|ESPN.com: Draft Kit||[Print without images]|
Editor's Note: This article was originally published in July 2010. We are bringing it back in archive form for your convenience.
Psst. Hey, football. Guess what?
The numbers geeks are coming for you.
We've seen this movie play out in baseball. As Bill James and his inheritors introduced more and more compelling metrics to decipher what makes a successful player and a successful team, the old guard -- consisting of ex-players, scouts and traditional media members frankly too lazy to learn anything new -- cajoled and mocked, fretted that the geeks were taking over their sport, made big productions about how you wouldn't understand if you haven't played the game. These boosters of concepts, like "chemistry" and "momentum," were mostly who you heard when you turned on a ball game or talk radio, because they were (and mostly are) entrenched in major media outlets. But like a river gradually digging a canyon through rock, "newfangled" data like OBP, OPS, BABIP, xFIP and dozens of other useful tools have gained traction, to the point now that most any sports-talk jock goes out of his way to mock, for example, wins as a true measure of a pitcher's worth.
And by the way, let's not downgrade fantasy baseball's role in the shifting of this dialogue. Sure, Michael Lewis's "Moneyball" (published in 2003) was a big step in opening eyes toward nontraditional metrics, but most successful fantasy players through the 1980s and '90s were eminently familiar with sabremetrics as a method to finding value. In fact, this unholy alliance (and overlap) between sabremetricians and fantasy players probably increased resistance to thinking about player evaluation in new ways. Hence you got smug on-air broadcasters saying, in short, that they don't like fantasy sports, that in some cases they don't understand them, and that fantasy players are too nerdy to listen to. And that overlaps with an ESPN guy like Bill Simmons, who admits he spent a few years basically ignoring baseball completely because he thought it had grown too complicated (i.e., the stats geeks were winning). But just look at ESPN to see how things have changed. Now we have fantasy segments on "Baseball Tonight." Simmons ran a mea culpa column about sabremetrics this spring. We run stats across the Bottom Line on your screen. We put people like me on TV. The revolution was, in fact, televised.
Now comes football. If you've been playing fantasy football for more than a few seasons, you remember a time where the term "targets" was considered weird, geeky and exotic. When I started writing about football for a living, I had a relatively difficult time tracking down targets ... and this was just five years or so ago. Now, of course, targets are a required part of any basic statistical evaluation of pass catchers. Obviously, that's just the tip of the iceberg.
Far be it from me to proclaim myself at the vanguard of sophisticated statistical evaluation in football. I'm not. Nowhere close to it. But I really like the guys who are. Guys such as K.C. Joyner and Aaron Schatz and Sean Lahman and Bill Barnwell and Will Carroll and dozens of others do the hard work of creating and evaluating metrics that help us better understand the NFL. And when we understand how to better evaluate players, as models grow more predictive, our fantasy teams get smarter. The resistance is, as it was with baseball, firm. Terms like DVOA and DYAR still feel a little like Greek, and frankly may not be fully formed. But I love reading Pro Football Prospectus anyway, and you should, too.
Now, that said, I think there's a case to be made that, for a few reasons, statistical analysis in football may never quite get to the point that it's gotten in baseball. First, the nature of the sports is awfully different: baseball focuses on the one-on-one battle of pitcher versus batter, a confrontation that gets replicated hundreds of times in every game, whereas football is the ultimate contingent sport, with dozens of moving, difficult-to-analyze-in-a-vacuum parts on every play. Also, careers simply don't last as long in football, and injuries are prima facie more common, meaning it's difficult to get a long-term grasp on every (or even most) players' abilities. It's difficult to say that numbers in football can give us the same predictive feel that numbers in baseball do. By way of example, I'm looking at the 2008 version of Pro Football Prospectus, which listed as its top three prospects Jerious Norwood, Jason Hatcher and Brandon McDonald. Oops. (Of course, that book gets tons of stuff right, too.)
My job in this column isn't to regurgitate the good work that sites such as FootballOutsiders.com, ProFootballFocus.com and others do, and you should read them for smart, sophisticated analysis of what they chart on a play-by-play basis. Instead, I want to lobby for metrics that don't necessarily come from charting every play, and are often available to anyone with an Internet connection and an Excel spreadsheet. And I also hope to help push the boulder uphill, so that some of these statistics become more prevalent everywhere. Here, then, are nine football stats we can all use to evaluate players at a deeper level, to help us discover value in the NFL and fantasy football.
