Modeling NFL Production in Young WRs

I first ran across RotoViz when I Googled “Principal Component Fantasy Football” — it returned the first Google hit1 — because I’m a huge nerd. I quickly became enthralled with a fantasy football site that was all about metrics and analytics. Lo and behold, today I’m lucky enough to write for RotoViz,2 which in my humble (and almost surely biased) opinion is the best fantasy football site there is. That’s because we have groundbreaking articles such as Shawn Siegele’s 2014 piece Breakout Age is the Skeleton Key in addition to industry leading metrics like Jon Moore’s Phenom Index. Add to it the best apps on the web such as the RotoViz Screener, and we get a combination that lays the foundation for this particular article. I’ve pulled together Moore’s Phenom Index numbers from 2004 to present, including the individual age and market share components from the linked article above, and meshed them with each receiver’s NFL statistics using the RotoViz Screener App to create a statistical model that predicts PPR fantasy points per game for receivers in their first four years in the NFL to a high degree of accuracy. I’ll show you which variables are significant in predicting success in each of the first four years of a receiver’s career, and how those variables change in significance throughout a WR’s early career. Then I’ll use these models to find the best candidates for success in 2016.

The Model Factors

I created a separate model for each of the first four seasons of an NFL receiver’s career. The factors I looked at included:
  1. Year (SEAS)3
  2. Logarithm of draft position (L.DPOS)
  3. Draft age (AGE.Z)4
  4. Final year collegiate market share (MS.Z)5
  5. PPR points per game the prior year (PPR.N)
  6. Games played the prior year (GMS.N)
For the rookie season, I obviously did not have bullets 5 and 6, but for the rest of the seasons I considered models using all six variables. One final note, these models only use data points that were nonzero in the response. In other words, I only looked at seasons in which a player put up fantasy points. If a player didn’t play, he is not included in the results. Thus, we don’t get to see bust rates, but simply the model results for players who did play. Here are the results.

Rookie Year Model
  1. Try it!  (back)
  2. Thanks Charles Kleinheksel (insert eyeroll GIF here)…  (back)
  3. The NFL has trended toward a more passing oriented league over the last decade.  (back)
  4. I used the z-score as defined in Jon Moore’s Phenom Index article.  (back)
  5. Again the z-score as used in Phenom Index.  (back)

Subscribe to the best value in fantasy sports

You're all out of free reads for now and subscribing is the only way to make sure you don't ever miss an article.

By RotoDoc | @RotoDoc | Archive

Comments   Add comment

  1. Jason says:

    You're killing it today Doc. My one question is a guy like Terrance Williams. He has some things working against him (Age, potential low volume pass offense, etc). But I just can't give up on him as a post-hype sleeper.

    In year 4 and even year 3, his old age should've have hurt him anymore according to your model. However, he seemingly didn't produce as well in year 3 due to Romo being hurt and basically everything going wrong for the Cowboys. If you look at weeks 1 and 11, the two weeks Romo played the full game, Williams scored 11 and 17 PPR points in both of those weeks. This is an incredibly small sample, but I really think Williams has a solid floor of a lower-end WR3 this year with TD upside playing alongside one of the most efficient QBs in the NFL. His OC is a plus too.

    What is the call on Williams then? The Cowboys defense is shaping up to be awful this season. Is there any way to factor in a lost type of season like Williams had in his year 3 into a model like this?

  2. Looking forward to the answer to this one. Williams is really interesting because he benefits in the 3rd year model from his massive final year market share being considered independent of age, since he was a 22-year-old senior. But then gets dinged in the 4th year from the quarterback nightmare the Cowboys had. I suppose you could run the 4th year model again, using his projection from the 3rd year model as his PPR.N. It's probably a wash though, because he probably lost a TD or two from Romo's absence, but gained targets and yards from Dez's injury.

  3. Jason says:

    I think the past bet on this one is just to use human intuition. I'll need to check out more of Williams in the preseason, but I am willing to take a flier on him in rounds 13-15 because I think he has a solid floor of a guy who can be a safe bye week fill-in (as opposed to a guy like Coates or Perriman who are extremely boom or bust). Laurent Robinson had a 10+ TD season playing with Romo and Dez so it's also possible the Cowboys are very efficient and Williams has a season like that where he gets some random type TD upside over the course of the full season.

    Williams is also in a contract year if you pay attention to those sorts of things. Going down narrative street here, but he will likely want to do everything in his power to get one (reasonably) big contract especially given his age. The Cowboys may also pepper him with more targets to see if they want to keep him or give up on him as their long-term WR2.

  4. I'm not sure where Amari ranks in the pantheon on the young WRs in the NFL. I know he profiles as a more athletic Julio Jones and broke out at a very early age but I think another year is needed to determine just how special he may or may not be. The headwinds against him this year is Oakland might be decent (not playing from behind). The head coach is a former DC and the offense is run-based, and the run game with Latavius and now DeAndre Washington should continue being good. Then there are the 5 games against Denver, KC and Carolina. I think I'd still take ARob against him straight up in dynasty.

Discuss this article on the RotoViz Forums