Stat Analysis

Advanced analytics on player and team performance

Introducing Targets Above Expectation

Guest column by Bryce Rossler, Sports Info Solutions

Considering the shortcomings in the current metrics available to evaluate receiver performance, SIS sought to develop a metric that would characterize a receiver's ability to earn targets based on his role within an offense. Statistics like Target Share and Yards per Route Run, while valuable, are results-based measures that do not account for play-level information.

Examining the game situation, pre-snap alignment, route, and coverage paints a better picture of how a receiver performed relative to the league average. For example, a slot receiver who is running an out route on first-and-10 against Cover-3 is expected to be targeted X percent of the time. That figure is then compared to whether or not the receiver is actually targeted, and all these comparisons are tallied and expressed on a per-100 route basis.

Screens and jet sweep passes are eliminated from the sample because they have a target rate that is virtually 100 percent for one pass-catcher, making it problematic to assess anyone else running a route on the play.

The resulting statistic is called Targets Above Expectation (TAE):

Leaders in TAE, 2018 (min. 150 routes)
Player Team TAE
Julio Jones ATL 8.1
Keenan Allen LAC 7.8
Odell Beckham NYG 7.3
A.J. Green CIN 7.2
Jarvis Landry CLE 5.6

The bottom five players in TAE -- Chris Conley (-7.6), Chris Hogan (-7.7), Chris Moore (-8.8), Bennie Fowler (-9.8), and T.J. Jones (-11.2) -- are not a particularly interesting bunch, but what is interesting is that Tyler Lockett fared poorly in this metric, ranking 99th out of 119 eligible players.

Lockett had the best season by any wide receiver in Football Outsiders' DVOA history (min. 50 targets). Furthermore, he had 14 catches and six touchdowns on passes thrown at least 20 yards downfield, each of which tied for second in the NFL. His 82 percent catch rate and perfect passer rating of 158.3 on those throws were both NFL bests.

There are a couple reasons for this, with the first being that more than 10 percent of Lockett's targets came on broken plays, which we do not factor into the formula because they speak more to scramble drill performance than to route performance. The second is that Lockett ran 44 flat routes -- which is arguably not an inefficient way to use him -- and was targeted just once on such routes. Lastly, he was arguably under-targeted downfield, as he posted a TAE of -1.4 on such routes. These factors combined to give a lower-volume, high-efficiency player a very low TAE.

This metric can also be calculated on a route level, a situational level, an alignment level, and a coverage level, and it can be used to perform a variety of analyses, some of which will be highlighted below.

Identifying Inefficiencies in Player Utilization

By applying additional filters to the TAE metric, it's possible to gain an idea of how a player's use could be optimized moving forward. Further defining alignment, coverage type (i.e., man vs. zone), and/or route provides insight into how a player wins and what role he may be best suited for.

For example, Carolina Panthers receiver Devin Funchess (who has since signed with the Indianapolis Colts in free agency) averaged 8.9 TAE on 6.4 routes per game as a slot receiver. However, he spent most of his time on the outside (17.2 routes/game), where he ranked 62nd of 127 eligible receivers in TAE (-1.36). This was in spite of the fact that Funchess earned targets from the slot at the fourth-highest rate in the league:

Slot Leaderboard, 2018 (min. 5 routes/game)
Player Team Slot TAE Rank (of 135) Routes/Game
Antonio Brown PIT 12.5 1st 10.4
A.J. Green CIN 12.0 2nd 10.6
Julio Jones ATL 11.1 3rd 15.5
Devin Funchess CAR 8.9 4th 6.4
T.Y. Hilton IND 7.9 5th 13.6

While this metric should not be interpreted to mean that Devin Funchess is the fourth-best slot receiver in football, it does indicate that Carolina may have deployed its personnel inefficiently last year, especially when you consider that Jarius Wright and D.J. Moore averaged 19.9 and 7.6 routes from the slot per game and ranked 72nd and 105th in slot TAE, respectively.

These analyses can be conducted on a route level, as well. Philadelphia's Alshon Jeffery is one player whose performance could have benefitted from a reimagined route tree.

Jeffrey was targeted at a well above-average rate on drags, curls, flats, slants, outs, and whips, ranking third of 90 eligible receivers (min. 100 routes). He ran 141 such routes, posting a 9.9 TAE rate, but spent more time running posts, corners, and nine routes, where he ranked 38th of 62 receivers with a -2.8 TAE on 157 routes.

While it's not realistic to limit a player to just a handful of routes, using TAE to identify where a player wins is a potential way to identify how to maximize his skill set.