While by no means a pure metric for a running back's value, YACo gives us a sense of a back's elusiveness, toughness and hunger. Surely, we're less interested in running backs who get touched in the backfield and then dance around for three yards and get pummeled, and more interested in the ones who run over a strong safety and keep on trucking. And surely, the RB who plays behind an elite offensive line tends to feel contact much further down the field than the RB who has to deal with opposing linebackers in his face on every carry. But make no mistake: no single kind of runner dominates YACo, i.e., it's not just a "sprinter" or "plugger" category. It's also important to note that YACo is a subjective stat; in doing research for this story, I found large discrepancies among sources, and the reason is obvious: different trackers have different definitions of "contact." Because their site is free to the public, I'm going to use ProFootballFocus's numbers. Here, then, are PFF.com's YACo leaders from the past two seasons.
Was Chris Johnson's breakout completely unforeseeable? Of course not (and not only because he was so good after contact in '08). But certainly I think we can call the inclusions of Rashard Mendenhall and Fred Jackson on last season's list eye opening, considering they finished 14th and 15th in total rushing yards, respectively. For the purposes of looking for potential gems for '10 and beyond, it might actually be more useful to examine how players fared after contact on a per-attempt basis. Here were the top men in that category the past two years (among RBs who played at least 25 percent of their teams' snaps).
In this light, Jamaal Charles' breakout second half is all the more impressive (especially because his offensive line isn't great and hasn't gotten substantially better), and guys such as Felix Jones, Justin Forsett and Beanie Wells -- all of whom figure to get cracks at larger roles in '10 -- may be in for big performance boosts.
Once you've got YACo, this one's easy to figure: YACo% is a running back's Yards After Contact divided by his total rushing yards. Again, this is no panacea for a running back's overall value, but it delivers information across two dimensions: how elusive a rusher is, and how much room he's typically given by his offensive line. Here are the leaders and the laggards in YACo% from '09, among RBs with at least 100 carries.
Some of these results aren't shocking. Thomas Jones ran behind a really good Jets line last year, and probably had more instances when he wasn't touched near the line of scrimmage than almost any back in the league. A few things here are instructive, though. Clinton Portis's days as an elusive runner appear to be over. Marshawn Lynch didn't make many folks miss, either. In the best of all worlds I think you'd like to see Frank Gore do more after contact, because he's not an elite-speed guy. Maybe Correll Buckhalter is a little less of a threat to Knowshon Moreno again this year than I've been worried about (he's not a burner, so if he's going down that easily, Buckhalter doesn't figure to produce).
You can't use this set of "after contact" lists as absolute indicators of success or failure, certainly, but I'll tell you which backs got my positive attention after sifting through past results: Justin Forsett, Fred Jackson and Jonathan Stewart. I'm a bit more impressed by them now.
Raise your hand if you're sick of the term "red zone." Sure, it served its purpose; about 30 years ago, the media picked up on the notion that NFL teams treated personnel and strategy differently when the ball was inside the 20-yard line. (I wasn't writing about football at that point, but I wonder if crotchety broadcasters and writers regularly gave the whole, "In my day, we didn't talk about red zones. All we cared about was the end zone! These kids and their newfangled metrics!" speech) For fantasy purposes, though, we acknowledge that knowing how frequently a running back carries the ball inside the 20 is vaguely useful, but it's much more important to know whether he gets it in goal-to-go scenarios. Here's a look at the top-10 ball carriers inside an opponent's 10 in each of the past three seasons (this data comes from STATS Inc.; unfortunately, I haven't found these stats regularly updated and free on the web, but if you have, please post a comment and let everyone know).And here's a look at those same seasons, and the top 10 ball carriers from inside an opponent's 5-yard line.
As I noted in my story about what to do with the No. 1 overall pick, Adrian Peterson's goal-line production last year was probably unsustainable: He had 14 touchdowns from inside the 5, a number that's likely to drop to a more reasonable eight-ish this season; it's fascinating to see that AP didn't register as an elite short-yardage weapon in either of his previous two seasons (when he had nine and three carries inside the 5, respectively). It's nice to see that Brandon Jacobs keeps getting cracks at short touchdowns despite his litany of injuries. It's interesting that Knowshon Moreno gets used so frequently inside the 10, but less frequently (14 times to be exact) inside the 5. Marion Barber's boosters might be chagrined to know how many short cracks he got in '09 (he converted six for scores), seeing as how much of his value this year figures to come from resuming a "closer" role. Boy that Bears offensive line must've been really bad last year, because it's not like Matt Forte didn't get a bunch of chances. And only two players appear on all six lists: LaDainian Tomlinson and Thomas Jones. I don't foresee big years for either, but it would be a mistake to completely rule out any fantasy value.
