Writers of Pro Football Prospectus 2008

30 Aug 2006

FO on BSMW: The Selection of Laurence Maroney

This week's Football Outsiders feature on the Boston Sports Media Watch site takes a look at the selection of Laurence Maroney: how it represents a sea change from previous Belichick drafts, why it may have been a necessary risk, and its potential for success. Sadly, Ken Walter's streak of appearances in FO on BSMW columns ends this week at ... 2.

Posted by: Bill Barnwell on 30 Aug 2006

54 comments, Last at 31 Aug 2006, 3:46pm by Bill Barnwell

Comments

1
by Vern (not verified) :: Wed, 08/30/2006 - 1:18am

Is there ANY correlation between these college "production" stats and NFL success? If so, where is that shown? NFL success has been shown to correlate with DVOA. Not so for college-to-NFL transitions that I am aware of.

Don't you first have to show what stats correlate strongly to success and/or to failure before drawing comparisons based on them?

Otherwise, this is like saying "guys who's names have 4 vowels tend to dominate" or the proveribial "so and so is always better on Tuesdays after a rain delay."

2
by Purds (not verified) :: Wed, 08/30/2006 - 1:33am

Vern:

I hear what you're saying, but the article is a little more thoughtful than you portray. Nine of the 10 guys on the 550+ carries list were pretty darn bad. Here's the list with 2005 DVOA and then yards

Dayne 25.3%, 270 yds
Irvin: 0
Thomas: -28.5%, 80 yds
A. Davis: ?
Betts: -14.8% 78 yds.
Perry: 5.7%, 279 yds
Duckett: -14.7%, 388 yds
M. Barber: -11.6%, 115 yds
Enis: 0 (but 16% rookie year)

In addition, FO has repeatedly noted the NFL RB problem for guys who get over a certain number of carries (is it 300 or 350)?

That said, Maroney has a chance to be good, especially if he shares the load for a year or two.

As a Colt fan, I was sad to see Edge go, but the carries he had were just so high. These guys are ultimately only human.

3
by Purds (not verified) :: Wed, 08/30/2006 - 1:35am

Sorry.

The note on Enis should have been -16% his rookie year.

4
by Terry (not verified) :: Wed, 08/30/2006 - 1:51am

Maybe I'll contribute something a little more intelligent later, but I'm pressed for time currently, so let me just add:

I cannot believe Ron Dayne averaged six yards a carry.

5
by Andrew (not verified) :: Wed, 08/30/2006 - 1:54am

I think this article would be much more convincing had the author compared Maroney's stats to running backs who are considered successes in the NFL. Ideally, you'd have a "negative control" (unsuccessful backs) and a "positve control", (successful backs) then look at both and see which of the two the "experimental" (Maroney) compares best with.

Right now, all he's given us is a bunch of negative stats, and all I'm convinced of is that drafting Big 10 running backs is a bad idea.

6
by JKL (not verified) :: Wed, 08/30/2006 - 8:29am

I like most of the stuff on FO, but I found this fairly worthless. I guess I am not convinced that what TJ Duckett did or did not do in college has a whole lot to do with Maroney.

Right now, all he’s given us is a bunch of negative stats, and all I’m convinced of is that drafting Big 10 running backs is a bad idea.

And this is exactly what was said about Larry Johnson. Throw in the Penn State thing to boot. On my list of scouting a running back, the name of the conference he played in is down on my list of reasons to take a player. Maybe teams should be taking more players from the current Mountain West schools, because LaDainian Tomlinson, from a current MWC school, is pretty dang good.

The only valid basis for comparison is Marion Barber III, since they actually played on the same team together. In my opinion, Maroney was the superior back to Barber in college, and the scouts tended to agree. If Barber had a decent rookie season, I think that forebodes well for Maroney.

7
by Ima Pseudonym (not verified) :: Wed, 08/30/2006 - 10:10am

diluting the brand?

Don't want to be harsh about it, but I haven't found any of barnwell's pieces interesting yet (and I'm a long time FO reader and a pats homer). They seem to me to only scratch the surface of whatever topic they are covering. I feel a little uneasy that some people will get their first exposure to FO from these articles.

Maybe fewer columns that are more thoroughly researched would be better.

