Forecasting WR Performance

Now that the NFL draft has come and gone, we’re left to sort out the implications for our dynasty rookie picks and redraft fantasy teams. We all know that draft position to predictive of future NFL success, but how much should we weight it versus our pre-draft evaluations? By finding the right set of relevant variables, we can develop a numbers-based formula that would have told us that Allen Robinson (drafted 61st overall) was a much better prospect than Davante Adams (53rd), and Tyler Lockett (69th) looked more like past NFL successes than Nelson Agholor (20th).

In this article our goal is to predict NFL success for recently drafted wide receiver prospects, using collegiate production statistics, NFL combine measurables and NFL draft data. We’re going to define NFL success in a way that would likely lead to a profit on your dynasty rookie draft picks, and that’s reaching a threshold season-long finish in a player’s first three years. For wide receivers we chose a top-24 finish (PPR scoring) as the most logical threshold to define NFL success. Here is a link to the rookie running back analysis.

To determine the variables that, in addition to draft position, are statistically significant for predicting early NFL success, we used a logistic regression model. The logistic regression model builds a formula for predicting a binary outcome - in this case NFL success - by weighting the relevant independent variables.

The Results

The three variables we found best for predicting early NFL wide receiver success are (in order of statistical significance):

• Draft position (logarithmic formula to capture exponential decay of value)

• Career market share of receiving yards

• Final-year market share of receiving yards

Here are the top-15 predict scores (roughly equivalent to the likelihood of success) for the 2000-2013 data set used to train and test the model.

Top 15 Wide Receiver Scores (2000-2013)

Name	College	Year	Draft Pos	Car MS Rec Yds	MS Rec Yds	Top-24	Predict
Calvin Johnson	Georgia Tech	2007	2	0.42	0.51	1	0.99
Charles Rogers	Michigan State	2003	2	0.43	0.49	0	0.99
Larry Fitzgerald	Pittsburgh	2004	3	0.41	0.45	1	0.97
Demaryius Thomas	Georgia Tech	2010	22	0.43	0.61	1	0.96
Braylon Edwards	Michigan	2005	3	0.34	0.47	1	0.96
A.J. Green	Georgia	2011	4	0.34	0.39	1	0.90
Plaxico Burress	Michigan State	2000	8	0.38	0.43	1	0.90
Lee Evans	Wisconsin	2004	13	0.43	0.43	1	0.90
Dez Bryant	Oklahoma State	2010	24	0.35	0.60	1	0.89
Quincy Morgan	Kansas State	2001	33	0.48	0.44	1	0.86
Rashaun Woods	Oklahoma State	2004	31	0.39	0.53	0	0.85
Roy Williams	Texas	2004	7	0.34	0.40	1	0.85
Josh Reed	LSU	2002	36	0.40	0.53	0	0.85
Kelley Washington	Tennessee	2003	65	0.42	0.54	0	0.81
Koren Robinson	North Carolina State	2001	9	0.36	0.36	1	0.79

Nine of the top-10 receivers hit the top-24 threshold in their first three seasons, including perennial fantasy studs Calvin Johnson, Larry Fitzgerald and A.J. Green.

There are a few busts in the top-15, but the model also appropriately gives additional credit for the high market shares of late first round picks Demaryius Thomas and Dez Bryant.

Thomas had elite market share numbers as part of the low-volume Georgia Tech offense, but his raw numbers looked somewhat pedestrian, averaging only 60 receiving yards per game during his college career. It’s possible that Thomas fell during the draft process as evaluators focused on his yardage total, not more predictive statistics like market shares.

The 2014 and 2015 draft classes were not part of the model development due to the lack of seasoning, but it’s helpful to look at those scores anyway to see if the model has any predictive value with an out-of-sample group.

