Predicting WR Success
By: Devin Sperle
As we all know, finding talented players in rookie drafts is an essential component of building a championship roster. The problem is that identifying that talent is a difficult task. Can Player X be successful in the NFL? Is he someone that will put up fantasy numbers for my team? These are all questions we ask ourselves, but rarely are we able to do anything more than guess.
I set out to identify the key factors in identifying wide receiver fantasy success. I collected data from all drafted WRs from 2008 – Present. I started out with about 20 factors, including things like college production, combine results, and physical statistics. For each receiver I noted whether or not he achieved a top-24 WR fantasy season during his first 3 seasons in the NFL. I ran a logistic regression for the data and began removing variables deemed “statistically insignificant” by the model.
When all was said and done, I was left with a predictive model that contained 3 variables: draft position, college market share of yards and touchdowns, and breakout age (the age at which the player first achieved a market share of 20% or higher). I was slightly surprised to see that there was no correlation between physical attributes or combine results and fantasy success.
No predictive model is perfect, mine included. However, using the data from 2008 – Present, this model is 76% accurate in predicting whether or not a wide receiver will achieve a top-24 fantasy season during his first 3 years in the NFL.
I’m not planning on revealing my predictive formula, at least for the time being, but I do want to give a quick snapshot of the results for the 2015 and 2016 draft classes. These numbers are the statistical probability (or percent chance) that each WR has of achieving a successful fantasy season.
REMEMBER: I am not saying that this model is perfect or that the results should be written in stone. There is a chance these WRs still bust, or that others not in the top 5 end up as the best WR in their class.
2016 Draft Class
1. Josh Doctson (0.635)
2. Leonte Carroo (0.618)
3. Tyler Boyd (0.603)
4. Corey Coleman (0.542)
5. Will Fuller (0.528)
2015 Draft Class
1. Amari Cooper (0.774)
2. DeVante Parker (0.624)
3. Devin Funchess (0.567)
4. Breshad Perriman (0.484)
5. Phillip Dorsett (0.479)
I plan on doing more work with this model and creating others for the QB, RB, and TE positions as well. As always, if you have questions, comments hit me up on Twitter @dynastyfbfocus or send an email to firstname.lastname@example.org