Dynasty

Tight End Prospecting: 2017 Draft Class

Perhaps the biggest story of the NFL combine was the athleticism of the 2017 tight end draft class. The median dash time was 4.64 seconds for the field of 15 participants, and six of these 15 prospects leaped 35 inches or higher. With this kind of athleticism and the college production to match, at least one lesser-known TE is likely to be a great bargain in your dynasty formats.

The TE position is historically difficult for rookies. Over the last 10 years, the top rookie TE averaged a 43-462-3.5 line, which amounts to 67 standard fantasy points (111 PPR). With underwhelming players like John Carlson, Jace Amaro, and Tim Wright on this list, hoping to hit the top rookie TEs doesn’t seem a viable short-term strategy. Still, the 2017 TE class is deep enough to consider a shrewd, early investment for a long-term payout.

With TEs like Greg Olsen, Delanie Walker, and Jimmy Graham already in their 30s, the time is ripe for rookies to evolve into NFL stars. While the consensus is that O.J. Howard or David Njoku will be stars, you’ll see that Jim Cobern’s (@Jimetrics) TE database, combined with my logistic regression model, helps us identify three players with nearly equivalent NFL success profiles. The 30 percent success rate of the underlying  model suggests that at least one of these three players is expected to be a long-term NFL starter (start 64 or more NFL games in their career).

Applying the Model

The real power of this statistical model is how accurately it identifies TE busts. During the validation years, this model correctly identified 98 percent of players who have yet to log 64 NFL starts. This model depended significantly upon relative youth, college market share of receiving yards, speed score, and bench press. Interestingly, college strength of schedule was insignificant to the model (p=0.365) in the test years (1999-2006), which may bode well for some of the smaller-school prospects on this list.

So let’s see which players come up short in one or more of the model’s components, in order to help you combat the recency bias of the combine. The player names are sorted by the RotoViz Draft Prospect Rank to help you spot TEs who might outperform their projected draft position:

Rotoviz_2_d

At a glance, you can see that the top seven players test near the upper tier of at least three of four of the critical factors for TE success. The aforementioned logistic regression model will weight these factors by their relative contribution to the (log) odds of long-term NFL success and estimate this probability. The scale below has been modified to show the log-odds of success. This visualization allows the reader to see the additive contribution of each factor in passing the estimated 15 percent chance of success that proved optimal in the validation years (2007-2011), which is all we care about here.

Rotoviz_2_e

Since bench press was observed to be categorical in its contribution to NFL TE success (21 or more, less than 21, did not participate), you can see how high bench reps might be the difference between passing or failing this “rule of 15 percent.” Yet still, six of the nine prospects above the threshold passed upon athleticism, youth, and college production alone.

So let’s rank our prospects according to their estimated success probability, with the understanding that this puts the player into approximate tiers, rather than saying each player is better than the next.

Rotoviz_2_f

Those three TEs I mentioned at the beginning with similar success probabilities to Howard and Njoku are now clear. Jonnu Smith, Bucky Hodges, and Adam Shaheen all project for at least a 60 percent chance of being a long-term NFL starter.

To those curious about Jake Butt, we cannot estimate his probability of success with this model since he wasn’t healthy enough to run at the combine. For the remaining top 15 prospects, we see that the top six players with elite bench numbers are in the upper tier. The next tier looks to be composed of the three players with modest bench press numbers but who still passed the model, and the remaining five players who failed the 15 percent threshold make up the bottom tier. That the tenth ranked TE, Jonnu Smith, falls in the highest tier is perhaps the very gold for which we were digging.

Conclusions

In a nutshell, this model suggests NFL starting TEs need to score well in three of four factors: youth, college production, athleticism, and strength. Yet in this class, we have six prospects score high on all four factors. Let’s compare our implied probability of success to prior draft classes to see how much better this year’s TEs project.

