Tyler Lockett, Tyreek Hill, and the Many Uses of Slot Receivers

Seattle Seahawks WR Tyler Lockett
Seattle Seahawks WR Tyler Lockett
Photo: USA Today Sports Images

The term slot receiver conjures a specific image in my mind. It's of Wes Welker. The 2007 Patriots owe much of their historic offensive outburst to Tom Brady and Randy Moss, perhaps the best to ever play their respective positions. But the 5-foot-9, 185-pound Welker enjoyed his own breakout with the team, catching a league-leading 112 passes and cracking the top 10 wide receivers in PPR formats. Welker went undrafted out of Texas Tech in 2003 thanks to the physical tools one would expect of a player of his stature. His skills were short-area quickness and guile, things more difficult to measure than traditional combine tests such as the 40-yard dash. In that historic season, Welker provided a template for some of the best fantasy values of the last decade-plus such as his Patriots successor Julian Edelman (a seventh-round draftee), Doug Baldwin (undrafted), and Cooper Kupp (third round), all underdrafted at first in fantasy because they were underdrafted in real life.

Slot receivers still provide some of the best fantasy values in 2021. But as teams have embraced positionless football and become more creative in deploying their receivers in non-traditional alignments, that template has become more difficult to follow. Cole Beasley was an obvious slot receiver, and he produced his best fantasy season in 2020 when his quarterback Josh Allen transformed into an accurate passer. But CeeDee Lamb wasn't an obvious one. He's 6-foot-2 and an athletic marvel, and he went in the middle of the first round of last year's draft. Nevertheless, he matched Beasley with a 94% slot target rate in his freshman season. Sports Info Solutions charting makes it clear that every receiver spends some time in the slot. Even the best outside-capable receivers such Tyreek Hill (73%) and Davante Adams (59%) see more of their targets from the slot than out wide.

The grayness of real receiver roles made me realize that my former black-and-white projections choice to bucket receivers as slot receivers, wide receivers, and deep receivers missed the mark. Projected slot target rates offer a better solution. They ease the small-sample concerns of early-season opponent factors for teams that might have faced just one or two stereotypical slot or deep receivers. They regress receiver opponent adjustments when defenses fare differently in coverage against different types of receivers—a desirable outcome since it is difficult to predict which defenders will cover which receivers when matchups rarely repeat. And they even help identify No. 1 receivers (those likeliest to draw shadow coverage from the best non-nickel corners) in offenses that target a Welker-like slot receiver more than anyone else.

Without weekly, in-season access to actual slot target rate data, I used receiver characteristics to project slot rates. And despite the diversity of modern slot receivers, I found an algorithm that works by combining heights, weights, and average depths of targets:

Projected Slot% = 1.853 - 0.024*H - 0.024*B - 0.052*A

H=Height in Inches Over 70
A=Average Depth of Target

Correlation coefficient of projected vs. actual slot target rates of heavily targeted receivers from 2016-2019: 0.67

To demonstrate, here are last year's wide receivers with 100-plus targets with their projected slot rates for 2021 and their actual slot target rates from last year to compare.

Projected Slot Target Rates, 2021
Wide Receivers Coming Off 100-Target Seasons
Player Tm H B A Proj Slot%
Diontae Johnson PIT 1 25.2 9.27 74.2% 27%
Cole Beasley BUF 0 27.4 8.86 73.5% 94%
Robert Woods LAR 2 26.2 8.56 73.1% 82%
Russell Gage ATL 2 25.0 9.17 72.8% 77%
Jarvis Landry CLE 1 29.0 8.87 67.2% 81%
Tyler Boyd CIN 4 25.3 9.32 66.5% 91%
Cooper Kupp LAR 4 26.3 8.91 66.2% 90%
CeeDee Lamb DAL 4 24.3 9.85 66.2% 94%
Tyler Lockett SEA 0 26.1 10.99 65.5% 73%
JuJu Smith-Schuster PIT 3 28.4 8.56 65.4% 88%
Keenan Allen LAC 4 27.1 8.96 64.1% 67%
Tyreek Hill KC 0 26.5 11.47 62.1% 73%
Robby Anderson CAR 5 23.7 10.48 61.9% 64%
Davante Adams GB 3 28.4 9.30 61.6% 59%
Stefon Diggs BUF 2 25.9 11.11 60.6% 51%
Brandin Cooks HOU 0 27.1 11.47 60.6% 64%
Justin Jefferson MIN 5 24.0 10.61 60.5% 59%
DeAndre Hopkins ARI 3 28.4 9.75 59.2% 29%
Amari Cooper DAL 3 27.8 10.08 59.0% 48%
Allen Robinson CHI 5 26.2 10.00 58.4% 56%
Terry McLaurin WAS 3 27.0 10.59 58.2% 46%
Adam Thielen MIN 4 25.7 10.97 57.0% 59%
Jerry Jeudy DEN 3 25.3 11.67 56.7% 47%
Michael Gallup DAL 3 26.4 11.42 55.4% 23%
Tee Higgins CIN 6 25.6 10.57 54.5% 56%
DeVante Parker MIA 5 26.5 10.68 54.2% 34%
Calvin Ridley ATL 3 25.1 12.26 54.1% 37%
Marvin Jones DET 4 25.4 11.74 53.7% 67%
DJ Moore CAR 1 30.0 11.51 51.0% 53%
A.J. Green ARI 6 25.6 11.48 49.8% 24%
A.J. Brown TEN 3 30.3 10.96 48.4% 42%
Chase Claypool PIT 6 27.6 12.04 42.1% 42%
DK Metcalf SEA 6 28.0 12.29 39.8% 30%
Mike Evans TB 7 27.4 12.41 38.2% 39%
Actual Slot% data from Sports Info Solutions

Diontae Johnson makes me look bad immediately with his disparate 74.2% projected and 27% actual slot rates, But he is his own brand of outside receiver, and defenses will likely treat him that way. By and large, the projected and actual slot rates mirror each other. And that isn't just true of the Beasleys (73.4% vs. 94%) of the world. The algorithm makes strong predictions for players who do not fit the slot-receiver stereotype such as Keenan Allen (64.1% vs. 67%), Brandin Cooks (60.6% and 64%), and Allen Robinson (58.4% and 56%).