8
by karl (not verified) :: Wed, 08/30/2006 - 10:22am

Gonna have to agree with #5 on that one. You shoulda done something comparing these guys to the 3 UM running backs of recent success - James, McGahee, Portis.

THE U.

9
by B (not verified) :: Wed, 08/30/2006 - 10:47am

I like the idea of looking at backs who had a lot of carries in college, but 550 is setting the bar way to low. Even if a guy gets drafted after two years, that's only 275 carries a year, way below the 350 carries a year that FO points out as the breakpoint in the NFL. Also, considering the difference in talent level, it's likely that carries have a lot less wear on a back in college than in the pros.

10
by C (not verified) :: Wed, 08/30/2006 - 11:06am

Re: #9

275 carries in 12 games ~= 23 carries / game

350 carries in 16 games ~= 22 carries / game

So if you look at it on a carry-per-game basis, 275 carries in a college football season is comparable to 350 carries in a NFL season.

11
by Andrew (not verified) :: Wed, 08/30/2006 - 11:16am

B #9:

I like the idea of looking at backs who had a lot of carries in college, but 550 is setting the bar way to low. Even if a guy gets drafted after two years, that’s only 275 carries a year, way below the 350 carries a year that FO points out as the breakpoint in the NFL. Also, considering the difference in talent level, it’s likely that carries have a lot less wear on a back in college than in the pros.

But lets think about this more closely. For men, their last years of growth are ages 18 to 21, with the final growth often occurring in one last spurt of leg growth. It doesn't seem outlandish to think that to add to this the bulking up in weight and muscle, the large increase in weight and other training during this time, and the amping up the impact of football collisions from the pool of players dramatically improving from High School to College could all be factors that would dictate the wisdom of limiting carries for a running back during these years. The player's body is simultaneously being stretched from growth, compressed from collisions, and strained by unprecedented workouts. Surely there is something to this.

12
by Boston Dan (not verified) :: Wed, 08/30/2006 - 11:18am

Please keep this quiet. I'm not sure if anyone in my fantasy league realizes that Maroney is going to be the man this season. I still hear people talking about Dillon as a high pick. I guess they haven't watched the preseason games at all or they're engaging in hardcore disinformation tactics.

13
by Ima Pseudonym (not verified) :: Wed, 08/30/2006 - 11:36am

Re 11:

Do you have any evidence to actually back up that speculation? Why not suppose that it is important to subject the body to the sorts of stresses it will be ultimately be subjected to while it is still maturing. Might this not help ensure that the body develops so as to be best prepared for the trauma it will later face?

14
by Will Allen (not verified) :: Wed, 08/30/2006 - 11:38am

Yeah, given that Barber and Maroney played on the same team, Barber had a decent rookie year (ball protection seems to be his biggest weakness as a pro), and Maroney clearly had far superior NFL potential, since he is faster and more elusive, Maroney was a good risk for the Pats. I saw Maroney's entire career at Minnesota, and the guy had no ego problems in sharing the spotlight with Barber, and by all reports had an excellent work ethic. He is no running back diva, which bodes well with him fitting in on a Belichik roster.

I always liked the Pats, but the selection of Maroney converts me into a full-bore irrational Patriots homer. You do know that Belichick is a genius, don't you, and how could that Vinitieri be such a Benedict Arnold.......?

15
by Sophandros (not verified) :: Wed, 08/30/2006 - 11:40am

Troy Davis had phenomenal rushing stats in college...

16
by Vern (not verified) :: Wed, 08/30/2006 - 11:58am

I still come back to, why does the conference matter (as long as you're talking about a quality 1-A conference)? Is the contention here that the Big 10 has some special properties, rules, or play styles that only impact RBs?

If not, how is "Big 10 RBs" any different than "RBs with black hair" as a grouping?

I understand the stats were hard to get, but shouldn't we (for starters) compare to all 1-A "Majors" for RBs with similar carries?

17
by Disco Stu (not verified) :: Wed, 08/30/2006 - 12:27pm

College teams play 70% of their games within conference. And year in/year out styles of play tend to persist in each conference. So to me seperating Big Ten vs. SEC stats in college is like the BP guys seperating PCL vs. International League stats for AAA ball. BP translates stats across minor leagues and levels in order to compare. It seems like FO is starting to develop that capability- evidenced by the "Big East recievers/SEC backs tend to outperform" rule, so maybe down the road we can get a translation system. But for now segreagting by conference for comparisons makes sense.

