Stat Analysis
Advanced analytics on player and team performance

Introducing YAC+, Part II

Introducing YAC+, Part II
Photo: USA Today Sports Images

by Bill Barnwell

For those of you who haven't seen it, please read last week's introduction of YAC+ before getting to this article.

In this follow-up, I want to take a look at what YAC+ says about the player pool as a whole. Is YAC+ a reliable, consistent statistic? Is it subject to massive year-to-year swings? I'll identify a few players and what their stats say about these relationships along the way.

Let's start with the year-to-year consistency of YAC+. Anecdotally, it would make sense that the players we introduced as having elite YAC skills in our last article -- Darren Sproles and Miles Austin -- would have those sort of skills from year-to-year. We noted that Sproles would have led the league in YAC+ per catch in 2008 if he'd qualified with one more catch -- but, on the other hand, he was below-average in 2007. Austin only had a combined 18 catches in 2007-08, which makes that data pretty useless, but he was below-average in 2007 and above-average in 2008.

For a league-wide check, we measured the year-to-year correlation of YAC+ per catch for players with 30 or more catches in consecutive seasons. I also split the groups out by position, to see if one particular group was more or less consistent than another. Obviously, the fact that there are only four years of data to draw upon limits the predicative value of these stats, but there are still some really interesting things to discuss here. The table below reports those year-to-year correlations, with the sample sizes in parentheses.

Table 1: Year-to-Year YAC+ Per Catch Correlation
  2006-07 2007-08 2008-09
All Receivers 0.44 (85) 0.17 (84) 0.38 (90)
RB -0.03 (11) 0.00 (11) -0.38 (14)
WR 0.38 (54) 0.17 (55) 0.36 (55)
TE 0.23 (17) -0.41 (18) 0.32 (21)

Well, that's not what I was expecting. The year-to-year YAC+ correlation ranges from fairly weak (2006-07, 2008-09) to just about useless (2007-08). Sadly, we're about five years away from figuring out whether 2007-08 was the outlier, but what a strange dip in reliability!

During that season, tight end YAC+ totally flipped, which is worth exploring. The 2007 leader for YAC+ per catch was Donald Lee (1.2); Lee promptly fell off to -2.2 YAC+ per catch in 2008, and after being replaced by Jermichael Finley, was at -1.9 in 2009. Just behind him was Desmond Clark (1.0), who also fell below-average a year later (-0.2). On the other side of things, David Martin had a dismal -2.2 YAC+ per catch in 2007, and a year later, he was almost exactly average (0.02). Vernon Davis was the next-worst, at -1.1 YAC+ per catch, and he improved to 1.1 YAC+ in 2008, although he dropped back down to -0.5 YAC+ per catch in 2009.

The consistency of wide receiver YAC+ also fell off in a much larger sample, though, and there are a few players to note there. The league leader in YAC+ in 2007 was Greg Jennings, who put up a very impressive 3.3 YAC+ per catch. Jennings' YAC+ progression over the past four seasons: 3.0, 3.3, -0.1, 2.4. Perhaps Packers fans can help us identify a specific factor that led to the YAC decline in 2008. Obviously, Green Bay changed quarterbacks that year, but did it really affect Jennings differently from other players? As for injuries, Jennings suffered a rib injury and a concussion in 2008, but he had ankle and groin injuries in 2007 that would theoretically cost him more in terms of running after a catch.

Jennings wasn't the only one who saw a swift downturn. There were four other qualifying receivers who had YAC+ per catch above 1.0 in 2007. One -- Patrick Crayton -- retained his YAC+ rate, going from 1.6 in 2007 to 1.5 in 2008. The other three all fell way off: Roddy White (1.3 YAC+ per catch in 2007 to -0.2 in 2008), Reggie Williams (1.9 YAC+ in 2007, -1.8 in 2008), and Anthony Gonzalez (1.6 YAC+ in 2007 to -0.7 YAC+ in 2008).

The trend wasn't as pronounced on the other side of the coin. 16 receivers had a YAC+ per catch below -1.0 that year, paced by Bryant Johnson, who had a YAC+ of -1.9 in 2007. Only five of these receivers produced a positive YAC+ figure in 2008.

