ESPN Radio
Facebook TwitterText
Share This: |

What do Advanced Stats Say About CJ Spiller?


There was one redeeming aspect to the 2012 season – a year that was filled with some of the most miserable losses of the Buffalo Bills' 13-year playoff drought.

One player's breakout performance almost made the late-game pick vs. Tennessee, the 500-plus yards allowed to San Francisco, the wrong route New England and the punts from the opponents' 35-yard line all worth watching.

That player, of course, is C.J. Spiller.

His incredible season gave Bills fans a reason to believe that their struggling franchise could actually have a superstar again – something they haven't truly had since Thurman Thomas signed with the Dolphins.

Spiller gained 1,244 yards, caught 43 passes, scored eight touchdowns and averaged an incredible 6.0 yards per carry.

The question now is: Can he get even better in 2013?

First, how good was he?

In terms of advanced stats, he ranked third in Football Outsiders' Defense Adjusted Yards Above Average. And second in Win Probability Added according to

Spiller reached rare air with his 6.0 YPC. In fact, the only running backs since 1990 to average 6.0 per carry or more in a season are: Spiller, Adrian Peterson, Barry Sanders and Jamaal Charles. He became the 52nd running back (with more than 100 carries) since 1990 to average more than 5.0 yards per carry in a single season.

Spiller hit his stride back in 2011 while filling in for an injured Fred Jackson. He finished the season at an outstanding 5.2 yards per carry in 107 carries. By improving on his 5.2 YPC, he became only the sixth running back since '90 to repeat a 5.0+ season.

But the likelihood that he could average 6.0 YPC again is low.

Of the 52 who have averaged more than 5.0 YPC, only two players gained more yards per attempt the next season. On average, those running backs dropped by 1.0 yards per carry the next season.

This chart shows how backs did the season following a year of averaging more than 5.0 YPC:
(Blue is the 5.0+ season, Orange is the next season)

One of the explanations for the average yards per carry taking a dip for 96.1% of running backs who were over 5.0 YPC is that they saw more work the next season. It's a simple case of regression to the mean.

The thing about an NFL season is that every year is a small sample size. In hockey, you wouldn't judge a goalie's save percentage after 16 games. In baseball, pundits would laugh at the notion of judging a hitter by his 16-game batting average. Yet in football, we make assessments on positions based on a very tiny sample. For a player like Spiller – and I know it seems funny to say – but 300 carries doesn't really tell us whether he'll be a superstar or a one-year wonder.

It seems crazy to say, but the truth is that a halfback could face easy defenses and have a little good luck and end up with a ridiculously season and a half. Many forgettable backs have jumped to single-season stardom by averaging more than 5.0 yards per carry. Back in 1996, Napolean Kaufman averaged 5.8 YPC. Thirteen years later, Felix Jones averaged 5.9 yards per carry. Both faded significantly the next season, falling well short of their team's high expectations.

There's a great chance Spiller's yards per carry will drift back toward the mean. Consider that he will probably get 70 percent of the carries this season instead of splitting reps 50/50. That's what usually happens when RB's go for more than 5.0 per attempt. Since 1990, 32 of 52 backs who averaged more than 5 in a single season found themselves taking a higher number of carries the next season. Many of the 20 who didn't played fewer games due to injury the following year, but averaged more carries per game.

More carries, bigger sample, better chance he returns to the mean

Thing is, the small sample size makes it difficult to determine what Spiller's mean actually is. For the league, the average handoff goes 4.3 yards. Spiller's career YPC is 5.4, in 388 carries. Since 1990, only Jamaal Charles averages more than 5.0 with his amazing 5.8 YPC.

Does that mean Spiller will drift back in the 4's? Not exactly. For regression (or progression) to the mean, it's for the individual, not exactly for the league.

For example, if a batter hits .300 in 200 at-bats, but for his 1,000 at-bat career he's been a .220 hitter, it's likely he'll hit a slump eventually and end up closer to .220 than .300 in his next 800 at-bats. But if a hitter is hitting .220 when he's normally a .300 guy, he's likely to end up closer to .300.

What's C.J. Spiller's mean? How can we figure it out?

First, what unsustainable factors might have effected his YPC?

One was a fairly  favorable schedule of poor teams vs. the rush. The Bills faced the 30th, 29th, 28th, 27th and 26th (twice) and 24th worst defenses in 2012. They also saw the Patriots' ninth ranked D twice, Houston and San Francisco.

Things could get a little harder on Spiller and the Bills' O-line, which is now minus Andy Levitre. They'll see last year's No. 1 rush defense in Tampa Bay, the second best team Pittsburgh, 12th ranked Cincinnati and 14th ranked Carolina. Spiller will still be able to face the 32nd New Orleans and 30th Jacksonville. A few harder teams might take a little chunk out of his high YPC.

