CanonDynasty

Foreman, Fournette and Mixon Lead the 2017 RB Success Model

We have roughly a month until the 2017 NFL draft, when we will learn where our favorite (or not so favorite) prospects will land this coming season. While draft position and landing spot are huge factors for forecasting the success of any running back prospect, I’ve found that we can accurately predict whether a running back will be successful largely based on his production profile and athletic measurables.

We know that collegiate production isn’t everything for wide receivers, it’s the only thing. For running backs, the situation is wholly different. Production matters, but size-adjusted speed is king for determining which running backs will be successful in the NFL.

You can define success many ways, but I’m choosing to use a top-12 fantasy point season (PPR) for running backs. The model’s dependent variable for early NFL success is whether or not a player had such a season within his first three years in the NFL.

We used age, production, and combine measurables to train and test the updated 2017 running back model. The model used 350 running back prospects that entered in the NFL from 2000-2014, splitting the data roughly 2-to-1 into training and testing sets.

After plugging dozens of different production and combine statistics into the model and slowly taking away, one-by-one the least statistically significant, we were left with four (two combine, two production) that provide the most explanatory and predictive power (listed in order of statistical significance):

1. 40-yard dash

2. Weight

3. Final season rushing yards per game

4. Final season receptions per game

As you’d expect, the model favors faster, heavier prospects who had strong rushing and receiving production in their final college season. The 40-yard dash is by far the most influential statistic for predicting NFL success, followed by weight. The model’s c-statistic on the test set is nearly 0.80, which is generally considered a strong score for a logistic regression model.

We are big believers in the predictive value of the three-cone drill here at RotoViz, and I found in my regression tree analysis that when looking at strictly combine measurables, the three-cone drill is significant for slower backs. But many RB prospects choose to skip the agility drills at the combine, leaving us with a hard choice to either exclude them from the analysis or estimate the missing times. I estimated the missing times using a linear regression on a prospect’s weight and 40-yard dash, which are strong predictors. Even with these estimates, my analysis found that adding three-cone times to weight and 40-yard dash didn’t enhance prediction. I don’t think you should ignore agility in your running back prospect analysis, but I wouldn’t give a prospect additional credit for a fast three-come time if he already has strong weight-adjusted speed.

To get an historical perspective on the types of prospects the model favors, here are the top-15 scores for the entire 2000-2014 data set. Remember, draft position is not one of the inputs in the model, I only added it here for reference. You can think of the “Top 12 Predict” score as the likelihood that the running back will meet the model threshold of registering at least one top-12 PPR season in his first three years as a pro.

Player School Draft Year Draft Position Weight Forty RuYds/Gm Rec/Gm Top-12 Top-12 Predict
Chris Johnson East Carolina 2008 24 191 4.24 109.5 2.8 Yes 0.58
Darren McFadden Arkansas 2008 4 210 4.33 140.8 1.6 Yes 0.58
Matt Forte Tulane 2008 44 218 4.46 177.2 2.7 Yes 0.58
Kevin Jones Virginia Tech 2004 30 228 4.38 126.7 1.1 Yes 0.57
Michael Turner Northern Illinois 2004 154 244 4.49 137.3 1.6 No 0.57
JJ Arrington California 2005 44 214 4.40 168.2 1.8 No 0.56
Demarco Murray Oklahoma 2011 71 213 4.41 86.7 5.1 Yes 0.55
Latavius Murray Central Florida 2013 181 223 4.38 100.5 2.5 Yes 0.55
Ladainian Tomlinson Texas Christian 2001 5 221 4.46 196.2 0.9 Yes 0.50
Rashard Mendenhall Illinois 2008 23 225 4.45 129.3 2.6 No 0.50
Reggie Bush USC 2006 2 203 4.36 133.8 2.8 Yes 0.50
Adrian Peterson Oklahoma 2007 7 217 4.40 144.6 1.4 Yes 0.47
Jonathan Stewart Oregon 2008 13 235 4.48 132.5 1.7 No 0.47
Larry Johnson Penn State 2003 27 228 4.55 160.5 3.2 Yes 0.45
Ronnie Brown Auburn 2005 2 230 4.43 76.1 2.8 No 0.43

You’ll see that this draft-agnostic model was good at predicting success, even though only eight of the 15 above went in the first round of the NFL draft. The model does have some misses, but even technical misses like Michael TurnerRashard Mendenhall and Jonathan Stewart were more near-hits or late-bloomers than abject failures.

