Thoughts on WR Stats

In the last post I discussed some questions about how to measure unsuccessful plays for receivers. Typically, receiver performance is measured by total receiving yards, but that's not a very good stat for a lot of reasons, primarily because it ignores negative plays. But should all incomplete passes count against the receiver? Maybe he dropped a good pass, or maybe he ran a bad route. What about interceptions? Those are highly negative plays which sometimes are the fault of the receiver.

I think the consensus from the comments was that all plays, including unsuccessful ones, should count for and against the receiver. We already have conventional stats that count up only successful plays--total yards and yards per reception. Plus, many times the receiver is at least partly to blame for negative outcomes. We count drops against quarterbacks in their stats, and we count interceptions that are due to bad routes or lackluster efforts by receivers against the passer, so completeness dictates we treat receivers the same way.

But as I said in the previous post, the right thing to do is to start by articulating what we are asking the stat to measure. What is it that we want to know?

Football is the team sport of all team sports. As I've written several times, it is just unrealistic to think we can truly separate the contributions of individual players from his teammates with a stat. I think the best we can do is ask, "How good or bad are the outcomes when a team throws to a particular receiver?" In the end, it's not an individual stat but a 'partial' team stat. That's why it's better to include incomplete passes and even interceptions.

I'm asking all these questions because I'm developing a database of advanced individual stats. For QBs and RBs, the concepts are pretty straightforward, but for WRs and TEs it's a little tricky. The individual stats will be the WPA and EPA stats that I've been throwing around lately, plus a few more that I'll introduce shortly.

There were some really great thoughts and suggestions in the responses to the previous post, so let me touch on some of them.

JMM emphasized that all stats are really team stats, and I would agree. Alchemist correctly points out the major flaw with most individual football stats when he wrote, "When a WR is open but is not thrown to, this looks on the play-by-play as a non-play. But of course, the WR was contributing as much as he could. When a WR is given close double coverage and is not thrown the ball, he has contributed handsomely to the play by drawing 2 defenders away from the action. But this looks on the play-by-play to be a non-play for the WR. When a WR throws a key block, this is not reflected on the play-by-play."

Combining those two thoughts points in  the direction of some kind of +/- or With or Without You (WOWY) stat like they use in hockey and basketball stats. It would be great if we could do the same thing with football, but the nature of the sport doesn't offer the same continuity and flow required for a WOWY stat that the other sports do. The sample sizes are usually far too low as well. Only in unique circumstances could we really do a WOWY stat for football.

Lab suggests we track the data both ways, with and without negative outcomes. I think that's smart and I'll do that in house.

James suggested breaking up a passing play into its Air Yards and Yards After Catch. That would be fun to look at, but the play-by-play doesn't break out the YAC.

A few guys suggested breaking apart the blame for bad plays between the passer and receiver according to a weighted formula or other technique. That's definitely worth looking into. Some comments worried that counting all passing stats for both passers and receivers would end up double-counting passes when we look at a team's passing stats. That's not how it would work. I can tally up passing outcomes on a team basis, counting each play once, and then separately count pass plays fully for both passers and receivers.

Huw and an anonymous commenter both suggested some sort of 'par' system. Receiver performance would be compared to a baseline for their team's passing game as a whole, which would somewhat account for how good or bad the other components of the team are--the passer, the line, etc. We can also look at the proportion of targets a receiver draws for his team. That doesn't solve the problem of good receivers drawing double teams, but it does tell us about his ability to get open.

Another anonymous commenter pointed out that the play-by-play roughly identifies the location of the field where each pass was thrown. Because deep passes are harder to complete, we can break out a receivers' performance in deep pass situations. I've been thinking about a stat akin to Bill James' "range factor" for baseball that would factor in target percentage, catch rates, and depth of throws. More on that later.

I generally don't believe in one overall stat that encompasses everything. There are various dimensions to any player's performance, and we can learn more if we can look at each dimension separately. I like looking at a 'line' of stats to get an impression of what kind of player a guy is.

In the end, individual stats for football are always going to be flawed, and receiver stats highlight the difficulties in making sense of them. But they still have value. Ask yourself this: How do you know Andre Johnson is a really good receiver? Is it because of your keen professional scouting eye that can detect the greatness of a player in the one game the Texans are on TV each year? Or is it because of his monstrous stats he consistently puts up? There are hundreds of players in the NFL, and it's impossible to measure them all qualitatively. At the very least individual stats are a good starting point, and the better the stat, the better the starting point.

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

7 Responses to “Thoughts on WR Stats”

  1. Anonymous says:

    "Is it because of your keen professional scouting eye that can detect the greatness of a player in the ONE GAME the Texans are on TV each year?"

    My poor Texans :(. Their day will come.

  2. Anonymous says:

    >>How do you know Andre Johnson is a really good receiver?

    Because I heard a guy in the bar telling his friend that AJ is greal.


    I am somewhat skeptical of this proposed WR stat, but will keep an open mind.

    I think additional data is needed that is not easily available. Then more interesting advanced stats can be tracked.

  3. Zach says:

    I think you should also adjust for the QB's performance when he throws to his other teammates. Is Pierre Garcon really that good, or is it Peyton Manning? Well, see how Peyton does when he throws to Garcon (stats like YPA, comp%, WPA per attempt, etc), and compare that to how he fares when he throws to Wayne, Collie, Clark, and the rest of his teammates.

  4. Anonymous says:

    FO has done some related work:

  5. Brian Burke says:

    Thanks. It's a misnomer to call that +/-, but it's definitely along the lines of what I was thinking for establishing a baseline catch rate.

  6. Anonymous says:

    You count the Passes that hit are in their area of influence. From the waist to the top of their helmet. Also Passes they could have caught had they laid out or simply leaned forward.

    I know this is nearly a year late but here are my thoughts having played. Not professionally but I think I understand the game well enough to pitch my case.

    Not every Pass is going to be a thing of beauty.

    We roast guys like Smith for throwing high, but praise the Staffords of the NFL because they put the ball where nobody but Megatron can get it.

    We should also count Interceptions against Receivers if the ball can be played and the Receiver doesn't go up to knock it away. As well as tipped balls that had no business of being touched if they can only get one hand on it and knock it up for the Opposing Defense to recover before it hits the ground. I played. I was taught to knock those down and give the QB another chance or allow the ST unit to do what they do. No reason to give the other team a takeaway in that scenario.

    Passes out of bounds count as an uncatchable Pass.

    You really can't count bad Route Running even though it plays a HUGE factor in turnovers. How do you prove it was the fault of the Receiver or the QB. For all we know the QB had a moment that he wished he had that one back so he could throw it to where it needed to go. Also chemistry affects this when the Receiver and the QB aren't on the same page with specific Blitz packs, not to mention Man up/off or Zone coverage because the Receiver or QB haven't had the Reps in practice due to injury. Some things you just cannot account for.

    Look at Braylon Edwards this season. He missed so much time due to the knee and then the shoulder, that he just couldn't get back onto the same page with Alex Smith. Harbaugh's offense may look simple but you cannot miss the Reps for huge chunks of time and expect there to be chemistry between the Receiver and the QB. It's not going to happen. You have to be there or you have to have a bunk area for the players to spend the extra time with the playbook, game plan and film study to compensate.

    Crabtree took 3 years to get on the same page consistently.

  7. Anonymous says:

    These stats u talk about is matter of opinion me and u can watch the same mistake and blame two different people then how do u do the stats

Leave a Reply

Note: Only a member of this blog may post a comment.