For judging a wide receiver's playmaking ability and explosiveness, YPC (yards per catch) is out, and YPT (yards per target) is in. Do I particularly care that Robert Meachem averaged 16.0 yards per catch last year? Well, sure, it's nice: it means Meachem finished 15th among qualified leaders (wideouts who caught at least 30 passes). But it's more revealing that Meachem produced his 722 receiving yards on 63 targets, giving him 11.5 YPT, which placed him No. 1 in the league (again, among WRs who caught at least 30 balls). YPT is better than YPC because it factors in unsuccessful plays, as well as successful ones. Here are the qualified YPT leaders and laggards from 2009.
I'm concerned about how the Saints spread the football, but if you can convince yourself Meachem is ready to take a step into permanent starterdom, he's a great sleeper. Remember, though, that he had the fewest receiving yards of all qualifiers in '09 (tied with Torry Holt). Maybe Hakeem Nicks is the Giants' answer to Plaxico Burress after all. With Vincent Jackson out of action for at least three games and possibly a lot more, Malcom Floyd doesn't look like a poor late-round flyer. And how about Jason Avant? I find it hard to believe he'll have much fantasy value as Philly's third receiver, but my mind just got opened a bit.
Now, YPT isn't a holy grail, because it's reliant on a wideout's quarterback and his offensive line. But to an extent, it captures whether a WR has a tendency toward running bad routes or dropping the ball, something YPC doesn't do. And YPT certainly gives emphasis to vertical threats while diminishing possession receivers, probably to an unfair degree. By the same token, big yards and big plays are what we prefer out of our fantasy wideouts, aren't they?
Another metric to establish who's a downfield threat -- and thus likely to give you those huge days by a wide receiver that can win you fantasy games by themselves -- is the extent to which a receiver is actually a deep target, versus whether he caught several shorties and ran them a long distance himself. Of course, yards are good for fantasy receivers any way you get them, but it's safe to argue that if a player is consistently getting deep downfield looks, he's likelier to keep up big-play production game to game, and year to year. To figure AY%, subtract a receiver's Yards After Catch from his total receiving yardage, and divide by those total yards. Here were last year's best and worst (among qualifiers).
Once again, we like Malcom Floyd's upside as a sleeper. Mike Wallace is a burner who looks like a game-breaker, especially when Ben Roethlisberger comes back from his suspension. I put Torry Holt in a different category: a guy who can't run anymore, so he pretty much just jogs out of bounds after the catch. Meanwhile, so much for our hopes about Hakeem Nicks. It looks like an awful lot of his yardage was the catch-and-run kind, which again isn't inherently bad, but it doesn't seem as though the Giants made him a must-bomb kind of guy in his rookie campaign.
CPT removes yardage from the equation, and begins to zero in on the effectiveness of a passer/receiver combination. We'll get to a more QB-independent metric in a moment, but provided a team's signal-caller hasn't changed from year to year, it's illustrative to look at how effectively a receiver is able to get on the same page with his thrower, whether his thrower tends to be accurate, and whether the wideout's hands are strong. Here were the qualified (more than 30 catches) CPT winners and losers last season.
You'd expect the guys who are low in Yards Per Target to be high in Catch Per Target (since the ball doesn't travel far in the air), and guys such as Wes Welker and Davone Bess comply. But again we see Robert Meachem as a sure-handed player despite his downfield ways, and it's heartening to see a young vertical threat such as Sidney Rice present good chemistry with his quarterback and good hands that compare with those of wily veteran Hines Ward. Meanwhile, some receivers with notoriously bad hands (Roy Williams and Braylon Edwards come to mind) are on the "bad" list, as are several receivers with unsettled quarterback situations. Speaking of which, Calvin Johnson, everyone's favorite Megatron, had the lowest CPT (exactly 50 percent) of any wideout who caught 50 passes or more last season. But was it him, or was it Matthew Stafford's growing pains?
