McBride Leads Inaugural Travis Ratings

Colorado St. TE Trey McBride
Colorado St. TE Trey McBride
Photo: USA Today Sports Images

NFL Draft - Trey McBride of Colorado State is the clear leader among this year's tight end prospects, according to Football Outsiders' new Travis projections for college tight ends. But he falls far short of the top receiving tight end prospects of the last 20 years.

Yes, it's time to introduce another Football Outsiders rookie projection system to join our usual lineup of QBASE, BackCAST, Playmaker Score, and SackSEER. Our new system projects receiving yardage for tight ends and I'm calling it Travis. That stands for:

Tight end Receiving Value System

I could capitalize it something like TRaViS but I decided that looks weird so we'll go with just "Travis" for now. You know who it is named after, although the system didn't exactly do a great job of picking out Mr. Kelce as a sleeper prospect. We'll get to that in a minute.

As always, Football Outsiders' 2022 NFL draft coverage is presented by Underdog Fantasy!

Underdog Fantasy

Here is a look at the Travis scores for this year's tight end prospects. Each Travis score represents a projection of how many receiving yards that player will gain per year through his first five NFL seasons, i.e. the same output as Playmaker Score.

Travis Scores, 2022
First Last College Scouts
Inc. Rk
Trey McBride Colorado St. 59 395
Greg Dulcich UCLA 79 250
Isaiah Likely Coastal Carolina 88 231
Daniel Bellinger San Diego St. 127 197
Jelani Woods Virginia 103 190
Jeremy Ruckert Ohio St. 113 179
Cade Otton Washington 123 172
Charlie Kolar Iowa St. 141 164
Cole Turner Nevada 109 154
James Mitchell Virginia Tech 168 139
Chigoziem Okonkwo Maryland 277 118
Jake Ferguson Wisconsin 149 116
Grant Calcaterra SMU 227 114
Teagan Quitoriano Oregon St. 293 86
Austin Allen Nebraska 217 86
Jalen Wydermyer Texas A&M 183 85
Derrick Deese San Jose St. 235 51

This is not a good historical year for tight end prospects. Most years will have at least one prospect over 400 yards. For this season, we just have McBride close to that and then a big drop to the next set of prospects.

The main element in the Travis scores is projected draft position. For most of the data, I used NFL Draft Scout's projected rounds, similar to how projected draft position is used in SackSEER and Playmaker Score. In recent years, we used instead a translation of the player's rank on the Scouts Inc. big board on ESPN+. Tight ends projected to go in the top three rounds get more of a boost. Historically, pre-draft ratings have done a pretty good job of judging which tight ends would be successful in the NFL.

The second element in Travis is the player's best season in receiving yardage divided by total team pass attempts. Again, this is similar to Playmaker Score. Greg Dulcich's 725 yards on 335 total UCLA passes is more impressive than Charlie Jolar's 756 yards on 444 Iowa State passes. (Greg Dulcich's moustache is even more impressive than his 725 receiving yards.) McBride and Isaiah Likely had the best scores in this metric last season.

The next element is something we don't use in Playmaker Score: 40-yard dash time. I used pro day times for players who didn't run at the combine. For wide receivers, we don't use 40 time because it ends up predicting success for the same players who do well in the other elements of Playmaker Score (yards per team pass attempt, rushing attempts, etc.). For tight ends, using 40 times did improve accuracy of Travis. McBride had the most impressive 40 this year with 4.56 at his pro day. Jelani Woods of Virginia had 4.61.

The final element of Travis is age under 23 as of September 1 of the player's rookie year. Particularly young players have a better track record of development beyond their college performnaces. This seems to even out around age 23, which is the typical age for a player who was in college for four seasons. I used age instead of a player's class (junior, sophomore, etc.) because COVID has made figuring out what class a player is in a bit complicated. Daniel Bellinger gets a bit of a boost as the youngest member of this year's tight end draft class.

The big surprise here is that touchdowns didn't come out as a significant variable in predicting NFL success for tight ends. No matter whether I made the dependent variable total yardage or total fantasy points, touchdowns did not come out as significant, so you won't find them in the Travis projection equation.

Here's a look at the top tight end prospects of the last 20 years according to Travis scores. Most of these players were first-round picks. Ben Troupe, who went in the second round, 40th overall to the Titans in 2004, stands out as a bust, although current Cleveland tight end David Njoku has certainly been disappointing.

