Writers of Pro Football Prospectus 2008

19 Sep 2005

Trying to Make Prospects' Numbers Add Up

Here's an interesting story on how the 49ers are using some advanced statistical methods to evaluate college and pro players, something no other NFL team is currently doing. Much like the statistical revolution in baseball, not every GM or head coach believes that the nerd with the pocket protector and a background in mathematics is necessarily the best resource when it comes to X's and O's. But San Francisco's Paraag Marathe is doing his part to change that perception. (Thanks to G. Fischer for the link; bugmenot/free registration required)

Posted by: P. Ryan Wilson on 19 Sep 2005

15 comments, Last at 20 Sep 2005, 10:58pm by Jim A


by Jim A (not verified) :: Tue, 09/20/2005 - 12:44am

For an offensive tackle, the data includes such things as sacks allowed, missed assignments and positive assignments. For a wide receiver: balls caught vs. opportunities, yards after the catch, yards after contact, critical-situation catches and touchdowns vs. receptions. The computer then spits out a rating.

I wonder where Marathe gets these kind of stats on college players, especially considering how many Division I teams there are. I'm not even sure where he gets those offensive linemen stats for NFL players without compiling them himself (obviously the coaching staff can do this easily for the 49ers players, though). Does STATS Inc. assemble stats on offensive line assignments?

by Colin (not verified) :: Tue, 09/20/2005 - 3:35am

I have to guess, and I think it's a pretty good guess, that the actual college coaches who go over their own tape have their own records of this sort of stuff. As part of the good ol' boys network of coaching, I'm sure it wouldn't be too hard for a fellow in his position to convince coaches to give him that information. It may not be perfect, but it's better than the alternative of having to compile all that shit himself. Even if he did have to, though, it doesn't seem unreasonable that he might hire four or five recent graduates who have good football minds but didn't have the skills to play the pro game and have them go over tape 5 days a week for 35 grand a year. I know alot of people who would love that job, especially as a potential springboard into a scouting or player personel job.

by Basilicus (not verified) :: Tue, 09/20/2005 - 4:47am

I'd go for that kind of job. Now an NFL scout will happen upon this website, be impressed by my gumption, and hire me as offensive coordinator of the upstart Portland SeaGrizzlies. The rest will be movie magic. Starring Cuba Gooding, Jr. as the Bearer of SeaGrizzly Light.

by Bill Belichick (not verified) :: Tue, 09/20/2005 - 6:42am

And I always thought it was about watching the guys on film.

by MikeT (not verified) :: Tue, 09/20/2005 - 8:56am

Hey, let me do my Aaron impersonation for a sec...

Lots of teams are currently doing some sort of advanced statistical analysis at the front office level to evaluate players and manage the salary cap. A couple of teams you might have heard of -- the Patriots and Eagles -- are at the forefront of this type of analysis. The Chiefs and Titans are also doing this kind of work.

The Niners are the most open team about using statistical research, and it has made them a lightning rod for people who want to stoke some Scouts vs Statheads controversy.

Of course, the joke is that "stat geeks" sit at their spreadsheets and never watch games. That's like saying that the scouts never look at a player's stat line and never use numbers to prove their points ("He ran a 4.25 forty and can squat 600 pounds!"). Statistical research is a tool, just like scouting.

by MikeT (not verified) :: Tue, 09/20/2005 - 8:58am

Yo Jim:

From what I have read, the Niners hire a team of high school coaches on a per diem basis to come in and watch lots of hours of film and make all of the breakdowns. That's where a lot of the info comes from.

by Jim A (not verified) :: Tue, 09/20/2005 - 9:01am

As part of the good ol’ boys network of coaching, I’m sure it wouldn’t be too hard for a fellow in his position to convince coaches to give him that information.

But I'd guess Marathe is about as far away from the good ol' boys coaching network as a personnel exec can get.

by Jim A (not verified) :: Tue, 09/20/2005 - 9:04am

Thanks Mike. That's what I would guess too. But I bet they still don't review every Division I game.

by mel (not verified) :: Tue, 09/20/2005 - 10:35am

why is the link plain text int the feed??????????
this is the internet you know....

by Terry (not verified) :: Tue, 09/20/2005 - 4:07pm

Marathe refusing an interview request for what appeared to be a sympathetic story is pretty dumb, I think.

by Moses (not verified) :: Tue, 09/20/2005 - 6:46pm

Marathe refusing an interview request for what appeared to be a sympathetic story is pretty dumb, I think.

Well, since you have no idea about how the media has treated Marathe like a piece of **** from the first time he said anything about seeing if moneyball concepts could be translated into an additional measurement tool for evaluating players, and allocating salary cap resources among the positions, perhaps your comments were pretty **** ignorant. The last time he opened up, he was mocked by the Merc, the Chron & CoCoTimes. Then, when he was on the coaching search (and Nolan's biggest fan and prime in-house advocate Nolan be interviewed) he was mocked some more. Marathe has NO REASON to think the media will treat him fairly or accurately.

But, even as "favorable" as this article is, it has its moments of derision as he's being knocked for a model in its infancy and is sometimes throwing out some strange results. Also, Marathe isn't saying replace scouting or film watching. He's trying to find an additional edge.

But I’d guess Marathe is about as far away from the good ol’ boys coaching network as a personnel exec can get.

And you'd be guessing wrong. The 49ers are in NATIONAL (the scouting combine) and have their own in-house regional scouts. Marathe also has two scout R&D assistants and spends time with Nolan, McCloughan and/or Gamble depending on what aspect of his "Moneyball" theory he's working on...


Really, it'll work or it won't. If it works, fine. If not, it's York's money to **** away as he chooses.

by Carl (not verified) :: Tue, 09/20/2005 - 7:18pm

"no other NFL team is currently doing."

I agree with MikeT. I know, for a fact, that the Colts contract out for stat analysis. Apparently, Polian doesn't like firing up SPSS to do his own regressions anymore.

And the irony is that I know, for a fact, exactly how Buffalo scouts use game drive analysis to better understand their opponents. It's actually quite sophisticated and involves a fairly advanced knowledge of geometry and probability theory.

Sometimes the coaches or scouts don't see it that way, but it is nevertheless so.

JimA, you're right about Stats, Inc. numbers for O-line work. They're NOT as good as I think they could be.

One nice thing, however, is that they now include every player on the line for the play.

by Carl (not verified) :: Tue, 09/20/2005 - 7:20pm

"Basically, what Marathe does is gather position-specific data on each player, give each piece of data an A-through-D grade and feed it into a computer along with other pertinent info such as medical history and character."

I love how he constructed that sentence. I have this image of a guy in a lab coat holding a tray of punch cards to "feed" into the blinking Univac.

by Carl (not verified) :: Tue, 09/20/2005 - 7:22pm

"According to a league source familiar with Marathe's rankings, his top-rated computer-produced quarterback was Louisville's Stefan LeFors."

His sources? Marathe ("Hey, I told them to pick LeFors!") and Stefan LeFors ("Make me the top pick!").

by Jim A (not verified) :: Tue, 09/20/2005 - 10:58pm

Moses, I don't doubt that the 49ers do plenty of scouting. I just wouldn't think Marathe, with his background, has many close contacts within the ranks of college or NFL coaches that he can call up personally and request their teams' proprietary stats, as Colin suggested in #2.