18
by GlennW (not verified) :: Wed, 08/30/2006 - 1:13pm

This workload hypothesis does not bode well for Adrian Peterson, already at 559 college carries and saddled up to be ridden like a horse again this season. Peterson missed all or most of four games last year and played hurt throughout-- otherwse he'd be up over 650 carries in just two seasons at ages 19-20. Peterson may turn pro after this (his junior) season, but he'll probably have ~900 carries on the frame in just three college seasons.

19
by Bill Barnwell :: Wed, 08/30/2006 - 1:14pm

Couple of quick responses.

- If you feel like these articles are just scratching the surface, well, there's a reason for that. These aren't full-fledged research pieces for the FO audience, but very much pieces to introduce the FO methodology and statistical analysis to readers of the Boston Sports Media site who aren't familiar with/necessarily friendly to statistical analysis.

- I didn't say that Big Ten backs perform/underperform relative to other conferences precisely because I didn't get a chance to perform that research.

That being said - here's a list of SEC first round backs from the same era:

Fred Taylor JAC
Robert Edwards NE
John Avery MIA
Shaun Alexander SEA
Jamal Lewis Ravens
Deuce McAllister NO
Ronnie Brown MIA
Cadillac Williams TB

Compare that to the Big 10 backs. Maybe I should have put that in.

As for the conclusions to be drawn - you can't compare Duckett to Maroney and say with authority that they'll perform remarkably similar. When you take ten backs from the Big 10, though, and nine of them have underperformed expectations, I feel reasonably comfortable saying that Maroney's going to have to buck the odds, which is exactly what I said.

20
by Sophandros (not verified) :: Wed, 08/30/2006 - 1:24pm

Bill, you forgot Joe Addai on your SEC list, since we're looking at Maroney.

Plus, this gives an opportunity to make this a Big 10/SEC pissing match AND a Patriots/Colts pissing match all in one! Let's not forget that Brady played in the Big 10 and Manning played in the SEC...

Yes, I'm evil.

21
by Bill Barnwell :: Wed, 08/30/2006 - 1:32pm

Sorry Soph - the database we've compiled is only through 2005, hence the lack of Addai as of yet. Besides, he doesn't mean anything yet when we are comparing NFL performances.

And the fact that you bring up Brady/Manning is just cruel.

22
by zlionsfan (not verified) :: Wed, 08/30/2006 - 2:08pm

One trend that I have noticed on this site is that there seem to be many occasions where people cite individual players as counterexamples, as JKL does above, and then imply that the counterexamples disprove the theory.

Very few FO articles I've read, either on this site or elsewhere, suggest that something is true for all players in a group. They tend to say that something is likely for players in that group.

Yes, Larry Johnson is an example of a back from a Big Ten school who has had, to this point, success in the NFL. But he doesn't disprove the "Big Ten RBs tend to disappoint" theory any more than LaDainian's success, in and of itself, is a reasonable basis for drafting more Mountain West alumni, er, players who've attended ... anyway, RBs from Mountain West schools. Now, if other Mountain West RBs have NFL success, especially if they approach that level, then yes, you could consider that theory. However, we'll need some NFL GMs to help us out ... the list of drafted Mountain West RBs from the last few drafts is somewhat small.

I wouldn't necessarily reach the same conclusion that Andrew did. I'd be more likely to look at Big Ten heritage on the same level as schedule: if two RBs are pretty much even, I'll probably take the non-Big Ten back, or I'll take the SEC back.

Now, about MAC QBs ...

23
by Pat (not verified) :: Wed, 08/30/2006 - 2:10pm

College teams play 70% of their games within conference.

I think you're being generous - the out of conference games most teams play aren't really games, except for 1, maybe 2. No one cares how Larry Johnson does against Northern Illinois.

But obviously, that just strengthens your argument.

24
by Disco Stu (not verified) :: Wed, 08/30/2006 - 2:23pm

Re #22- I agree- I think I made this point on one of the draft message boards... Tom Brady was a 6th round pick, so you don't need to take a QB til the 6th round. No, that's not what it means. Drives me nuts.

