsletter

Get notified periodic notifications about our content and future subscription deals.

You May Also Like

Five to Trade Before the NFL Draft

Some players have value and could be affected by the Draft – now is the time to trade these five players!

The Next Kenneth Walker?

Wyoming has a running back who could be a VOLUME PIG in 2024; here’s the lowdown on Harrison Waylee and his opportunity!

Five to Pick Up Before the Draft

@Knighthawk7734 has five players to target that may be on your waivers and see an increase in value after the Draft!