Now the part we’ve all been waiting for: Let’s apply our historically accurate model to the 2017 draft class. Here are the top-10 scores.

** Updated following D’Onta Foreman’s Pro Day**

Player School Weight Forty RuYds/Gm Rec/Gm Top-12 Predict
D’Onta Foreman Texas 233 4.48* 184.4 0.6 0.53
Leonard Fournette Louisiana State 240 4.51 120.4 2.1 0.44
Joe Mixon Oklahoma 228 4.48* 106.2 3.1 0.42
Joe Williams Utah 210 4.41 156.3 1.0 0.35
Jeremy McNichols Boise State 214 4.49 131.5 2.8 0.29
Christian McCaffrey Stanford 202 4.48 145.7 3.4 0.28
Dalvin Cook Florida State 210 4.49 135.8 2.5 0.23
Marlon Mack South Florida 213 4.50 98.9 2.3 0.14
Aaron Jones Texas-El Paso 208 4.56 147.8 2.3 0.12
TJ Logan North Carolina 196 4.37 50 2.2 0.11

* 4.45 pro day time with 0.03 added based on historical Combine-Pro Day differentials.

It’s safe to say that D’Onta Foreman hasn’t been a draftnik favorite, coming in as RB5 in the most recent iteration of the RotoViz Scouting Index. While other top backs basked in the spotlight of the combine in Indy, Foreman sat out the drills with a stress fracture in his foot. Well, he didn’t hold anything back at the Texas pro day with a little time to rest. The statistical significance of pro day times vis-a-vis the combine is questionable at best, but I think it’s safe to say that the most productive top prospect (on the ground) of the 2017 class ain’t slow. The only blight on Foreman’s resume is his lack of receiving production. If his low reception numbers were more about system than skill, he could be a one of the greatest draft-day bargains we’ve seen.

It shouldn’t be a shock to see Leonard Fournette near the top of the rankings. Even as some draft observers have soured on his rushing ability following a relatively lackluster final college season, any disappointment surrounding Fournette as a prospect may simply be a function of high expectations. Fournette has bold check marks in the all-important weight and speed categories, and his production is more than sufficient for a top-tier prospect. No one is expecting Fournette to be a force in the passing game, but his decent receiving numbers give an unexpected boost to his candidacy as the top back of the 2017 class.

Joe Mixon is a polarizing prospect for good reason. He wasn’t allowed to participate at the combine, which adds a layer of uncertainty to his true athleticism. It’s my personal opinion that Mixon shouldn’t go until at least the late rounds of the draft – if at all – as his well-documented transgressions outweigh any potential benefit he can provide at the devalued RB position. But the Mixon redemption tour has been in full swing the last couple months, and the lack of blowback for the the Chiefs after drafting Tyreek Hill last season makes Mixon likely to go much earlier.

Matt Freedman has done yeoman’s work in the honor of the late-round running back, giving us his workhorse metric for finding overlooked prospects. So should we be concerned that Mixon wasn’t close to a workhorse and didn’t even lead his own team in rushing attempts? I think the answer is no, as Mixon’s talent is viewed on a much higher plane than the backs Freedman profiled, and his raw production numbers are more than sufficient for forecasting success at the next level.

Joe Williams and Jeremy McNichols are your 2017 running back sleepers. It’s fair to question whether Williams will be able to replicate his propensity for break-away runs in the NFL, but we at least know that he won’t be lacking the speed to do so. Williams doesn’t have ideal workhorse size or much in the way of receiving production, but the fact that one of the most productive backs in college football didn’t even make Dynasty League Football’s March Rookie ADP list means he’ll be well worth the nominal required investment.

McNichols doesn’t score as high as Williams in the model, but I still think he’s the most undervalued RB of the 2017 class. McNichols doesn’t have a blazing 40-yard dash, but it’s a better size-adjusted time than some RB commonly mocked in the first round. More importantly, McNichols is a back who can get it done in both the running and passing game, giving a chance to land a potential workhorse at his current RB10 rookie ADP.

