What's new
Fantasy Football - Footballguys Forums

This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

Towards an Objective Measure of Talent in RBs (1 Viewer)

wdcrob said:
Thanks Construx (and everyone), I actually saw that article while I was doing this project and was glad to see one piece of this was confirmed by someone else.
Click to expand...
No problem. FWIW, I also did a break down the tape article there (here) on a couple of RBs and Jackson really impressed me the most on tape. That would seem to fit with your conclusion as well. Now that I think of it, I was least impressed with Michael Bush on tape and you have him in the 4c category. I sure hope your theory continues to match my tape-watching. :confused:

 
Wow, awesome post wdcrob :goodposting:

Already looking forward to seeing your analysis of the 2008 RB class

I'm glad that your top 3 resembles mine as well. Even prior to the NFL Draft in April I had Peterson, Lynch and Brandon Jackson as the top 3 backs available in the draft (and not much after them worthy of noting)

All 3 happened to end up in more than ideal situations as it stands

 
I had Jackson as the #3 RB too and was counting on picking him up very late in a startup dynasty. Then he went to Green Bay. Since I have Morency rated as basically a mug, I decided to take Jackson at #57 (this was May) because I didn't want to risk not getting him.

My leaguemates flipped out. Sure hope I'm right. :fishing:

 
KnowledgeReignsSupreme said:
az_prof said:
[First, thank you for an interesting read and theory. What I don't understand is why age of rookie year should matter. Do you have a theoretical explanation? It just doesn't seem like it should matter.
Click to expand...
Doug Drinen has a theory on this:http://www.pro-football-reference.com/arti...todThomAn01.htm

oh and :JasonSehorn:
Click to expand...
--The younger guy is simply a better talent. If he's able to do at 21 what the other guy couldn't do until he was 24, he's probably just better.This is the point someone made about going pro as a junior.

--Assuming they both peak at age 27, the young guy has 6 years of improvement left, whereas the older guy only has 3. It's as though his talent has more time to accrue interest.

I think this is more relevant than people above were giving it credit. Especially given the baseball findings mentioned in that article.

Was thinking about all this again last night... and realized that if I'm right (no way to tell yet - could just be a model that 'backtests' really well) I'm basically saying that there's an inefficiency in the NFL RB market. Actually, two inefficiencies:

1) GMs are biased towards heavier backs, and underestimate talent b/w 205 and 220.

2) They aren't accounting for the additional improvements that happen with backs between 21 and 23 years old.

 
There's another inefficiency which happens in baseball MUCH more than in football: talent blocking your path.

In baseball, by and large, young players come up through the minors and spend significant time there - a year or more. so the team who drafts them will do some need-based drafting, but what was a need at one point may cease to become one. You see this with teams that have top prospects at a position blocked by an establihed vet who is good.

This doesn't really happen at the NFL level much. You'll see it here & there: Deuce McAlister, Cedric Benson are examples of players drafted by teams with good players in front of them at those spots, but developmentally, almost all players come through college, so you get to see things on that level and judge.

More relevant to this study would be when a player is blocked at college, and moves to another school, or waits until an established star graduates or leaves to take over. this doesn't mean they didn't have NFL talent at the younger age, it means they didn't have an opportunity to show whether they did or not have such talent.

 
stevegamer said:
More relevant to this study would be when a player is blocked at college, and moves to another school, or waits until an established star graduates or leaves to take over. this doesn't mean they didn't have NFL talent at the younger age, it means they didn't have an opportunity to show whether they did or not have such talent.
Click to expand...
Might need to do an informal study on this. My hypothesis would be that NFL caliber younger talent stuck behind a good but not NFL caliber starter will tend to push the vet aside quickly. Anecdotally it seems college fans know when they guy lower on the depth chart is going to displace the older starter sooner rather than later.
 
We don't have their draft positions yet obviously, but here are their other two measurements, knowing that weight may not be 100% accurate.

Code: 
Name				  Wt	   Age on 9/1/08Darren McFadden	205	21.03Jonathan Stewart	230	21.47Rashard Mendenhall	210	21.22Felix Jones	200	21.33Ray Rice	200	21.62Steve Slaton	195	22.67Jamaal Charles	200	21.70James Davis	205	22.68Chris Johnson	200	22.96Tashard Choice	205	23.80Mike Hart	196	22.41Matt Forte	223	22.74Marlon Lucky	210	22.52Ian Johnson	194	22.95Dantrell Savage	195	23.56Ryan Torain	213	22.08Yvenson Bernard	202	22.87
 
What I find so interesting is that the original premise by wdcrob was that younger was measurably better. And when you look at the list I posted, the "highest" ranked RBs are also the youngest, with my top 7 all being under 22 years old other than Slaton.