Rotoviz_2c

With more tight-ends in the upper-half of the probability scale this year than the past five years combined, there’s little doubt this is the strongest TE class in years. You shouldn’t need a chi-square test to tell you that six of 19 (32 percent, 2017) and five of 67 (seven percent, 2011-2016) are statistically significantly different (p = 0.012). However, such a p-value should tell you that the 2017 class is systematically different (younger, faster, stronger, more productive) from the past five years of incoming TEs.

To translate these scores to action, you might dig into lesser-known prospects like Jonnu Smith, Adam Shaheen, and Darrell Daniels to see what this model is detecting. However,  you’d be wise to avoid players like Jeremy Sprinkle, George Kittle and Michael Roberts as projected busts. Since this threshold of 15 percent success was so historically powerful at ruling players out, you might use the table below to screen future TE acquisitions.

DraftedPlayer NameHeightWeightBenchMSYAgeSPD_ScoreP(starter)Rule of 15%
2017O.J. Howard78251220.7320.7570.9360.6148PASS
2017David Njoku76246210.8070.9920.7990.6935PASS
2017Evan Engram75236190.7820.6870.9730.3373PASS
2017Bucky Hodges78257220.7560.9330.920.707PASS
2017Jordan Leggett77258180.6510.8030.760.2272PASS
2017Gerald Everett75239220.8560.6150.7840.5047PASS
2017Adam Shaheen78278240.9140.7250.8120.6132PASS
2017Jeremy Sprinkle77256.0.4860.6640.7180.0929FAIL
2017Michael Roberts76261.0.5140.5610.6140.0619FAIL
2017Jonnu Smith75245220.9830.9410.8720.7784PASS
2017George Kittle76247180.5840.1980.9380.0911FAIL
2017Blake Jarwin77246210.3240.6420.430.1257FAIL
2017Darrell Daniels75254170.3790.7550.9320.1868FAIL
2017Pharoah Brown78255240.5270.5690.3560.1342FAIL
2017Scott Orndoff77254.0.7810.2650.3360.0281FAIL
2017Hayden Plinke76258180.7350.0340.2490.0162FAIL
2017Evan Baylis77244170.2740.390.1480.0108FAIL
2017Cethan Carter75241190.3670.1580.6640.028FAIL
2016Hunter Henry77250210.8560.9650.7170.6565PASS
2016Austin Hooper76254190.6060.7810.7340.1923PASS
2016Ben Braunecker75252200.9870.3950.7320.1791PASS
2016Jerell Adams77244180.6440.3560.730.0879FAIL
2016Joshua Perkins75223120.6340.3860.4350.0457FAIL
2016Nick Vannett78256170.3710.3730.2110.0149FAIL
2015Jeff Heuerman77254260.8450.7770.4410.3767PASS
2015Jesse James79261260.610.9920.3880.3453PASS
2015Brian Parker76265250.7070.3530.8730.3381PASS
2015Mycole Pruitt74251170.9280.5220.9450.3068PASS
2015Maxx Williams76249170.9540.9880.4960.3038PASS
2015Nick O'Leary75252210.6720.6840.2090.1584PASS
2015Will Tye74257.0.8910.2210.9810.1447FAIL
2015James O'Shaughnessy76248160.6460.4480.7380.1085FAIL
2015Tyler Kroft78246170.7150.730.4030.1033FAIL
2015Matt Lengel79268260.2380.0380.9560.1024FAIL
2015Ben Koyack77255160.4030.8520.5250.097FAIL
2015Matt LaCosse78257200.3150.710.7020.0917FAIL
2015C.J. Uzomah78262190.2930.3370.9110.0651FAIL
2015Clive Walford76251200.8530.2070.5060.0568FAIL
2015A.J. Derby76255150.550.1670.7490.051FAIL