For two backs commonly seen as first-round talents, Christian McCaffrey and Dalvin Cook don’t have scores as strong as you’d expect. I wouldn’t say that the model hates either, but their scores are significantly lower than you’d hope for at their likely ADPs. While not part of the model, McCaffrey’s score should be viewed in a positive light considering to his 6.57 three-cone time, which is 0.38 seconds faster than you’d expect based on his weight and 40-yard dash.

On the flip side, Cook bombed the agility drills, registering a 7.27 three-cone time (0.27 seconds slower than expected).1 This doesn’t mean that Cook won’t succeed at the next level, but there are red flags waving in unison for those looking to invest significant draft capital in the stud RB.

Others Running Backs of Note:

Player School Weight Forty RuYds/Gm Rec/Gm Top-12 Predict
Kareem Hunt Toledo 216 4.62 113.5 3.2 0.08
Alvin Kamara Tennessee 214 4.56 54.2 3.6 0.07
Samaje Perine Oklahoma 233 4.65 106.0 1.0 0.05

Kareem HuntAlvin Kamara and Samaje Perine are all seen as top-10 prospects according to NFL Draft Scout. Evaluators are enamored with Kamara’s tape and were impressed with his excellent burst at the combine.2 But the success model doesn’t share their optimism. Kamara, along with Hunt and Perine, lack the weight-adjusted speed the model loves. It isn’t that these prospects are doomed,3 but you also can’t ignore their relative lack of potential based on historical data.

The entire 2017 class (FBS combine participants):

PlayerSchoolWeightFortyRuYds/GmRec/GmTop-12 Predict
D'Onta ForemanTexas2334.48184.40.60.53
Leonard FournetteLouisiana State2404.51120.42.10.44
Joe MixonOklahoma2284.48106.23.10.42
Joe WilliamsUtah2104.41156.310.35
Christian McCaffreyStanford2024.48145.73.40.28
Jeremy McNicholsBoise State2144.49131.52.80.28
Dalvin CookFlorida State2104.49135.82.50.23
Marlon MackSouth Florida2134.598.92.30.14
Aaron JonesTexas-El Paso2084.56147.82.30.12
TJ LoganNorth Carolina1964.37502.20.11
Brian HillWyoming2194.54132.90.60.1
Elijah McguireLouisiana-Lafayette2144.5386.72.20.09
Kareem HuntToledo2164.62113.53.20.08
Alvin KamaraTennessee2144.5654.23.60.07
Jamaal WilliamsBrigham Young2124.59137.50.70.05
James ConnerPittsburgh2334.65841.60.05
Samaje PerineOklahoma2334.6510610.05
Chris CarsonOklahoma State2184.5862.11.40.03
Wayne GallmanClemson2154.675.51.30.03
Corey ClementWisconsin2204.68105.80.90.02
Stanley-Boom WilliamsKentucky1904.51900.60.02
Dare OgunbowaleWisconsin2134.6536.11.70.01
Jahad ThomasTemple1904.6279.42.80.01
Devine ReddingIndiana2054.7686.32.10
Deveon SmithMichiganNANA65.11.2NA
Donnell PumphreySan Diego StateNANA152.41.9NA
Elijah HoodNorth Carolina232NA782.3NA
Justin DavisSouthern California208NA60.71.4NA

Subscribe for a constant stream of league-beating articles available only with a Premium Pass.

  1. I still have yet to figure out why prospects with poor agility don’t just sit out the drill.  (back)
  2. Vertical and broad jump of 39.5 and 131 inches, respectively.  (back)
  3. Arian Foster and Frank Gore had long, successful careers with predict scores of 0.01 and 0.04, respectively.  (back)
By Kevin Cole | @Cole_Kev | Archive

Comments   Add comment

  1. There is a formula, but it's not easily calculated like a linear regression. Not sure it would be much value to share.

  2. @colekev_FF Ssoooo he ran a 4.45 at his pro day, thats pretty amazing at that size

  3. I think you're right that the tree nodes are smaller and more difficult to rely on for statistical significance. The value of the trees was more to put the combine drills into a digestible format based on past results. The trees can be overfit, or closely follow past data at the expense of being predictive.

  4. Thanks for doing this. The tree model always bothered me since it created binary branches out of continuous data, and resulted in findings where a 0.01 difference in forty or agility time would create massive swings in "success" predictions.

Discuss this article on the RotoViz Forums

12 more replies