Another aspect of Spiller's rush average that might be difficult to sustain was the amount of 20+ yard runs. He gained more than 20 yards on a rush once out of every 17.25 carries, which was second best in the NFL only to Peterson. But how repeatable are a high number of 20+ yard rushes? Should we expect one once every 20 runs or will that number drop?

Here's a look at what some of the league's best have done in the last three seasons in terms of 20+ yard runs.


Year Charles Johnson Gore Foster Lynch Peterson Rice McCoy Turner
2012 25.9 34.8 36.9 39 35 12.9 51.4 28.5 44.4
2011 n/a 23.8 25.6 39.7 71.3 29.7 32.3 19.5 27.3
2010 23 24.3 33.9 27.3 40.4 31.4 76.8 29.5 37.1

As you can see, there's some level of consistency in busting big plays for some backs, but wild variation between others. We don't really know which category Spiller will fall into, but his skillset suggests he probably belongs in the former category. Still, one per 17.25 is an extremely difficult rate to repeat. If his 20+ rate dropped to one per 25, it would drop his YPC, but still be amongst the league's best.

And the tape indicates that Spiller can pull off 20+ yard runs with regularity due to his elite-level elusiveness and vertical quickness.

Take a look at this: A screen grab of a game against the Patriots in which eight defenders surrounded Spiller, who was at a dead stop at the time. Somehow, he exploded for 25 more yards. Spiller can see three steps ahead, it seems. He can juke and accelerate from zero to full speed at a Barry Sanders-like pace.


Another factor that causes running backs' averages to dip is how they are used. Some are more successful out of certain formations or are a product of certain situations and systems. Spiller will be presented with a new system this season. However, a closer peek at ESPN's splits shows that the Bills' RB excelled in pretty much every situation.

2 WR: 5.5 YPC

3 WR: 6.9

4 WR: 6.1

Shotgun: 6.0

2 Back: 6.1

Lone: 6.0

Amazingly, Spiller was able to run with great success in every direction as well.

Middle: 6.5

Left: 6.3

Right: 5.0

Left Sideline: 5.1

Right Sideline: 7.1

Here's where we're at so far: The league generally regresses, but Spiller has already been an outlier. He faced fairly easy defenses, but is incredibly elusive, can break 20+ yard runs and can run toward all directions.

OK, what else are we missing?

 Oddly enough, that factor is the Bills playing from behind.

When the Bills were trailing, Spiller averaged 7.4 YPC in 133 carries. When the Bills were ahead, he ran for just 4.2 YPC in 76 attempts. The explanation for this seems pretty cut and dry: Teams were playing the pass. Nearly half (94 of 207) of the former Clemson back's carries came with less than four defenders on the front line. If the Bills are playing from behind again, defenses will spend more time worrying about preventing big pass plays than Spiller's 10 yard carries.

This works both ways. If E.J. Manuel is incredible and Mike Pettine's defense stout, we could see a drop in Spiller's average per carry simply because the Bills will be giving him the ball with the lead – when teams are expecting their opponent to pound away at the clock. Plus, if he's getting the ball more often (and isn't “winded”) the Bills might be ahead more.

What should we expect from Spiller in 2013? Certainly more carries, but hopefully not that many more.

The passing game is a huge part of Spiller's value to the Bills and, in part, what should make him one of the Top 3 running backs in the league. Rush yards are like Batting Average, all-purpose yards is like On-Base Percentage. It doesn't matter how Spiller gets successful plays, just that he has them. In fact, 48% of the time he touched the ball last season, it was considered by AdvancedNFLStats as a “successful play.” That was one of the league's highest.

Last year he went 43 for 56 in catches/targets. Spiller was used brilliantly at times by Chan Gailey as a pass catcher and has the rare ability to go deep or even line up in the slot. Nate Hackett, take note.

The case for 2,000 yards on the ground was supported by earlier this week. But if Spiller is catching passes at a similar rate and gaining around 10 yards per catch, it seems it's an extreme long shot he'd rush enough to get to 2k. He has 388 carries in three seasons for a total of 2,088. The Bills shouldn't be handing off to Spiller that many times.

Hackett, Doug Marrone and Spiller himself should aim for an all-purpose yards goal rather than just rush. Last season, he gained 1,703 yards in 250 touches.

A little Twitter poll conducted here indicates that would be right on par with fan expectations.

Most Bills fans don't expect 6.0 YPC, but they expect to see more Spiller. And the odds are that if Spiller jumps from 207 carries to around 300 carries and is targetting 50 times in the pass game, he will very likely average fewer yards per carry but make a bigger overall impact.

So sis regression to the mean is likely to result in an average of 4.5 to 5.0 and with more carries, but a reasonable projection for Spiller could be 2,000 all-purpose yards, a Pro Bowl and the official rise of Buffalo's first superstar in a long time.

Call it a case of regression to the awesome