 
Last edited by a moderator:
I have done some similar sort of work with RBs before. One factor that also seems to matter is height. A tall back (6'2'' or taller) may have a good few seasons but wears down quickly. I did this a long time ago but IIRC I would much prefer a short RB over a tall RB. Guys like Brandon Jacobs, Adrian Peterson, and Larry Johnson are more likely to have 2-3 great seasons then wear out.

I am sort of curious about the age issue. Could the difference just be juniors that are talented enough to come out are better than seniors? It would make sense that this causes the difference you see in age.

 
Last edited by a moderator:
sn0mm1s said:
I have done some similar sort of work with RBs before. One factor that also seems to matter is height. A tall back (6'2'' or taller) may have a good few seasons but wears down quickly. I did this a long time ago but IIRC I would much prefer a short RB over a tall RB. Guys like Brandon Jacobs, Adrian Peterson, and Larry Johnson are more likely to have 2-3 great seasons then wear out.I am sort of curious about the age issue. Could the difference just be juniors that are talented enough to come out are better than seniors? It would make sense that this causes the difference you see in age.
Click to expand...
I think that was generally the conclusion although being younger and having a little more time to both grow in to your body and learn the position was also mentioned.
 
I've actually spent a good amount of time refining this. Nothing too earth-shattering, but there are a few things wrong with what I posted earlier. And I've added a few tweaks that seem to improve things. It's simpler, with fewer categories, if nothing else.

Brandon Jackson and Laurence Maroney remain as the best test cases. Neither has really proven themselves yet, and both are considered elite prospects in this system.

I'll post an updated version after the draft with predictions for the 2008 draft class and some other unproven RBs that it's worth keeping an eye on.

 
No. I've got to go back and standardize one of the things I'm measuring and it's going to take forever. And it's nowhere near as good as the RB thing, which I've been playing with for almost a year now, off and on.

After the draft I'll post something on the WRs too, with whatever I've got. The one thing it's actually good at is separating 1st round WR busts from those that will be successful. Sort of. And after that there's only limited utility.

For what it's worth, here are the WRs the system liked (1994-2007):

Marvin Harrison Syracuse

Tedd Ginn Ohio State (maybe, there aren't enough with his profile to tell for sure)

Roy Williams Texas

Randy Moss Marshall

Torry Holt North Carolina State

Andre Johnson Miami

Larry Fitzgerald Pittsburgh

Calvin Johnson Georgia Tech

(gap)

Lee Evans Wisconsin

Joey Galloway Ohio State

Terry Glenn Ohio State

Santana Moss Miami

and here are the ones it didn't:

Charles Rogers Michigan State

Charles Johnson Colorado

Troy Williamson South Carolina

Yatil Green Miami

Mike Williams Southern California

Reggie Williams Washington

Bryant Johnson Penn State

J.J. Stokes UCLA

Matt Jones Arkansas

Ike Hilliard Florida

Travis Taylor Florida

Peter Warrick Florida State

Rod Gardner Clemson

Donte' Stallworth Tennessee

Johnnie Morton Southern California

Eddie Kennison Louisiana State

Javon Walker Florida State

and here are the ones in-between:

David Terrell Michigan

Michael Westbrook Colorado

Koren Robinson North Carolina State

Braylon Edwards Michigan

David Boston Ohio State

Plaxico Burress Michigan State

Keyshawn Johnson Southern California

 
Thanks for the WR info..

BTW...I'm a big fan and I'm looking forward to your take on the 2008 class.

Big Props!

 
I think your research and insights are awesome. Great work. I play in a TD heavy scoring system dynasty league and the 4a catagory is super-gold in my league system. I'm going to focus on that going forward. I've missed many times historically in my rookie, very high draft picks.

My basic RB system though is size, age, speed etc. don't matter...its just a matter of watching RB's in game situations in the NFL and seeing if they have 1) moves and 2)vision. Bettis had both. Barry Sanders had both. They don't look anything alike at RB. That is gold beyond size, shape, age or speed.

 
Go back and compare Bettis with my 'elite' backs selected in the same draft position. He's nowhere near as good.

According to FBG Bettis averaged 12.7 PPG (PPR) in seasons that started before he was 30 years old.

Comparing that to Fred Taylor, big back drafted in almost the same exact spot as Bettis (#9, vs #10 for Bettis), you find that he averaged 16.7 PPG (PPR).

And those results are consistent with everything I found. All other things being equal the guys on my 'elite' list almost always outscore similarly drafted backs who are the 'wrong' weight or age.

In fact, going back and looking at my updated spreadsheet, NONE of the established big backs on my 'elite' list have a PPG (PPR) as low as 12.7. Only Cedric Benson and Chris Perry are below that. Perry may never get a chance to show what he can do, and the jury's still out on Benson. The other 12 backs have averaged 16.6 PPG (PPR), and I didn't adjust that to take out seasons before they were feature backs, or were over 30 years old.