Side point- these conference trends are really interesting to me... does anyone know which conferences produce the best players at each position?

25
by jason (not verified) :: Wed, 08/30/2006 - 2:26pm

With all due respect to Bill, I have to agree with some of the other comments here panning this article. I think there are a lot of holes, namely:

1. The Pats selection of a runningback in the first round was not surprising at all. I know many people had the pats going LB in the first round, but if you look at bb's draft history, the emergence of ben watson, and the importance of the rb position both to the pats offense as well as brady's life expectancy, picking a rb in the first round was a no brainer.

2. While a bit contrary to the point above, I think it is a mistake to look at past Pats drafts and conclude that "Patriots organization has identified tight end and defensive line as positions that are both undervalued and worthy of multiple high draft picks"

Clearly these positions are important, but I think it speaks more to the Pats choosing the best fit for the team at their draft position than it does an emphasis on any position or position group

3. While possibly true, the comments equating Corey Dillon to 30 year old cheese strike me as premature, preseason notwithstanding.

4. The rb stats provided in the article are interesting but not compelling because they don't normalize for offensive system, offensive line, or rb style. All of these factors can impact the durability of a runningback as much as, if not more than, the number of carries.

Not accounting for the impact of these factors leaves the analysis as incomplete.

5. Say what you want about Ricky Williams, but I don't think it is accurate to suggest that there is a causative relationships between his durability issues and the number of carries in college. If anything, his durability issues would seem to most likely spring from a combination of his high nfl workload as well as the overall impotence of the passing component of the offenses he played on.

26
by Tom (not verified) :: Wed, 08/30/2006 - 2:36pm

I'm sensing the need for somebody to do a research project along the lines of the "where are tackles drafted" piece, but for running backs. Does anybody else want to volunteer?

27
by Pat (not verified) :: Wed, 08/30/2006 - 2:49pm

If anything, his durability issues would seem to most likely spring from a combination of his high nfl workload as well as the overall impotence of the passing component of the offenses he played on.

Depends. The teams could've used him that hard precisely because he was used that hard in college, believing he could handle it.

28
by Sophandros (not verified) :: Wed, 08/30/2006 - 3:05pm

For the record, LaDainian Tomlinson played his career in the WAC, not the Mountain West, as TCU just moved to the MWC this year.

29
by GlennW (not verified) :: Wed, 08/30/2006 - 3:09pm

> Very few FO articles I’ve read, either on this site or elsewhere, suggest that something is true for all players in a group. They tend to say that something is likely for players in that group.

True, but it also seems that sometimes a legitimate disclaimer such as this one gets ignored in the analysis of the individual player and team situation, attempting in practice to place far more accuracy to the theory than was ever intended or is statistically supportable. I think I've even heard applications to the extreme opinion that a RB should almost never be drafted in the first round. I don't agree with that even in this example with the Patriots and Maroney, even where Maroney is far from a sure thing (Jason gives good reasons for supporting the pick in #25).

30
by Vern (not verified) :: Wed, 08/30/2006 - 3:55pm

Is it really that hard to just NOT group RBs by conference for comparison? I suspect you'll find that Larry Johnson and any other successful RB would have a lot more in common with other successful backs (SEC or not), and that the few SEC failures would have a lot more in common with other failures, whether Big 10 or not.

I'd certainly find it more interesting to estimate odds of success based on a metric that can actually be attributable to the player, rather than by some patch attached to the players old jersey.

Maybe it's some new stat that would group all the successes together and separate them from the failurs, like YPC - some measure of OL strength * offensive system factors.

Conference just seems like something that should be very low on the list of possible factors to look at, along with grass vs. turf field, in that it is a such a gross generalization of playing styles, programs, and opponent strengths.

31
by Craigers (not verified) :: Wed, 08/30/2006 - 3:57pm

Troy Davis had phenomenal rushing stats in college…

Admittedly, though, Davis's size was always two strikes against him in the NFL. Davis has played very, very well up here in the CFL... we were been very privileged to have him in Hamilton (where I live) until he was traded last year.

I think, based on what I saw of him here, that he could certainly have played in the NFL, either as a third-down back or as a featured tailback in a pass-heavy offense.