Contained in that group of 16 players is a familiar lightning rod for our criticism. Chris Chambers has been, arguably, the most consistently below-average producer of YAC of any receiver in football. His YAC+ figures from 2006 to 2009 go -1.3, -1.7, -2.2, and -0.3. Remember -- this isn't attempt-based, so it doesn't account for Chambers' poor catch rate. It's strictly based on what he does after he's caught the ball.

On the other hand, consider a player with a style similar to Chambers, Vincent Jackson. Obviously, Jackson is a much better player, but his YAC+ figures have also been disappointing: -1.2 in 2007, -1.0 in 2008, and -1.3 a year ago. This led me to wonder whether YAC+ is too harsh on players who catch jump balls, such as Chambers and Jackson. Then I considered a third sort of player who gets those kind of passes, Calvin Johnson. He's had a positive YAC+ in 2008 (1.8) and 2009 (0.7). So the answer is: I don't know what it all means.

Although it's a very small sample size in each season, running backs have exhibited virtually no year-to-year YAC+ consistency. In 2006, the YAC+ leader among backs was Larry Johnson (3.6 YAC+ per catch); a year later, he was exactly league-average, and he hasn't qualified for the leaderboard since. Behind him was Mewelde Moore (3.1), who didn't qualify again until 2008, when he was at 0.2 as a member of the Steelers. Maurice Jones-Drew (2.5) was in fourth in 2006, and he's one of the few backs who has been relatively consistent in YAC+, with a figure of 2.8 in 2007, 2.0 in 2008, and then a slightly above-average total of 0.3 in 2009. That 2.8 total in 2007 was tied for the league lead with Ronnie Brown, who promptly fell off to -0.7 YAC+ in 2008.

What happened to the leaders of 2008 in 2009, though? Well, several players with impressive YAC+ figures didn't qualify a year later, including J.J. Arrington, Darren McFadden, Jerious Norwood, and Derrick Ward. Steven Jackson, who was in second place in YAC+ in 2008, fell slightly below-average in 2009. Chris Johnson, who was tied for the worst YAC+ among running backs in 2008 (-0.7, alongside Brown), had the second-best YAC+ amongst all players in 2009. I think the small sample size certainly complicates things, but I don't think it's possible to make a case that gaining above-average YAC is a consistent skill for running backs.

Other interesting notes from the past four years:

  • Reggie Wayne has yet to post an above-average YAC+, with figures of -1.5, -0.4, -0.5, and -0.1, respectively, from 2006-09. However, this could show one of the weaknesses of YAC+; although each pass is compared to passes of similar length and location, these aren't necessarily similar routes. A 15-yard go down the sideline looks the same as a 15-yard curl on the sideline, and as we all know, the curl is Reggie Wayne's specialty.
  • What about the rest of the Colts? Dallas Clark has posted slightly positive YAC+ figures in each of the past two years, but the only other receiver on the Colts from 2006-09 who put up a well-above average YAC+ per catch figure was Pierre Garcon last year (1.8). Remember that YAC+ is, as of now, not adjusted for team.
  • Santonio Holmes posted a YAC+ of 3.0 in 2006, but has been just slightly above-average since; meanwhile, Antwaan Randle El (the player Holmes replaced) put up a league-worst -1.9 YAC+ that year. He was between -0.2 and -0.4 YAC+ in 2007, 2008, and 2009.
  • Hakeem Nicks led all wideouts with 3.6 YAC+ per catch a year ago. The only wide receiver with a better figure over the four-year span was Devery Henderson in 2008, who put up 4.1 YAC+. Henderson regressed to 1.4 YAC+ in 2009, and was actually below-average in 2007, with a YAC+ of -0.7. I would suspect that Nicks's YAC+ and raw YAC figures will also decline in 2010.

We'll finish off our series on YAC+ later this week with a look at quarterback YAC+.


16 comments, Last at 19 Aug 2010, 1:55pm

1 Re: Introducing YAC+, Part II

RE: Jennings

The past two seasons the Packers didn't rely so heavily on the quick slant that they thrived on before. Also Jennings might have been playing deep-threat decoy for the other guys.

Furthermore, the Packers as a whole also dropped far in YAC. Drops are also up.

-packer fan

2 Re: Introducing YAC+, Part II

I don't think YAC+ is being "harsh" on jump ball receivers. It is simply describing them. It matches observation quite nicely, that jump ball receivers go down field jump up and catch the ball with a defender draped all over them, and then get tackled.