 
Last edited by a moderator:
wdcrob said:
Go back and compare Bettis with my 'elite' backs selected in the same draft position. He's nowhere near as good.

According to FBG Bettis averaged 12.7 PPG (PPR) in seasons that started before he was 30 years old.

Comparing that to Fred Taylor, big back drafted in almost the same exact spot as Bettis (#9, vs #10 for Bettis), you find that he averaged 16.7 PPG (PPR).

And those results are consistent with everything I found. All other things being equal the guys on my 'elite' list almost always outscore similarly drafted backs who are the 'wrong' weight or age.
Click to expand...
I'm with you...I don't play in a PPR league but a TD heavy league. Davenport has more fantasy points than FWP in my scoring system. Sad system, but you have to adapt. Your 4a grouping is the gospel criteria for me.
 
Group 1a: 16.2 PPG (14 backs, or 11%)

Group 4a: 15.0 PPG (10 backs, or 8%)

Group 1b: 12.9 PPG (9 backs, or 7%)

Group 2a: 10.4 PPG (8 backs, or 6%)

Group 4b: 09.3 PPG (14 backs, or 11%)

Group 2b: 09.1 PPG (13 backs, or 10%)

Group 2c: 06.1 PPG (7 backs, or 5%)

Group 3-: 04.5 PPG (27 backs, or 21%)

Group 4c: Low PPG (26 backs, or 20%)

Above is your grouping for PPR scoring which generally favors smaller backs who catch alot of screen passes and such. Is there anyway you could adjust it for TD only scoring system? I have a hunch that 4a will be "off the charts"...just a hunch. I'm just trying to see about my dynasty league drafting of rookie RB's going forward.

 
Captain Spaulding said:
Group 1a: 16.2 PPG (14 backs, or 11%)Group 4a: 15.0 PPG (10 backs, or 8%)Group 1b: 12.9 PPG (9 backs, or 7%)Group 2a: 10.4 PPG (8 backs, or 6%)Group 4b: 09.3 PPG (14 backs, or 11%)Group 2b: 09.1 PPG (13 backs, or 10%)Group 2c: 06.1 PPG (7 backs, or 5%)Group 3-: 04.5 PPG (27 backs, or 21%)Group 4c: Low PPG (26 backs, or 20%)Above is your grouping for PPR scoring which generally favors smaller backs who catch alot of screen passes and such. Is there anyway you could adjust it for TD only scoring system? I have a hunch that 4a will be "off the charts"...just a hunch. I'm just trying to see about my dynasty league drafting of rookie RB's going forward.
Click to expand...
I've actually done away with that - there are only four viable prospect groupings now. I don't have the TD info though. Just took PPR points.
 
Out of curiousity, wdcrob, have you ever considered building your models off of a limited sample (say, 1990 to 2002) and then using the unused sample years (in this case, 2003, 2004, maybe 2005 if we have enough data to judge) as a test?

The big problem judging a model's effectiveness on the data that was used to create the model is that there's all kinds of bias. I mean, imagine that only three WRs had been drafted in NFL history. One was 4'10", one was 5'5", and one was 6'6". The 4'10" went on to become an uberstud, the 6'10" one was a bust, and the 5'5" was serviceable. I now say that I've created this wonderful model that correctly identified which WRs went on to be studs, and look, it correctly identified the 4'10" as a future stud... but it didn't. It knew beforehand that the 4'10" guy was a stud and stacked the deck so that it would predict studhood from the guys it already knew was a stud. It's impressive that your WR system correctly identified Harrison, Moss, Fitz, etc as studs and Rodgers, Green, Stokes, etc. as busts... but did it really identify them as studs, or did it know before hand that they would be? Did it say "guys who fit this profile will be studs", or did it just say "guys who look like Marvin Harrison will be studs"?

By saving some years and withholding the data, you can then use those years to test the predictive power of your system, since they aren't tainting the projections or introducing bias.

 
SSOG, there just wasn't enough data to hold some of it back. I think there are only like 25 elite backs in 12 years. Also, the NFL has changed. If you go back far enough the minimum weight to be successful was clearly lower. You used to see quite a few good backs at 185-200 pounds. Now, almost never. (btw... if I had to pick two backs who 'break the mold' it'd be Warrick Dunn and Jerome Bettis. Those guys are just totally unique.)

The other problem is that a lot of the time you don't really know how good a back is until they've been in the league for at least five or six years and that limits things (only about half the backs in my sample have been there that long). Right now I can tell you that McGahee and K Jones are fantasy starters, as predicted by the system, but are they really the worst of all the 'elite' backs or do they still have a couple phenomenal years in them that will move them up the ladder? Every year that goes by adds a lot of useful information at this point.