Top Travis Scores, 2001-2020
First Last Year Travis Yr 1-5
RecYd Avg
Vernon Davis 2006 593 602
Eric Ebron 2014 570 564
Noah Fant 2019 567 635
Kellen Winslow 2004 564 493
Evan Engram 2017 559 605
David Njoku 2017 557 320
Todd Heap 2001 550 579
Greg Olsen 2007 536 504
O.J. Howard 2017 532 401
Jeremy Shockey 2002 525 727
Ben Troupe 2004 525 212
Jermaine Gresham 2010 521 544

Since "projected draft position" is the most important variable in Travis, obviously the system is going to give the best forecast to prospective first-round picks. Most of the players on that list of Top Travis Scores won't surprise anyone. What about players who went later in the draft? Players that Travis projected better than their actual draft position are highlighted by Jason Witten (a third-round selection with a Travis of 462), Jared Cook (392), Aaron Hernandez (313), and Ben Utecht (Travis of 153 despite going undrafted). Of course, there were also misses. Anybody a fan of 2019 seventh-round pick Caleb Wilson around here? He had a Travis of 179 and has zero NFL yards. High picks that had somewhat lukewarm projections from Travis included Jerramy Stevens, John Carlson, Gavin Escobar, and Richard Quinn. Similar to a lot of Football Outsiders' projection systems for rookies, Travis appears to be a little better at picking out busts than picking out sleepers.

Travis Kelce, the namesake of the system, had a Travis projection of 287, right between Austin Hooper and Jimmy Graham. So, that one was a miss.


29 comments, Last at 24 Apr 2022, 11:18am

1 Hello?

Ever heard of Gronk?  If he did not even make it in your ratings that is a bigger miss than anything scouts did!

3 Tight end Receiving Value…

Tight end Receiving Value System

I could capitalize it something like TRaViS

Except there is no 'a' between the R and the V, and no 'i' between the V and the S!

It's interesting that it sees Engram and Ebron as basically about their projection, even if both teams see them as disappointments. Winslow, too, although perhaps he underperformed in longevity after losing his damned mind.


As an aside: if you looked at performance on a per-snap basis, does that lead to draft position no longer being a significant variable? On a bunch of these, it seems like the best predictor for bulk performance is sunk-cost. Is it just sunk-cost, or are teams actually semi-optimal at assessing future performance? Or are all positions actually as fungible as RB, it's just teams are still stupid about the others?

5 Might I suggest

Tight end Receiving Actualization Value Insight System. Or maybe Tight end Receiving Adjudication Value Insight System. Or possibly even Tight end Receiving Added Value Insight System. I'm struggling with the 'A' - lots of things sort of, but not exactly, fit.

6 Blocking??

It is still an important component of the job, otherwise teams would all go 4-wide. The only marker I see for that is draft position. Is there simply no way to prognosticate 'Ralph blocks such-and-such well; Fred only such-well'?

7 Scouting reports generally…

In reply to by BigRichie

Scouting reports generally cover it, but I'm not sure if it's easily available data anywhere, other than maybe the advanced scouting services. Also, teams tend to value it differently. 

12 Well then ...

In reply to by Aaron Schatz

You do measure it once they get to the NFL, right? (honest question, tho' I figure 'yes') So is there any bust correlation there, where turned out the busters in the whole sure blocked lousy?

13 I'm not sure there is an…

In reply to by BigRichie

I'm not sure there is an objective block score system. PFF does grades, but as Muth discusses fairly often in his arguments, without knowing the assignment, it's tough sometimes telling who screwed up, even if it's obvious that someone did.

I'm sure teams track it, but:

  1. they ain't tellin'
  2. each team may score the same play/player differently

16 If I recall correctly

FO does tabulate missed blocks, pressures allowed and penalties. (or else has a buddy that does) Pressures allowed will matter little for tight ends, and I dunno, maybe there's not much variance in penalties either. But does FO then track missed blocks for tight ends?

20 It's honestly easier to tell…

It's honestly easier to tell who is a strong blocker and who isn't based on how they are used than it is based on the result of the play. (assuming some base level of functional team) If they are sometimes assigned to DEnds in the run game, one on one assignments in passing, etc vs being used always to chip and help out other linemen and such. 

18 Cole Turner

I would like the Packers to take a flier on him in the mid rounds because there's some Jermichael in him. 

23 I asked a similar question…

I asked a similar question on the Playmaker article, but I’ll ask again here - is there a reason why “receiving yardage divided by total team pass attempts” is being used as an input variable instead of yards per route run? 

24 My guess would be…

My guess would be availability of historical data. Team passing attempts is available for probably a century. Routes run, or even just snap count data I haven't looked for, but if it's out there I doubt we've got 20 years of it and they like to use at least 20 years of data to build the models. If the snap count data is readily available for college football then yeah I would think it would be a better input.

27 This was mentioned as a…

In reply to by DisplacedPackerFan

This was mentioned as a possible reason in the other thread. However, you also have to consider the opposite - college football is very different today than it was in 2001, and it’s possible that including guys like Jerramy Stevens and Kellen Winslow actually hinders  the model’s predictive accuracy. 

26 Not obvious which would be…

Not obvious which would be better.

If 2 receivers have identical YPRR, then I'd guess that the one who was more regularly on the field running routes is a better receiver than the one who spent more time on the bench, or blocking.

If 2 receivers have identical YPTA, then I'd guess that the one who had better per route production is a better receiver than the one who needed more routes to get there.

So maybe somewhere in between.