Thursday, April 19, 2007

Evaluating a running back

I wrote about running backs earlier this week in a column in which I argued, through statistics, that LSU's running game as a whole was one of the most effective in the SEC. I got thinking though, how can I statistically evaluate a single running back in a way similar to the way I evaluated quarterbacks, attempting to mesh the entirety of their production into one reasonably meaningful statistic?

I could do it exactly the same way as I did it for quarterbacks, using a formula like this

Runner rating = (yards rushing + 20*touchdowns - 30*fumbles lost)/rushing attempts + (yards receiving + 20*touchdowns -30*fumbles lost)/passes thrown to

I think that result will turn out OK, but I find it unsatisfying for several reasons. First, I'm not sure I can find a statistic on the number of times a ball was thrown to a back. I'd be stuck with pass catches, but that would miss the effect of any dropped passes.

Second, I just don't think it breaks down the information enough. It treats all rushing yards as the same, except for the ones that get you into the end zone. I think a run on 1st and 10 is much different from a run on 3rd and 2. On 3rd and 2, you're just trying to move the chains. On 1st and 10, you're really trying to break a big gain or set up a good down and distance for the rest of the set of downs. It comes down to, on 3rd and 2, a three yard gain is a big success. On 1st and 10, a three yard gain is a let down.

I want to know more than just yards-per-carry. I want to know if the guy broke off a lot of big runs, changing field position. I want to know if he was an effective short-yardage runner too. These statistics just don't do it.

So I think you'd have to break it up by downs, or at least count runs for a 1st down differently, and you'd want to have some way of tracking longer runs. Then again, I wanted to do that for quarterbacks, but again those statistics just aren't kept. I wouldn't want to short-change an excellent short-yardage specialist just because he doesn't have a lot of runs of more than a few yards.

And let's not even begin to discuss how it ignores blocking, which is no small thing. A guy who can't block probably can't be on the field in an obvious passing situation, and his presence on the field may be a tell that it's a running play.

I think the only way to account for the impact of blocking is to drop the yards receiving and use a measurement along the lines of a yards per pass attempt with the player in the game - yards per pass attempt without the player in the game. Good luck compiling that.

Maybe in the end we should not try to develop a single statistic for a running back. Maybe yards per carry, yards per reception, touchdowns, and fumbles should be examined individually. Blocking? Well, just try to forget that there's no good way to statistically measure a blocker.

Looking at our running backs, what stands out? Well, Keiland's 5.7 yard-per carry average is excellent, but Charles Scott's 6.0 yards per carry is eye-popping. I think it's important to realize though that Keiland Williams did it against the core of the SEC schedule, while Scott did it against a weaker early schedule. Hester averaged 4.7 yards per carry, which is very respectable. Justin Vincent and Alley Broussard got plenty of chances, carrying the ball over 130 times between them, but pulled up the rear getting only 3.7 and 3.8 yards per carry respectively. All of our top 4 rushers got between 4 and 6 touchdowns, so they were all basically equally productive on the scoreboard.

Jacob Hester's 35 catches for 7.7 yards per catch and three touchdowns was excellent, and while Justin Vincent didn't catch a lot, he averaged 12.2 yards per catch. Keiland Williams and Charles Scott were basically non-factors in the passing game, catching only 7 balls between them, with no touchdowns.

I think for this coming year, one of the things to watch for is how much Scott gets into the game. It seemed that as the season progressed, the coaches turned to Keiland Williams at the expense of Charles Scott. Early in the season, Scott got the carries, but then as Keiland got in the game, Scott left it. Will he get back in?

Also, look to see if Scott and Williams are used a little more in the passing game. Last year, Hester and Vincent were the passing down backs. Vincent is gone, but Hester remains, and he is very good at that role.

Also, how does Richard Murphy play into this. Will he step into the Justin Vincent role (but hopefully more effective)?

No comments: