📊How I Perform Opponent Adjustments
▫️Treat each play as a unique matchup
▫️Use Ridge Regression to solve play's resulting EPA given the Offense, Defense, & league-wide homefield advantage
▫️FBS vs FBS games only
▫️Method & code below:
https://t.co/jRCqzaNbb4
Just dumped pretty much all my Recruit Expected Value analysis I've ever done in this thread
- Position
- Height, Weight, & Age
- Recruiting Service
- Location & School Type
- Historic Performance
📊Recruit Expected Value v3
▫️Updated using latest & greatest data & model fit
▫️Assuming high-end 3 Star (Rank #300) is "replacement level":
- #1 Recruit is worth ~20X
- Top 10 Recruits: 8-15X
- Other 5 Star: 5-8X
- 4 Stars: 1-5X
▫️Spend NIL accordingly
📊Title Win Rate by Class Expected Draft Value
▫️From 100-300 EDV: linear increase of 0-20% Title Win Rate. 10 pts of EDV worth 1% increase in Title Likelihood
▫️Above 350 Win Rate increases rapidly, with 10 pts of EDV worth 3% increase in Title Likelihood
📊Recruit Expected Value v3
▫️Updated using latest & greatest data & model fit
▫️Assuming high-end 3 Star (Rank #300) is "replacement level":
- #1 Recruit is worth ~20X
- Top 10 Recruits: 8-15X
- Other 5 Star: 5-8X
- 4 Stars: 1-5X
▫️Spend NIL accordingly
📊Historic Draft Value By Recruiting Class
▫️Alabama 2017 is BY FAR the GOAT
▫️Florida 2010 highest Expected DV but no hardware (thanks WM)
▫️Actual DV above on near Expected DV = higher title likelihood.
▫️No title-certainty for high level of Actual or Expected DV
▫️Poor USC
📊Recruit Expected Value v3
▫️Updated using latest & greatest data & model fit
▫️Assuming high-end 3 Star (Rank #300) is "replacement level":
- #1 Recruit is worth ~20X
- Top 10 Recruits: 8-15X
- Other 5 Star: 5-8X
- 4 Stars: 1-5X
▫️Spend NIL accordingly