32
by GlennW (not verified) :: Wed, 08/30/2006 - 4:13pm

> Is it really that hard to just NOT group RBs by conference for comparison?

Adjustment for strength-of-schedule (as derived from a college-wide neural network etc.), period, would be preferable. But if you're trying to break it down to the level of an adjustment for strength-of-opposing-rushing defenses, etc., that's a lot of data and can get pretty complicated.

In general I agree with you though. Maybe Larry Johnson should have taken a mandatory Big Ten-deduction, but a downward adjustment from LJ's 8.0 yards/carry in 2002 is still going to leave you with a superlative performance, regardless of conference.

33
by JKL (not verified) :: Wed, 08/30/2006 - 4:16pm

Now that I have had time to look at it, let me explain some of my thinking here. Some others have captured some of the thinking I have.

I am not offering Larry Johnson as a way of disproving the rule. I am saying that applying this same sort of analysis would have led you to make a gross error on LJ, and why? because of what conference he played in compared to some guys 5-7 years earlier, where absolutely none of the players are the same? Rather than actually considering his own abilities?

I do not believe the "conferences are stagnant" position. I am in Big XII country. The conference has had its ups and downs as far as talent level, not every season is equal. I am sure the same is true for the SEC or Big Ten. Coaches and programs change. Styles change. Talent level and things like performance in bowl games fluctuates. Some years, the Big Ten is better, some years the SEC.

It may be true that Big Ten backs were overvalued 10 years ago (small sample size and all would lead me to be very hesitant about saying that with as much confidence as this article), but that does not mean they will continue to be, and are at this time. The most recent comparable by conference is, in fact, Larry Johnson, based on draft position and conference affiliation. Perhaps a comparison of RB's drafted from these two conferences in the decade prior 1986-1995, would shed some light as to whether the Big Ten's string of busts in the mid and late 90's tells us much of anything about Laurence Maroney.

Emmitt Smith, all-time rushing leader is there for the SEC, but other than that, I see alot of similarities between Big Ten and SEC. The SEC top 10 backs were Bo, Brent Fullwood, Tim Worley and Garrison Hearst. Big 10 only had Keith Byars. At the bottom of the round, would you rather have Neal Anderson and Rodney Hampton, or Lorenzo White and Robert Smith?

I guess my point is, just because there is what "appears" to be a trend (with small sample size) doesn't mean it will continue and still exists. The big ten running backs may have been overvalued 10 years ago, but that does not mean they are today.

But I am willing to be convinced, this article just did not do it.

34
by Pat (not verified) :: Wed, 08/30/2006 - 4:17pm

I’d certainly find it more interesting to estimate odds of success based on a metric that can actually be attributable to the player, rather than by some patch attached to the players old jersey.

It's not just about a patch. You're controlling for defensive quality and styles. Apparently, it's just easier to run on Big Ten defenses - or good Big Ten teams' offensive lines are just much better than Big Ten defensive lines.

Look at the two stars in the first-round Big Ten selection: they're both very different than the others. Larry Johnson has the highest yards/carry of any of them (especially in his last year). Eddie George has the lowest yards/carry, and the most touchdowns in one year.

It basically could be saying "unless you're really completely running over teams (like Johnson) or running with a completely different style than normal (George), 5-6 yards/carry in the Big Ten is just an average NFL running back."

35
by JKL (not verified) :: Wed, 08/30/2006 - 4:18pm

And I would also be interested in the workload research. I am skeptical. I think Barry Sanders, Emmitt Smith, and Eric Dickerson had pretty good workloads, but maybe my recollection is faulty.

36
by Pat (not verified) :: Wed, 08/30/2006 - 4:24pm

I am saying that applying this same sort of analysis would have led you to make a gross error on LJ, and why?

Look at LJ's stats again. 7.7 yards/carry in his last year - far and away better than any of the others. Less so for his career stats, but still the highest.

A fair analysis of Johnson would've concluded that yes, Big Ten backs have traditionally struggled in the NFL. But Johnson looked like the best back coming out of the Big Ten for 10 years.

Maroney doesn't look like that. Having seen Maroney in college several times, I do have to say that he never struck me as anything impressive. Minnesota's offensive line was widely considered to be one of the best in the Big Ten - when they got shut down completely in the game versus Penn State, Maroney did nothing.