A more quarterback-independent measure of a receiver's hands, Catch% takes targets entirely out of the equation. Instead, we look at (catches / (catches + drops)). Of course, as with our discussion of YACo for running backs, Catch% involves a subjective judgment, because one statistician's "WR Drop" might be another's "QB Overthrow." For the purpose of this analysis, I looked at Stats LLC's numbers, which unfortunately aren't available free to the general public. (Once again, if you know of a reliable and free site that keeps its drops statistic current, post a comment so folks are aware.) Let's have a look at last season's qualified Catch% winners and losers.
The receivers that recur from the poor side of the CPT list are Bryant Johnson, Roy Williams, Louis Murphy and Chansi Stuckey. Blame them more than their quarterbacks. And once again, our friends Robert Meachem and Mike Wallace flashed good hands in addition to good chemistry. It's not hard to understand why many of us are awfully skeptical of Dwayne Bowe turning in the monster season that his talent indicates, not when he drops the ball so much. And gee, wonder why Terrell Owens can't find work?
Remember my spiel about "red zone" carries when it comes to rushers? The same is true for pass catchers, and here I won't limit myself just to wide receivers, we'll consider tight ends and pass-catching running backs, too. Certainly, getting a target inside the 20 is nice. But it's far likelier a receiver who catches a pass in a goal-to-go situation will get into the end zone. So as we did with running back carries, let's look at the past three years for the leaders in targets inside the 10.
(Note that four others also were tied with 10 such targets last year: Sidney Rice, Roy Williams, Tony Gonzalez and T.J. Houshmandzadeh.) The only man to appear on all three such lists is Randy Moss. And it's pretty shocking to note that Greg Olsen got so much 10-and-in love last season.
And now let's also look at who leads on targets from inside the 5.(10 other players tied with five targets inside the 5 last season.) Moss isn't as much of a weapon inside the 5 as inside the 10, whereas Reggie Wayne gets most of his close-in looks from really close in. It's also quite interesting to note the absence of certain players. Where are Andre Johnson, Steve Smith (CAR), Greg Jennings and Chad Ochocinco? Clearly, this is a facet of their game you can't rely on. If these guys aren't scoring from farther out, they probably aren't scoring much. Of course, they do score from farther out, so it's not a reason not to draft them. But if you want a sense of why Reggie Wayne tends to be week-to-week reliable, here's part of it.
Know how in-game football announcers always moan about quarterback rating? They're right. We can do better. First off, the number (100.7! 82.3!) doesn't inherently mean anything, which is silly. Also, it reflects a different time in football: it weighs yards, interceptions and touchdowns in ratios that simply don't reflect the modern pass-happy game. When Roger Staubach retired in 1979, he had the NFL's highest career rating: 83.4. Thirteen different quarterbacks beat that mark last year. ANYPA makes intuitive sense when you look at it, plus it has the benefit of passing the smell test once you start comparing quarterbacks from this and previous eras. Here's the currently agreed-upon formula for the stat, which you'll find for free on pro-football-reference.com.
ANYPA = (pass yards - sack yards - 45*INTs + 20*TDs) / (attempts + sacks)
Throw out your old-fashioned QB rating! Here are the quarterbacks who've led in ANYPA the past three seasons.
This isn't a perfect stat, especially for fantasy, because it doesn't include running yards by a quarterback (hence Aaron Rodgers doesn't show up big either of the past two seasons). But it does do a fairly good job weighing quarterbacks' aerial abilities, it makes sense (on average, when Drew Brees threw the ball last season, it resulted in 8.31 yards' worth of positives for the Saints) and allows different QBs from different eras to be compared to one another more fairly.
Of course there are way more metrics bouncing around out there, many more complicated, more elegant and more comprehensive than the more basic ones I've tried to detail here. Stuff like Win Probability Added, Success Rate, Defense-Adjusted Value Over Average (DVOA), Defense-Adjusted Yards Above Replacement (DYAR) and much more. I love reading about all of 'em. You don't necessarily get those "ah-ha!" moments you do looking at sabremetric baseball stats like BABIP or xFIP, but you can start building cases for player value, which is what success (both in the NFL and in fantasy) is all about. And so the boulder slowly rolls uphill.
Christopher Harris is a fantasy analyst for ESPN.com. He is a six-time Fantasy Sports Writing Association award winner. You can ask him questions at www.facebook.com/writerboy.