2014-15 WR Scores

Name	College	Year	Draft Pos	Car MS Rec Yds	MS Rec Yds	Predict
Amari Cooper	Alabama	2015	4	0.36	0.44	0.94
Sammy Watkins	Clemson	2014	4	0.30	0.34	0.83
Devante Parker	Louisville	2015	14	0.24	0.57	0.74
Jordan Matthews	Vanderbilt	2014	42	0.37	0.50	0.73
Mike Evans	Texas A&M	2014	7	0.29	0.30	0.67
Odell Beckham Jr	LSU	2014	12	0.29	0.35	0.61
DeAndre Smelter	Georgia Tech	2015	132	0.36	0.53	0.49
Brandin Cooks	Oregon State	2014	20	0.27	0.36	0.46
Marqise Lee	USC	2014	39	0.38	0.32	0.42
Tyler Lockett	Kansas State	2015	69	0.36	0.41	0.42
Allen Robinson	Penn State	2014	61	0.30	0.46	0.40
Nelson Agholor	USC	2015	20	0.24	0.34	0.38
Paul Richardson	Colorado	2014	45	0.26	0.45	0.37
Breshad Perriman	Central Florida	2015	26	0.23	0.34	0.30
Davante Adams	Fresno State	2014	53	0.32	0.34	0.29

It should come as no surprise that the top draft picks in the last two classes, Amari Cooper and Sammy Watkins, top the list. But somewhat lower selections, like Jordan Matthews, Allen Robinson and Tyler Lockett, were also among the top scores, lifted by outstanding market shares.

The model did a decent job sniffing out some players to avoid in the past couple dynasty rookie drafts. First-year disappointment Nelson Agholor is lower than you’d expect based on draft position alone, and Marqise Lee only has a slightly higher predict score than teammate Allen Robinson, despite being drafted much earlier.

One name that jumps off the page as a potential dynasty trade target is DeAndre Smelter, who scores much higher than expected for a late fourth round pick. Smelter didn’t make it on the field his rookie season after suffering an ACL injury late in his college career, so we still don’t know how he’ll perform at the next level. There’s little in the way of target competition in San Francisco, and Chip Kelly has brought his fantasy-friendly, high-volume offense to town.

Now let’s move on to the 2016 draft class.

2016 Draft WR Scores

Name	College	Team	Draft Pos	Car MS Rec Yds	FY MS Rec Yds	Predict	Dynasty ADP
Tyler Boyd	Pittsburgh	Bengals	55	0.43	0.40	0.61	WR7
Corey Coleman	Baylor	Browns	15	0.26	0.39	0.56	WR2
Leonte Carroo	Rutgers	Dolphins	86	0.36	0.49	0.52	WR8
Will Fuller	Notre Dame	Texans	21	0.29	0.37	0.50	WR6
Josh Doctson	Texas Christian	Redskins	22	0.28	0.38	0.49	WR1
Laquon Treadwell	Mississippi	Vikings	23	0.23	0.26	0.25	WR2
Tajae Sharpe	Massachusetts	Titans	140	0.32	0.43	0.22	WR14
Sterling Shepard	Oklahoma	Giants	40	0.24	0.32	0.21	WR4
Rashard Higgins	Colorado State	Browns	172	0.35	0.39	0.18	WR13
Michael Thomas	Ohio State	Saints	47	0.20	0.32	0.15	WR5
Pharoh Cooper	South Carolina	Rams	117	0.26	0.39	0.13	WR10
Malcolm Mitchell	Georgia	Patriots	112	0.23	0.36	0.10	WR11
Devin Lucien	Arizona State	Patriots	225	0.30	0.30	0.06	NA
Mike Thomas	Southern Mississippi	Rams	206	0.26	0.32	0.06	WR12
Daniel Braverman	Western Michigan	Bears	230	0.24	0.37	0.06	WR20
Ricardo Louis	Auburn	Browns	114	0.15	0.31	0.05	WR19
Demarcus Robinson	Florida	Chiefs	126	0.27	0.21	0.05	WR16
Aaron Burbridge	Michigan State	49ers	213	0.20	0.38	0.05	WR21
Chris Moore	Cincinnati	Ravens	107	0.16	0.22	0.04	WR23
Jordan Payton	UCLA	Browns	154	0.19	0.29	0.04	WR22
Braxton Miller	Ohio State	Texans	85	0.04	0.14	0.03	WR9
Jakeem Grant	Texas Tech	Dolphins	186	0.18	0.25	0.03	NA
DeMarcus Ayers	Houston	Steelers	229	0.17	0.35	0.03	NA
Trevor Davis	California	Packers	163	0.13	0.14	0.02	NA
Kolby Listenbee	Texas Christian	Bills	192	0.15	0.17	0.02	NA
Cody Core	Mississippi	Bengals	199	0.10	0.16	0.01	NA
Devin Fuller	UCLA	Falcons	238	0.10	0.09	0.01	NA
Charone Peake	Clemson	Jets	241	0.08	0.17	0.01	WR18
Kenny Lawler	California	Seahawks	243	0.14	0.15	0.01	NA