But I won't go and say he won't be any good. In fact, I think he probably will - offenses in the Big Ten have been rapidly changing in the past few years, and that may end the "Big Ten RB curse".

37
by Pat (not verified) :: Wed, 08/30/2006 - 4:29pm

(as derived from a college-wide neural network etc.),

Woah! That's a bit overkill. You just need to do a deconvolution, or if you want to really be sleazy, just do an iterative process. Should be easy. I did a bit of it last year. Doesn't take too long.

I'd use per-play data (like yards/carry), rather than aggregate numbers, since teams want to win, not pile up, say, rushing numbers. Then just calculate the average yards/carry for a season for a team, then, for each game, calculate the percentage that the team's yards/carry is above the other team's yards/carry allowed, reverse for defense. Rinse. Repeat.

38
by GlennW (not verified) :: Wed, 08/30/2006 - 4:32pm

Barry Sanders had a huge workload for one season and one season only (344 carries), but did not exceed the 550-carry total for his college career. Before that he was sharing time with Thurman Thomas and wasn't getting much use.

I tend to believe that college workload is a factor, but a factor towards career longevity rather than a great indicator of success, period. Miles are miles. If the player is injury-free coming out of college, I'm not sure that it makes much difference. Given all the other uncertainties, NFL teams tend not to think what might happen 5 years down the road (i.e. beyond the length of the first pro contract). The player should probably be concerned though-- as with my example of Adrian Peterson, who is already getting beaten up at age 20.

39
by Bill Barnwell :: Wed, 08/30/2006 - 4:32pm

#30 - Vern, we could try and develop some sort of metric that ignores conference, certainly, but if players show evidence of performing better or worse in the NFL based upon what conference they come from, it'd be foolish to ignore that as well. That's why there's conference data in KUBIAK.

#25 - I'm certainly open to criticism of the piece, no offense taken. I'm confused as to how you figure that the success of Ben Watson led to the Patriots taking a RB in the first round, though.

I also think that we're basically saying the same thing when it comes to our analysis of Patriots' drafts - you're saying they take the best fit, I'm saying they see DL/TE as the best fit more often than not.

I also stand behind what I said about Corey Dillon, when you look at his usage history. The essay in PFP2006 and, in fact, the Patriots selection of Maroney makes me feel comfortable saying that.

You'll also note that, in my comment about Ricky Williams, I was careful to say that the evidence that college workload affects pro performance is "(at the very least) anecdotal". Not definite. Not even empirical! Simply anecdotal, strictly because it requires more research. I'd love to do that research, but I don't have the data available and/or the time necessary to compile that data.

40
by Bill Barnwell :: Wed, 08/30/2006 - 4:39pm

JKL, you're using small sample size as an argument to disprove the possibility that Big 10 backs are weak (and that Larry Johnson can't be judged by his colleagues therein), but then using it to prove that league quality changes (since, as you said, some years the SEC is "better" than others). How do we know that? Performance in four bowl games? Higher ranking? Teams with better records (which doesn't necessarily prove anything about conference quality but instead, individual team quality)? That's a significantly smaller sample than ten years of data.

I am certainly open to the idea that this area needs a LOT more research and this article didn't prove that Big Ten backs are worse, on average, than other conference backs. But just as my article didn't convince you of its truth, to be entirely honest, your argument hasn't convinced me that I'm wrong, either.

41
by Will Allen (not verified) :: Wed, 08/30/2006 - 4:45pm

Interesting point about offensive line strength relative to defensive line strength within a conference, Pat, and that can be further broken down into offensive line strength relative to below median defensive line strength. It'd be interesting to see how may offensive lineman per team from each conference over the past 15 years or so have gone on to play two or more years in the NFL. Of course, given how pass blocking is more important in the NFL, relative to the college game, and was particularly so until pretty recently, this metric wouldn't be perfect, either.

42
by GlennW (not verified) :: Wed, 08/30/2006 - 4:45pm

> Woah! That’s a bit overkill.

