Comments on Advanced NFL Stats: Is Adrian Peterson a Liability?

The next step seems to be then examining what Adri...

2010-02-25T13:07:38.933-05:00

The next step seems to be then examining what Adrian Peteron's annual salary is, and comparing his net WPA and EPA with the rest of the league's active running backs. He may still net out out positive gains, but is that gain offset by the salary cap hit he represents for the Vikings team each year?

There are no guarantees in the NFL. A team could ...

2010-02-09T09:40:45.210-05:00

There are no guarantees in the NFL. A team could be great one year and suck the next because of injury, player/coach changes, etc. Peterson's fumbles might be more costly than just a few points or loss. Surely there's some sort of opportunity lost cost. How many more times will his team get into the playoffs?

@Most recent anonymous: Didn't Harris have a r...

2010-02-04T16:49:43.682-05:00

@Most recent anonymous: Didn't Harris have a reputation for avoiding contact? A better example of a runner taking punishment would be Earl Campbell or Walter Payton, I would think.

@Ryan: We're not just talking about lost fumbles, though; we're talking about all fumbles.

Just a thought: It might be worthwhile to look at...

2010-02-04T13:32:00.835-05:00

Just a thought:
It might be worthwhile to look at where each fumble occurred. It seems to me that a great runner like Peterson loses more fumbles because the average runner loses it around the line of scrimmage, where he gets hit the most (and where there are a lot of O-linemen standing around to recover it). Peterson, I'm guessing, loses a lot of fumbles downfield, after breaking into the open and getting hit 10-20 yards downfield, and often in the middle of the field, where I'd imagine there would be a much smaller chance of a recovery by the offense, given how many defenders are around the ballcarrier at that point. Could explain why his fumbles lost % is higher than 50%.

I think one point that is more football and less s...

2010-02-04T13:30:42.460-05:00

I think one point that is more football and less statistics regarding AP's fumbling problem is the tenacity of his running style and his stubbornness to go to the ground. Peterson seems to sometimes relish contact when carrying the ball often getting stood up by two or three defenders before going down. The propensity to fumble after contact must increase. The propensity to fumble after contact with multiple defenders must skyrocket.

He seems to be playing the Franco Harris card, scratching and clawing for every yard but risking putting the ball on the turf, rather than the Emmit Smith card, conservative and efficient running that may not be as explosive but yields fewer fumbles.

Just a thought...

I'd be interested to know where Barry Sanders ...

2010-02-04T08:08:29.302-05:00

I'd be interested to know where Barry Sanders sits in the comparison

Just curious if fumble rates are different for ind...

2010-02-04T04:01:23.465-05:00

Just curious if fumble rates are different for indoor games than they are for outdoor games.

Also, a slight suggestion for your profits/revenue analysis. I would compare EPA to economic profits rather than accounting profits since both take into account expected production. Taking expected production into account makes the value even more sensitive.

Thought about posting this on PFR, but didn't ...

2010-02-04T01:48:36.444-05:00

Thought about posting this on PFR, but didn't want to be a year and a half late to the party. My first thought when I read Eddo saying that fumble rates have declined is that defenders are having to make more tackles in space. Making a solo tackle in space is hard enough without thinking about the strip. As a sam linebacker in high school, I was coached to go for the srip if I was the second defender to meet a ball-carrier in traffic, but in space my responsibility was only to finish the tackle.

Brian, did you ever consider possibly using a Pois...

2010-02-03T20:46:09.258-05:00

Brian, did you ever consider possibly using a Poisson distribution for modeling fumbles? Since fumbles themselves are pretty rare events, a Poisson might be better.

Also, have you considered trying to put together a group of RB's that run ob more than others do? You might have to establish some arbitrary cutoff and it could turn into a multiple endpoints argument but it'd still be interesting to play around with.

Just read Jason's post at PFR. Wow. Fumble rat...

2010-02-03T19:29:20.032-05:00

Just read Jason's post at PFR. Wow. Fumble rates have really come down, far more than I would have imagined. I should probably delete the stuff about W.Payton and F.Harris in the final draft (after I'm properly embarrassed).

Sometimes, but not consistently. It's the same...

2010-02-03T19:16:30.054-05:00

Sometimes, but not consistently. It's the same data you can pull off nfl.com. Most official scorekeepers will say 'pushed ob (M.Tackler),' but not always.

I was saying the same thing yesterday. Running ob rarely results in a fumble, and the few fumbles that do occur are almost never lost. Peterson does have a reputation for taking guys on. He claims guys are afraid of tackling him, so they just go for the strip. I think that's farcical, but it is revealing of how he thinks and how he takes on tacklers.