2015Nick Boyle76268200.5470.8190.150.0496FAIL
2015Geoff Swaim76252.0.1550.3280.620.018FAIL
2015Blake Bell78252140.3980.1340.3650.0135FAIL
2014Austin Seferian-Jenkins78262.0.9490.9510.9620.4931PASS
2014Eric Ebron76250.0.9290.9870.9280.4814PASS
2014Jace Amaro77265.0.9090.8990.8650.383PASS
2014Scott Simonson77249240.9510.8470.2280.3407PASS
2014Je'Ron Hamm75236.0.8870.8950.6560.2567PASS
2014Cameron Brate77249240.6080.6120.4620.2092PASS
2014C.J. Fiedorowicz78265.0.7560.7260.8020.2048PASS
2014Richard Rodgers76257.0.6470.9710.5380.1553PASS
2014Troy Niklas78270.0.630.9490.5680.154PASS
2014Nic Jacobs77269.0.5630.7230.8230.1524PASS
2014Crockett Gillmore78260180.8120.7450.3950.1256FAIL
2014D.J. Tialavea76267220.3350.6430.3750.1136FAIL
2014Xavier Grimble76257160.4170.950.2910.0712FAIL
2014Marcel Jensen78259240.3390.0510.5020.0439FAIL
2013Zach Ertz77249240.9550.7530.70.5818PASS
2013Tyler Eifert78250220.9110.7010.770.5731PASS
2013Travis Kelce77255220.8350.1890.9050.3323PASS
2013Dion Sims77276.0.7170.8260.9410.2976PASS
2013Vance McDonald76267.0.8650.6050.9720.2739PASS
2013Jerome Cunningham75250.0.7140.8870.7950.2508PASS
2013Nick Kasa78269.0.7520.7490.8610.2384PASS
2013Chris Gragg75244.0.5660.6270.9850.1794PASS
2013Luke Willson77251.0.7270.4580.9490.1598PASS
2013Brian Leonhardt77262210.9690.5370.0990.1595PASS
2013Jordan Reed75236.0.9580.6340.5860.1557PASS
2013Gavin Escobar78254.0.8860.8080.4850.1547PASS
2013Josh Hill77246200.6630.5840.7050.1371FAIL
2013Ryan Griffin78261210.6880.4550.3350.1368FAIL
2013Demetrius Harris78237.0.0640.9280.8670.101FAIL
2013Levine Toilolo80260190.420.9290.4240.0934FAIL
2013Joseph Fauria79259170.6540.4610.6520.0931FAIL
2013Mychal Rivera75242170.6170.6980.4050.0805FAIL
2013Jake Stoneburner75252180.5040.1430.9010.0636FAIL
2013Jack Doyle77254.0.9450.5660.1750.0513FAIL
2013Zach Sudfeld79253110.7650.0790.420.0292FAIL
2012James Hanna76252240.2720.640.9940.3494PASS
2012Michael Egnew77252.0.8660.7430.9510.3298PASS
2012Dwayne Allen75255270.6270.8290.4580.3103PASS
2012Orson Charles75251.0.7440.9780.7450.2792PASS
2012Coby Fleener78247270.7010.1710.9260.2752PASS
2012Cooper Helfet75240240.5370.5930.5740.2249PASS
2012Brandon Bostick75243.0.8830.5590.8760.2171PASS
2012Phillip Supernaw77242180.6610.8060.5360.1464FAIL
2012Chase Ford78255200.1640.9190.5780.0797FAIL
2012Danny Noble77247170.3890.5330.5910.0561FAIL
2012David Paulson75246.0.6160.4930.4730.0466FAIL
2012Cory Harkey76260140.4580.90.070.0409FAIL
2012Garrett Celek76250190.130.0970.6450.0143FAIL

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

By Phil Watkins | @Advantalytics | Archive

1 Comment

  1. May 12, 2017 at 6:36 pm —

    Update: At the time of publication, Adam Shaheen’s DOB was 10/24/1993 according to Zach Whitman. Since publication, it has been corrected to 10/24/1994. The difference of a year makes his normalized age 0.2202, which corresponds to a 33.3% chance of logging 64 or more NFL starts. This is still a healthy “PASS” according to the model.

Leave a reply