Hey, I'm an engineer, not a mathematician, so (out of ignorance) I'm not actually going to do any of the work here. The problem with your suggestion though (as I see it) is the fundamental problem in evaluating conference tendencies in the first place. The conference games don't tell us anything about any biases (here, for quality of rushing defense), and the few non-conference games contain many cupcakes, where a RB might look fantastic against a defense that itself looks great (on paper) within its own weaker conference. So you have to look at how your opponents measured up against their opponents etc., and that becomes fairly daunting especially given the sample size.

43
by Disco Stu (not verified) :: Wed, 08/30/2006 - 4:55pm

The Big Ten curse was a product of offensive systems designed around RBS. Michigan, for one, had a string of disappointing RB- Biakabatuka, Wheatley, Perry, Thomas. And also had a string of QBs who were better than expected pros- T. Collins, Grbac, Brady, and even Navarre. Whereas Florida had a string of disappointing pro QBs and some solid pro RBs.

Maybe a draft strategy is to take players who looked good despite systems in which they weren't showcased.

44
by GlennW (not verified) :: Wed, 08/30/2006 - 5:06pm

> Interesting point about offensive line strength relative to defensive line strength within a conference, Pat, and that can be further broken down into offensive line strength relative to below median defensive line strength.

Yes. Ki-Jana Carter ran behind one of the greatest offensive lines in NCAA history (and I still thought he'd be good in the NFL, and he might have been, but). Similar story for just about everyone who ever racked up 1500+ yards for Nebraska. I think such individual team factors blow away even the most significant conference-wide biases. This is just a tough nut to crack, really.

45
by Tom (not verified) :: Wed, 08/30/2006 - 5:07pm

Re #36 and others
I was among the skeptics of LJ coming out of college, based not on the numbers he put up, but how he put them up. If you go back and look at his 2002 statistics, he feasted against bad teams and comparatively struggled against the good run defenses he faced. He played six games against teams that ranked 89th and below nationally, including games against 4 of the bottom 10 teams. In those 6 games, he averaged 9.9 ypc and 246 yards per game. Four other games came against rush defenses ranked between 46th and 52nd; in those, he averaged just short of 100 ypg and 5.61 ypc. Against the top 3 rush defenses he faced, Michigan, Iowa, and Ohio State, 31st, 5th and 3rd nationally, he averaged 4.16ypc and 71 ypg. Frankly, from those stats, and from what I saw of him, I still believe he's a back who is especially capable of getting a lot of yards when running behind an offensive line with an advantage over the defense, but that his advantages decrease the more you reduce that difference. Frankly, unless the Chiefs still manage to put together a really good line this year, I expect him to decline dramatically.

46
by Pat (not verified) :: Wed, 08/30/2006 - 5:44pm

So you have to look at how your opponents measured up against their opponents etc., and that becomes fairly daunting especially given the sample size.

Getting an absolute read on a RB - that is, relative to the entire college football landscape - is completely impossible.

If you look at a connection graph in college football, it's like a bunch of islands with a few bridges between them. The teams in the conferences themselves are extremely connected - in the Big Ten, virtually everyone plays everyone else - but the conferences are ridiculously poorly connected.

That's why you can't get rid of looking for conference biases like this. They just don't play enough out of conference games for it to matter.

47
by Oswlek (not verified) :: Wed, 08/30/2006 - 5:51pm

Not to sound like a jerk or anything, but all I need to see that Maroney is going to be at least decent (barring injuries, of course) is my eyes. He has looked every bit the part of a first round running back, and this was against the other teams' 1s. It isn't as if he has come in and run over guys who will be working construction next week, he has been impressive against the starters.

Maroney has shown elusiveness and power. Several of his carries have come on running downs, after the D has sufficiently held Dillon in check. Yet, Maroney will usually slide threw the same blocking that gave Dillon 2 yards for 5-6. Plus, he has rarely been taken down by the first hit.

Not only that, but Maroney has been surprisingly effective as a blocker. Against Atlanta he stoned a blitzing LB who had come in untouched, so he had a running start.

I am not worried in the slightest about Maroney.

48
by jason (not verified) :: Wed, 08/30/2006 - 6:05pm

Bill -

We can certainly agree to disagree on Corey Dillon. You most certainly could be right, I personally would prefer to wait until week 4 or so before coming to that conclusion.