Just as important for the projections, the algorithm also recognizes the differences in receivers whose projected target rates on their own would not identify them as clear No. 1 or non-No. 1 receivers for their teams. Tyler Lockett and DK Metcalf illustrate that perfectly. We project Lockett for a higher percentage of catchable targets in Week 1 (31.0%) than Metcalf (28.5%). But because Lockett has a projected 65.5% slot rate and Metcalf has barely half that at 39.8%, Metcalf outstrips Lockett easily with a 17.2% versus a 10.7% projected wide target percentage, which then classifies Metcalf as the Seahawks' No. 1 receiver for the purposes of opponent adjustments.

Projected Catchable Target% Splits
Seahawks Wide Receivers
Week 1, 2021
Player All Wide Slot
Tyler Lockett 31.0% 10.7% 20.3%
DK Metcalf 28.5% 17.2% 11.3%
D'Wayne Eskridge 7.0% 2.8% 4.2%
Penny Hart 5.0% 1.7% 3.3%

That said, this approach does not preclude a heavier slot player from classifying as his team's No. 1 receiver. Tyreek Hill has a major slot rate projection of 62.1%, but that still leaves him with a 10.6% projected wide target rate that easily beats those of teammates Mecole Hardman and Demarcus Robinson.

Projected Catchable Target% Splits
Chiefs Wide Receivers
Week 1, 2021
Player All Wide Slot
Tyreek Hill 28.0% 10.6% 17.4%
Mecole Hardman 14.5% 4.1% 10.4%
Demarcus Robinson 13.0% 5.2% 7.8%
Byron Pringle 4.5% 1.8% 2.7%
Cornell Powell 3.0% 1.3% 1.7%

We use those projected slot rates and WR1 classifications to calculate defense factors, and then we use those defense factors to adjust our true-talent player projections for specific weekly matchups. As such, the true test of the change in projection approach comes from performance testing, and it did improve the correlation coefficients and root mean square error totals between our projected weekly fantasy point totals and actual fantasy point totals from the last few seasons. Still, I think the eye test is a great confirmation. Check out the various defensive yards per target factors. Those aim to predict the percentage of yards defenses allow over or under an average defense per target. Like park factors in baseball, they express those differences as percentages over and under an average of 100.0.

Slot Defense Factors

Once again, a top entry on the leaderboard raises some suspicion. The Patriots have an elite cover corner in Stephon Gilmore, but by the end of 2020, they allowed 28.5% more yards per target to No. 1 receivers than an average team? But that result makes more sense than at first blush. Gilmore missed a third of last season with knee and quad injuries. That forced him to miss all of the similarly named Stefon Diggs' 20 targets, 237 yards, and three touchdowns in his two Patriots matchups. And even when Gilmore played, he did not always shadow his opponent's perceived No. 1 receiver. He defended just two of Keenan Allen's 11 targets and one of Mike Williams' nine targets in Week 13. He did not defend any of Tyreek Hill's six targets in Week 4. The Patriots follow different strategies with their corners that sometimes have them double-teaming No. 1 receivers with two lesser coverage options and sometimes have them assigning their faster slot corner Jonathan Jones to the speediest of their opposing receivers. Really, the Patriots are a perfect example of why opponent adjustments split on projected slot target rates make more sense than those split on strict buckets of receiver roles.

Overwhelmingly for me, the WR1 yards per target factors pass the eye test. The Eagles, Vikings, Lions, Falcons, and Jets were infamous for their poor coverage units in 2020. They were all 14th or worse in pass defense DVOA—and so were the Patriots. In contrast, the Rams, Bengals, Ravens, and Saints employed some of the best—and healthiest—cover corners in football in Jalen Ramsey, William Jackson, Marlon Humphrey, and Marshon Lattimore, three-quarters of whom finished in the top 10 in either yards per target allowed or coverage success rate.

These opponent factors can change dramatically from season to season as players switch teams and within seasons as players suffer injuries, improve, and decline. They can be difficult to keep track of, especially since they affect efficiencies of myriad passing, rushing, and receiving statistics. That's why we try to simplify things in the presentation of our weekly fantasy projections. The summary view illustrates how opponent adjustments—as well as venue and weather adjustments—take players from their base projected fantasy point totals to the context-adjusted ones that best reflect our predictions for the week. And by showing that work, we allow you to sort the tables to find which players are poised to enjoy the biggest benefits and suffer the biggest penalties each week and make your best start and sit decisions in traditional and daily formats.


2 comments, Last at 27 Aug 2021, 6:26pm

1 Regression specification

I don't mean to be super critical, because ultimately I think this is interesting work. But, I do believe you regression specification likely suffers from problems with endogeneity.

Depth of target should be at least partially determined by whether or not the player lined up in the slot or out wide. In other words, the left hand side variable causes the right hand side variable in your regression, not the other way around. That's a no-no, so just be careful how you use and interpret the results of that one.