Dynasty ADP from RotoViz Dynasty ADP App

Wide receiver scores are relatively low this year due to their weaker than usual collective draft position. The notion of trading your 2016 dynasty picks for veteran talent or future picks has been widely discussed in fantasy circles, and the numbers bear that out.

On the positive side, the combination of Tyler Boyd’s top-notch market share vaults the uber-productive wideout to the top of the class. Some think of Boyd as a low-ceiling prospect due to his relatively disappointing NFL combine performance, but he certainly has a high floor.

Corey Coleman is near the top of the class, right where you’d expect him to be as the first receiver taken. Coleman is followed closely by Leonte Carroo, a market-share stud who the model predicts will outperform his late third round draft position.

Unfortunately, Carroo is buried on the depth chart in Miami, with Jarvis Landry and 2015 first-round pick DeVante Parker ahead of him in the pecking order. When assessing Carroo’s chance to surpass either on the depth chart, keep in mind that his predict score could be somewhat overstated by his smaller sample of games played (Carroo didn’t play his freshman year and missed another combined eight games during his sophomore and senior years).

The drop-off in predict scores between No. 22 and 23 picks Josh Doctson and Laquon Treadwell is significant. Some in the metrics community have voiced concerns about Doctson’s advanced age, roughly two-and-a-half years older than Treadwell. However, the model did not find age as a statistically significant variable, so it could already be incorporated into draft position. It’s possible that Treadwell’s market share numbers were held down not only by his younger age but also injury recovery. Nevertheless, the former consensus No. 1 receiver doesn’t look like a great risk-reward pick in rookie drafts at his current WR2 ADP.

The small-school phenom Tajae Sharpe finds himself among the big names of the 2016 class, reflecting his enormous market share numbers. Sharpe is a taller receiver at 6-feet 2-inches, but only weighs 194 pounds. Sharpe just turned 21 years old last December, so it’s possible he’s still filling out his frame. The fifth round pick won’t necessarily have immediate opportunity with the Titans, but none of his target competition (Rishard Matthews, Kendall Wright and Dorial Green-Beckham) has established themselves as a lead receiver.

Sterling Shepard and Michael Thomas are both slight fades in the model relative to their dynasty rookie draft ADPs. Each has a top-10 predict scores, but the scores aren’t as high as expected considering their relatively strong draft positions. The model doesn’t incorporate the opportunity available to rookies, so you may want to make your own mental adjustments to Shepard and Thomas if you like the respective landing spots with the Giants and Saints.

The biggest loser in the model is clearly Braxton Miller. The converted quarterback doesn’t have a production profile that comes anywhere near what you’d expect based on his third-round draft position. It’s possible that Miller could learn his craft and become productive in the NFL, but previous weak producers who had higher draft positions, like Cordarrelle Patterson and another converted quarterback, Matt Jones, are cautionary tales.

If you want to throw a few darts at the receivers in the later rounds of your dynasty rookie drafts, you could do worse than Rashard Higgins and Pharoh Cooper. Higgins, a fifth round pick, has a predict score sandwiched between second-rounders Shepard and Thomas, and could have immediate opportunity in Cleveland. Cooper was the first receiver selected by the Rams, and plays a position of need for the team, all while having the opportunity to play alongside No. 1 pick Jared Goff.