Brian - does your data allow you to tell when a ru...

2010-02-03T19:09:37.529-05:00

Brian - does your data allow you to tell when a runningback has run out of bounds rather than take the hit?

Merely hypothesizing here (I don't have the data to hand) but is it possible that a RB may appear to have a high fumble rate because they choose to stay in bounds and fight for more yardage when they find themselves outside?

It might well be that Peterson doesn't have poor technique, but instead he goes after the defender versus taking the safe option. Of course, then it doesn't mean he can't improve, just that he'd need to think more about his game.

Thanks, Jason. Maybe the distribution graph above ...

2010-02-03T15:52:54.834-05:00

Thanks, Jason. Maybe the distribution graph above supports your notion. There is an unnatural drop-off in the number of RB's with fumble rates > 1.5%. There's a big survivorship mechanism at work. As soon as a guy starts fumbling more than average, he's on the bench or cut, and no longer in the sample.

Here's the post I wrote a couple of years ago:...

2010-02-03T15:46:43.889-05:00

Here's the post I wrote a couple of years ago:

http://www.pro-football-reference.com/blog/?p=522

To correct Eddo's recollection, I suspected instant replay was a cause, but when we compare the years that replay was introduced, then removed, then re-introduced, we don't see any spikes that would suggest they were a huge factor. Fumble rates for running backs have been consistently declining since the 70's.

My theory was that the increase in passing efficiency has led to decreased fumble rates, because fumbling is less tolerated if passing becomes safer and more efficient. It would take an explosive player like Peterson to be an exception, because if he was mediocre and he fumbled, he would be replaced.

Another example this year is Jamaal Charles. His reputation as a fumbler is what kept him on the bench. Even when Johnson was suspended/released, they initially wanted to give the ball to a more sure handed but way less explosive Kolby Smith. It took several games (and a bad passing offense) for Charles to get his chance and run with it. If he was with Indianapolis instead of KC, I suspect he may not get that chance. You don't take the ball out of Peyton's hands--and we don't care if you can break the occasional 50 yard touchdown.

Maybe he is valuable overall, but to link this art...

2010-02-03T15:36:30.699-05:00

Maybe he is valuable overall, but to link this article with the underdog/favorite one, a top team like the Vikes probably need more a steady, low variance back than a boom and bust back.

Eddo-That's true. I suspected that, but didn&#...

2010-02-03T15:05:47.224-05:00

Eddo-That's true. I suspected that, but didn't have the historical numbers. If anyone has the PFR link, that would be great.

Also, to answer the question above, no these numbers don't have adjustments for opponents.

You can't really compare modern fumble rates t...

2010-02-03T14:56:32.802-05:00

You can't really compare modern fumble rates to those of Payton and Harris (and others from the 80's and earlier). P-F-R had a blog post a few years back on this, but fumble rate now is less than half what it was in 1980. This is likely due to instant replay (until this season, instant replay could only remove fumbles, not add them).

I also think that fumbling now is much worse than fumbling then. As you've pointed out, one main reason for running the ball, besides game theory, is that it's much less variable than passing. Turnovers lead to more variation, so if a goal of the running game is safe, consistent gains, fumbles are an awful thing.

Contrast to 25+ years ago, when the running game was roughly as effective as passing in gaining yardage. Then, you could deal with more fumbles, since you weren't running to be more consistent and safe, but rather to gain yardage.

It seems to me that while AP is worth having, fumb...

2010-02-03T14:43:28.635-05:00

It seems to me that while AP is worth having, fumbles and all, his fumbles cut down his value quite a bit. His overall production is not that of an elite RB, when fumbles are taken into acconut. Keep him, yes, but he's no superstar - as long as he fumbles like he does.

Are these stats normalized in anyway to account fo...

2010-02-03T13:15:02.320-05:00

Are these stats normalized in anyway to account for 'strength of schedule' or basically what defenses teams are going up against?

I meant Brian. I do that all the time. Yes, touch...

2010-02-03T12:26:20.836-05:00

I meant Brian. I do that all the time.

Yes, touches is a little flawed, but probably a lot less flawed that looking at carries alone.

Isn't using Fumbles/Touch a little flawed. I a...

2010-02-03T12:19:57.658-05:00

Isn't using Fumbles/Touch a little flawed. I assume RB's fumble less on receptions so that makes RB's who are main options in the passing game look better (EX: Ray Rice). Shouldn't you adjust for that?

also, I'm not sure why included Michael Westbrook in this sample. He's a WR.