Tyreek Hill Reigns Supreme with Route DVOA

When I first created my advanced stats DVOA and DYAR for receivers 20 years ago, there was a bit of a breakthrough about which plays we were counting. At the time, there wasn’t much of a concept of “targets” in the public data. Receivers were tracked by receptions and yards per reception. Official gamebooks did not count targets, and they did not appear on the original version of Pro Football Reference in 2003. DVOA and DYAR were among the earliest stats on the Internet that considered both complete and incomplete passes to wide receivers. That’s how we’ve computed receiving DVOA and DYAR ever since.

The Problem with Targets

Since then, there’s been another breakthrough about counting plays for receivers. With the rise of charting companies such as Pro Football Focus and Sports Info Solutions (and now FTN Data) came a count of routes run, not just targets. Gradually, it became clear that earning targets was a skill itself. Stats which used “targets” as the denominator had a problem where they didn’t really account for this skill. Stats which used “routes run” as the denominator didn’t have that problem.

Often, you would have two receivers who ran a similar number of routes with a very different number of targets. The more prolific receiver might have a lower catch rate or even fewer yards per reception and thus perform worse in a stat (such as DVOA) with targets as the denominator. But he would have more yards and more production and thus perform better in a stat with routes as the denominator. 

To give an example from last season, Tyler Boyd of the Bengals ran an estimated 497 routes during the regular season. (Why “estimated”? We’ll get to that in a little bit.) DK Metcalf of the Seahawks ran an estimated 496 routes. Now look at the difference between the two players despite a similar number of routes.

Tyler Boyd has a higher catch rate and more yards per reception, so it’s not a surprise that he does better in DYAR and DVOA. But DK Metcalf earned many more targets compared to Boyd and does better in yards per route run. Conventional DVOA and DYAR are missing out on Metcalf’s ability to get open and earn targets and thus more value.

It’s time to drag DVOA and DYAR into the modern era by looking at routes run instead of just targets. And so that was one of my main projects for this past offseason.

Introducing Route DVOA

In the future, I will have access to FTN Data’s charting of routes run to get a more accurate count of routes run. For this project, however, I primarily used publicly available data. That’s the reason for the word “estimated” when I give the count of routes for Boyd and Metcalf above. To determine routes run, I used the Next Gen Stats play-by-play participation data that is available as part of nflverse on github. Alex Vigdorman from Sports Info Solutions was nice enough to pull participation data for three games over the last three years where the data is missing from nflverse.

When wide receivers are on the field for a pass play, they almost always run a route. Yes, there are RPOs and screens where certain receivers are blocking, but almost every snap on a pass is a route. This is not true for tight ends and running backs, which is why I haven’t included those positions in this project yet. But for wide receivers, I made my life easy. If they were in the lineup on a certain play, I counted that as a route.

Every player who ran a route on a play then got counted for DVOA or DYAR. Just like in regular DVOA, an incomplete pass counted as 0 success points. But unlike in regular DVOA, any route run without a catch counted as 0 success points as well. We determined that there are an average of 4.34 routes run on any given pass play, so we divided all the DVOA baselines by 4.34. This produced both DVOA and DYAR numbers based on routes instead of targets.

Now, let’s take a look at Boyd and Metcalf with our new metric.

This seems to be a much more accurate portrayal of how important each receiver is to his offense. It’s not surprising that Metcalf now comes out much better. In fact, it’s a bit surprising that Boyd drops so much, but compare his 1.53 yards per (estimated) route run to the NFL average of 1.80.

The Top Wide Receivers of 2022

When we look at the top wide receivers of 2022 with our new Route DYAR metric, Tyreek Hill really stands out. Hill led the league in yards per route run. He didn’t have the most production in the league, but he ranked second in yardage behind Justin Jefferson despite running over 130 fewer routes. Kevin Cole had Hill as the Offensive Player of the Year in his new offensive plus/minus ratings this offseason and Hill also ends up with a sizeable lead in Route DYAR. Here’s a look at the top (baker’s) dozen for 2022:

Justin Jefferson had a ton of catches and yardage last season, but he also ran a ton of routes. That’s why Jefferson falls behind Hill in Route DYAR and falls behind both Hill and Stefan Diggs in Route DVOA. A number of very active players come out much better in Route DVOA than they did in the original receiving DVOA. Davante Adams was 39th out of 97 qualifying wide receivers last year in standard receiving DVOA, but shoots up to fifth in Route DVOA, just behind Jefferson and Amon-Ra St. Brown. Drake London, who played such a huge role in the Atlanta passing game – his target rate was 33.4% of routes, second behind Hill – goes from average in original DVOA to ninth in Route DVOA. (The minimum to qualify for rankings here is either 250 routes or 50 targets.)

Two things you’ll notice about the new Route DYAR and Route DVOA. First, these new stats have much larger standard deviation than the old receiving DYAR and DVOA. Second, because the volume of catches is so important for Route DVOA, the rankings for Route DYAR (a total stat) are closer to those for Route DVOA (a rate stat) than the rankings for original DYAR are to original DVOA.

Hill, Adams and London are among the players who get the biggest boost when we switch from original receiving DVOA to Route DVOA, but they are certainly not the only ones. Because of that larger standard deviation, it didn’t make sense to just look at the difference between the two versions of DVOA. Instead, I looked at “z scores,” which measure the difference in number of standard deviations over average with each metric.

Here is a look at the top 10 risers when we go from original to Route DVOA.

The original receiving DVOA suggested that DeAndre Hopkins was disappointing when healthy in 2022. However, Hopkins was one of only nine qualified receivers to earn targets on at least 30% of routes last season. Consider him with Route DVOA instead of original DVOA, and his performance goes from replacement level to nicely above average. Route DVOA suggests that Hopkins didn’t really lose much of his previous performance.

In original receiving DVOA, Hopkins’ rank rises from 38th in 2020 to seventh in 2021, and then drops to 77th in 2022.

In Route DVOA, Hopkins’ rank falls slowly from seventh in 2020 to 15th in 2021 and then 20th in 2022.

Diontae Johnson of the Steelers is another receiver who really stands out when we look at the changes from original DVOA to Route DVOA. We know that Johnson scores very highly in the ESPN receiver metrics based on Next Gen Stats tracking. So it feels wrong to see him all the way down at a dismal 89th in original receiving DVOA for 2022. Route DVOA recognizes that Johnson was by far the most important receiver in the Steelers offense, getting targets on 27.4% of his routes, and boosts him up just above average. In conjunction, his teammate George Pickens – who only earned targets on 16.6% of his routes – is among the biggest fallers when we go from original to Route DVOA.

DeVante Parker of the Patriots takes the biggest fall when we go to the new Route DVOA. Parker averaged a fantastic 17.3 yards per reception last year, so of course he was going to come out strong in receiving DVOA. Just over half of his receptions went for first downs. But compared to other nominal No. 1 receivers, he simply did not draw many targets. Jakobi Meyers, Nelson Agholor, and Kendrick Bourne were all targeted on a higher rate of their routes than Parker was.

Route DVOA = More Information Than Original DVOA

A couple of paragraphs ago, you may have noticed that DeAndre Hopkins‘ Route DVOA has been much more consistent than his original receiving DVOA over the last three seasons. This is true for most players. The best indication that we’re measuring something real here is that Route DVOA is far stickier than original DVOA. So far, I’ve computed Route DVOA for the years 2020 through 2022. Looking at players with a minimum of 200 routes in each year, the year-to-year correlation of original receiving DVOA is 0.22. The year-to-year correlation of Route DVOA, on the other hand, is a whopping 0.60. Getting year-to-year correlations that high is very rare in the football world. The year-to-year correlation of Route DVOA is even higher than that for yards per route run (0.51), at least among these players in these three seasons.

Looking for a connection to your fantasy football team or DFS contests? Route DVOA and DYAR also appear to be better metrics for predicting future receiving yardage or fantasy points. Here are the correlations of a number of stats to fantasy points (0.5 PPR) in the following season among players who hit our 200-route minimum in consecutive years:

• Original receiving DVOA: 0.28
• Routes: 0.42
• Yards per Route Run: 0.50
• Original receiving DYAR: 0.51
• New Route DVOA: 0.56
• New Route DYAR: 0.61

This Google Sheet has all the standard and Route DVOA/DYAR receiving numbers for players in 2022.

Where do we go from here? First, once we’ve established all our regular DVOA and DYAR ratings and have them updating on FTN each week, I’ll work on gathering the FTN Data charting of routes and creating Route DVOA and DYAR each week. Having the actual charting of routes will also allow us to produce Route DVOA and DYAR for tight ends and running backs.

After that, the next step will be to adjust route DVOA by the personnel groupings used on each play. It’s been shown numerous times that average yards per route run changes significantly based on personnel. The more wide receivers, the fewer the yards per route run since only one of those receivers can theoretically gain yardage on any particular play. (The linked article also shows how down and distance affect yards per route run, but of course DVOA controls for that already.)

The downside of Route DVOA is that we can’t use it historically. We have DVOA and DYAR going all the way back to the 1981 season, but we only have routes run for a few years. It would be very interesting to see what Route DVOA would think of a player such as Chris Chambers, who has some historically awful seasons in standard receiving DVOA because he drew a very high rate of his offense’s targets… and then did not catch them. In 2006, for example, Chambers drew 26% of Miami’s targets (which would have ranked in the top 10 last season) and only had a catch rate of 39%. That season comes out with an all-time low of -291 DYAR.