As far as my comments about Ben Watson; I apologize for the confusion. What I meant was, that it appeared that, with his emergence, the Pats would probably use more 1rb/2te sets in 2006 than years past in order to get their best offensive personnel on the field.

In order to effectively run your offense out of that formation, the Pats needed to acquire another back to help spell Dillon and share the load. Therefore, since they weren't able to get any free agents to spell Dillon, they absolutely needed to draft another rb.

49
by GlennW (not verified) :: Wed, 08/30/2006 - 6:11pm

Yeah, I think Maroney is going to be okay in the NFL too.

> That’s why you can’t get rid of looking for conference biases like this. They just don’t play enough out of conference games for it to matter.

I thought that's what you were trying to do with your iterative process example. I guess you just meant that you'd rate/rank the RBs within the conference against the exact conference schedule/strength of defenses (there are slight two-game schedule differences within the Big Ten for example, which could make a difference).

It's my understanding that even given the small number of non-conference-game data points, systems such as Sagarin's nonetheless rate conferences and teams based on that limited available data. That's what I originally suggested, but I understand that it's complex and not necessarily very meaningful (again, given the low sample size of non-con games).

50
by Rich Conley (not verified) :: Wed, 08/30/2006 - 6:29pm

re:46

While I agree with what you're saying, that there just arent enough non conference games to take the conference out of it, I dont agree that conferences are static. I just dont think comparing a running back 10 years ago in the Big10 to one of today is any more valid than comparing him to a running back in a different conference of 10 years ago.

Because the data is so sparse, I think the only useful comparison is guys who were in the same conference at the same time, and that just limits data further.

51
by Pat (not verified) :: Wed, 08/30/2006 - 6:43pm

I thought that’s what you were trying to do with your iterative process example.

Nah, there what you're doing is trying to get rid of the relative intraconference strength issues. That is, certain conferences are much, much deeper than others.

It’s my understanding that even given the small number of non-conference-game data points, systems such as Sagarin’s nonetheless rate conferences and teams based on that limited available data.

Yup. And that's one of the main reasons why they're not that great. Remove a non-conference game from a team's schedule, and the rating changes about 5 times more than removing a conference game.

The 2-game schedule differences are nothing. Basically, the average distance (i.e. if Penn State plays Notre Dame, and Iowa plays Penn State but not Notre Dame, its distance to Notre Dame is 2) between two teams in the Big Ten averages 1.2. The average distance between a team in the Big Ten and any random team in Division I is something like 3, if memory serves. The average distance in the NFL is something like 1.6. In fact, there's no team in the NFL that's a distance of more than 2 from any other. In college football, most teams are at a distance of 3, and some even at a distance of 4 (Northwestern - Penn State - Minnesota - Florida Atlantic - any other team in the Sun Belt).

I really do think that any attempt at figuring out college football statistics would be best off just treating the conferences as if they never played each other, and then trying to gauge relative strength of conferences with longer-term data.

52
by Sophandros (not verified) :: Thu, 08/31/2006 - 12:26am

35: If I recall correctly, Dickerson shared time with Craig James at SMU, and Barry Sanders sat the bench behind Thurman Thomas (put him in the HOF, please) for a couple of years, which limited his carries.

53
by JKL (not verified) :: Thu, 08/31/2006 - 7:23am

Re: 40. Bill, just so I am clear, I am not basing the statement that conferences fluctuate on just bowl games. If I made it sound so, it was my error.

I actually used to do my own power rankings based on pf/pa and yards, so not as indepth as DVOA (for recreational purposes only, ahem) and conference power would fluctuate. As an example, the Big XII of the last two years, mainly due to the North teams, is not nearly as strong as the Big XII of the late 90's, early 2000's.

And I am not even saying there definitely was not a RB curse or that RB's from the Big Ten were not overvalued 10 years ago. But it is plausible that the market corrected itself and/or the Big Ten changed, and teams drafting running backs are not overvaluing Big Ten runners now.

Perhaps we could look at other stretches where players from a position in a certain conference tended to underperform their draft position, and then see if that trend continued 10, 15 years later.

54
by Bill Barnwell :: Thu, 08/31/2006 - 3:46pm

I gotcha JKL. I think, realistically, we'll find out in a couple of years whether they're no longer overvalued.