Identifying Potential Roster Building Alternatives

The Tampa Bay Buccaneers, who released DeSean Jackson, may have felt that third-year receiver Chris Godwin is capable of duplicating Jackson's role in the offense and will be ready to assume a larger role in 2019. Of course, Tampa Bay was able to save $10 million in cap space by cutting Jackson, but he and Godwin did indeed earn targets very similarly on seven, eight, and nine routes in 2018:

Jackson and Godwin TAE Ranks on Deep Routes, 2018 (min. 100 routes)
Player Team TAE Rank (of 62)
DeSean Jackson TB 4.5 8th
Chris Godwin TB 4.4 9th

Identifying cost-controlled options who can earn targets at a similar rate and in a similar fashion to another player could potentially reveal cost-saving avenues.

TAE Season-to-Season Predictiveness

As the graphs below illustrate, a relationship exists between TAE and subsequent season receiving yardage totals for players with a minimum of 150 routes in both season N and N+1. The correlation coefficient for the 2016-2017 and 2017-2018 pairings combined was 0.52:

Helping Visualize Collegiate Performance

Just as it is for NFL receivers, TAE is descriptive for NCAA receivers.

Draft Prospects Among Top 25 in TAE, 2018 (min. 150 routes vs. Power 5 competition)
Player School TAE Rank
Dillon Mitchell Oregon 19.9 1st
Jakobi Meyers North Carolina State 8.9 6th
Stanley Morgan, Jr. Nebraska 7.7 8th
Kelvin Harmon North Carolina State 6.9 11th
J.J. Arcega-Whiteside Stanford 5.7 12th
N'keal Harry Arizona State 5.3 13th
David Sills V West Virginia 5.0
14th
Hakeem Butler Iowa State 5.8 19th
Jalen Hurd Baylor 3.8 20th
A.J. Brown Ole Miss 3.3 25th

Ohio State's Parris Campbell (27th) and South Carolina's Deebo Samuel (28th) fall just outside the Top 25. Only one draft-eligible senior among that group projected to be undrafted (Northwestern's Nagel Flynn); the rest are underclassmen.

Oregon's Dillon Mitchell, now with the Minnesota Vikings, sits comfortably atop the leaderboard; TCU's Jalen Reagor, who was ineligible for the draft this year, ranks second with 14.1 TAE.

Using the full suite of SIS stats, it's possible to further contextualize why a player who appears so dominant has not been reported to be valued highly by teams. The numbers indicate that Mitchell struggles with tracking the football, as he dropped 14.3 percent of his catchable deep (20-plus yards downfield) targets and failed to secure any throws that were charted as off-target but catchable (outside the frame or requiring adjustment) in 2018. This would suggest that Mitchell may have more value to a team that has either a quarterback with high-level accuracy, an established deep-ball tracker, or both.

There isn't currently enough longitudinal data to assess how predictive college TAE is of NFL TAE in the long-term, but the metric's ability to predict yardage in season N+1 for NFL players is reason to be optimistic in this regard.

Conclusion

TAE is useful for supplementing existing evaluation methods in the sense that it allows us to not just quantify how well a player performed, but to contextualize that performance as well. These analyses can be conducted for both collegiate and professional players and can be split by situation, route, coverage, and alignment.

A correlation exists between TAE and N+1 season yardage at the pro level, and there's optimism that a similar trend will emerge between the college and pros with a larger longitudinal college data pool.

Comments

7 comments, Last at 14 May 2019, 6:44pm

1 Re: Introducing Targets Above Expectation

I wonder if this could be used to convert the current per-target DVOA for receivers into a per-expected-target DVOA.

What value would we assign to extra targets? Or to missing ones?

Getting open enough to be a target is a large part of a receiver's job.

3 Re: Introducing Targets Above Expectation

I believe DVOA will punish a receiver if the ball is forced to him too often, since presumably forced passes have a lower success rate. If the receiver is getting targets above expectations, that might suggest the QB is forcing the ball to that receiver out of confidence in the receiver. Therefore, I would speculate that TAE incorporated into DVOA might better reflect the receiver's talent than DVOA alone.

I'm not sure, however, it would tell us anything that DYAR doesn't, when meshed with DVOA, as DYAR, as a counting measure, presumably captures the impact of more targets.

So in other words, my hypothesis is that DVOA incorporating TAE will give us roughly a similar ranking of individual receivers that DYAR does. But I'm far from experienced in these metrics, and I might be missing a key factor or two.