9 Re: Introducing YAC+, Part II

But it implies that they are less skilful at running after the catch than a player who runs a route of similar depth, but catches the ball behind the D and only has to stay upright to get good YAC as he runs it in for a TD. You are right, to an extent it is just describing them rather than judging them.

I suspect the standard deviation on YAC on long catches is quite high, as a lot will be jump balls and a lot will be long runs into an open field, with not as many in between. But then I don't understand standard deviation at all, so I may be completely wrong.

3 Re: Introducing YAC+, Part II

Obviously, Green Bay changed quarterbacks that year, but did it really effect [sic] Jennings different [sic] from other players?

I would guess that many things like that can effect change in YAC+.

The table below reports those year-to-year correlations, with the sample sizes in parentheses.

I assumed that the top row, "All receivers", shows the correlation from one year to the next in your YAC+ statistic, and the other three rows were subsets of "All receivers" but showing correlations similarly. However, the first and third columns have correlations for "All receivers" that are larger than any of the subsets. I think either your description of the table is wrong or you haven't successfully reported your results.

5 Re: Introducing YAC+, Part II

Maybe, but it's a descriptive stat, not a predictive one. It tells you how a receiver played in the past year, but it doesn't predict how they'll do in the future.

6 Re: Introducing YAC+, Part II

I wonder if supposed YAC+ studs like Miles Austin are consistent from game to game, of whether they put up big enough numbers in three games to make their entire season look good...

"Just look at that pumpkin."
-John Madden, looking at the moon.

8 Re: Introducing YAC+, Part II

The problem ist, that the statistic is not able to divide the data points by different causes.

The problem is always, that it is impossible to put value to the data points that did not occur. (How many catches would he have had, on the plays where the QB decided not to throw too him, although he was partially open? How many YAC would he have had on his drops in traffic) The elite receiver have these Negative data points also in their sample, because they are goto guys, and because the QB knows he is still getting a positive expected return out of them.

Therefore these stats should paint the good but not great guys as the league leaders, and this is a very unstable group from year to year.

10 Re: Introducing YAC+, Part II

Same here. This is definitely a metric right in its infancy. But I can see different things being added to it, such as the Ds average YAC+ to factor out people running for good YAC against terrible Ds, or possibly the average YAC+ compared to expected on the route that a player runs (a curl having lower YAC+ on a 10 yard pass than a 10 yard dig) which game charter could look for in the future.

I think its quite useful to show that players who are thought of as monsters after the catch might not actually be.

12 Re: Introducing YAC+, Part II

I kinda agree that these WR stats being rolled out are pretty uninteresting. I have a feeling they're going to go quitely away with ALY and other least beloved FO metrics that don't really seem to be measuring what they're supposed to.

It just seems like most of them don't say anything coherent or interesting or surprising that DVOA doesn't say. They require too much of a combination with on-field observation to be useful - or maybe, subjectively analyzing the game as you watch it is just as useful as these stats.

14 Re: Introducing YAC+, Part II

Even if all we end up learning from this stuff is that catch rate and/or YAC isn't consistent enough to be considered a skill (and I'm not claiming that's going to be the case), that's knowledge we didn't have.

11 Re: Introducing YAC+, Part II

Does YAC+ have any type of relationship with +/-? Wayne sticks out as being very good by +/- and very bad by YAC+.

16 Re: Introducing YAC+, Part II

well there are factors by merely watching the game that are pretty obvious, and this really is not that confusing as it is being made out to be. First off, i don't like the chris chambers/vincent jackson comparison at all even when chris chambers was younger, being that chamber is 5'10, even tho he plays bigger, VJax is pushing 6"5. Now the difference between VJax, Chambers and other "jump ball" wr's like Calvin Johnson and say a younger Randy Moss is this; elite speed. VJax and a young chris chambers have good speed; Megatron, Randy, Andre 3000 have elite speed, so those guys will get behind defenders and get more yards after catch vice, just running downfield and jumping up for a ball and coming down with it. They can also take short screen passes and bust them open, even though they excel as "jump ball" wr's, they have other skill sets too. Just watch the game its fairly easy to see this, and it's not oversimplifying it's actually glaring obvious when you watch these guys play.