Having said that...

I did go back and look at the top 15 RBs in terms of fantasy points scored since 1960, and 14 of them fit my profiles:

Walter Payton, Emmit Smith, Barry Sanders, Curtis Martin, Marshall Faulk, Thurman Thomas, Ricky Watters, Marcus Allen, Ladainian Tomlinson and Edgerrin James all fit the 'elite' small back profile. Eric Dickerson, Franco Harris and John Riggins fit the profile for big backs. (In fact this is where I started. I grouped the all-time greats together and started looking for commonalities.)

The only RB who doesn't fit the profile is Tony Dorsett. He was well over 23 when he entered the league, and only weighed 192 pounds. I didn't post anything about this earlier, but there's something of a 1st-round exception rule to all of this. The later you get drafted the more you'd better fit the profile to a "T." But get drafted in the 1st ~half of the 1st round and some of your warts get covered up. Dunn, Bettis, Robert Edwards, Dorsett. In general you'd rather have a 'profile' back if they were drafted in the same position, but it's very hard to whiff outright on an early 1st rounder unless they've got character problems (William Green, Lawrence Phillips). Ron Dayne and TJ Duckett are the only two (which sort of proves the point about what a freak Bettis was).

So while you're right that some of what I'm calling a 'system' is probably random and/or biased, I'm pretty sure there's something here. The results may be overstated, but they are useful even allowing for that IMO.

 
Last edited by a moderator:
I am still thinking about this system and really appreciate the effort that went into developing it.

However, after thinking about it, it seems to me that basically it is reinforcing that the draft gets it right and that guys who are good enough as Juniors to come out for the draft are going to be good NFL backs. This makes sense. If you need another year to prove your ability or to move up your draft value you probably have some aspects of your game that need improvement. If you are good enough as a Junior to enter the draft and be a first day pick it means that you were good enough to start early in your college career and to produce well enough to get the NFL scout's attention.

The size thing makes sense too because the big guys who fail by and large are just too slow for the NFL game and yet big as they are they can't run over NFL linebackers the way they did in college. On the other hand, the elite big backs are really some of the best backs in the NFL because they also have speed.

I would like to see what happens if you take into account 40 speed times. My guess is that most of those smallish backs who are in the best category have off the chart speed.

 
az_prof said:
I am still thinking about this system and really appreciate the effort that went into developing it.However, after thinking about it, it seems to me that basically it is reinforcing that the draft gets it right and that guys who are good enough as Juniors to come out for the draft are going to be good NFL backs. This makes sense. If you need another year to prove your ability or to move up your draft value you probably have some aspects of your game that need improvement. If you are good enough as a Junior to enter the draft and be a first day pick it means that you were good enough to start early in your college career and to produce well enough to get the NFL scout's attention. The size thing makes sense too because the big guys who fail by and large are just too slow for the NFL game and yet big as they are they can't run over NFL linebackers the way they did in college. On the other hand, the elite big backs are really some of the best backs in the NFL because they also have speed. I would like to see what happens if you take into account 40 speed times. My guess is that most of those smallish backs who are in the best category have off the chart speed.
Click to expand...
Good points. In some ways it seems like it can be common sense, but it's amazing how often we still make mistakes (Eric Shelton over Braylon Edwards?? :wall: ). My opinion on the size thing was that it wasn't just that defenders are bigger and harder to run over in college, but that big backs have an instinct developed in HS and college to take on the defender instead of try to get around him. It's sort of like the fight or flight instinct. The smaller guys have learned they are better off with flight but the bigger guys have been able to get away with fight all through college so when they are presented with a split second decision with a defender in front of them, they try to run him over. The double whammy in the pros is that it not only is more difficult and can lead to less success in terms of yardage gained, but it also exposes them to more contact and possibility of injury.
 
ConstruxBoy said:
We don't have their draft positions yet obviously, but here are their other two measurements, knowing that weight may not be 100% accurate.

Name Wt Age on 9/1/08Darren McFadden 205 21.03Jonathan Stewart 230 21.47Rashard Mendenhall 210 21.22Felix Jones 200 21.33Ray Rice 200 21.62Steve Slaton 195 22.67Jamaal Charles 200 21.70James Davis 205 22.68Chris Johnson 200 22.96Tashard Choice 205 23.80Mike Hart 196 22.41Matt Forte 223 22.74Marlon Lucky 210 22.52Ian Johnson 194 22.95Dantrell Savage 195 23.56Ryan Torain 213 22.08Yvenson Bernard 202 22.87
Click to expand...
Very nice Construx....now let's extrapolate a little. Here's an attempt at putting these guys in categories taking a few guesses at their draft positions.

1A - 16.2 PPG - (< 22.5 age, 205 - 221 lbs, Top 76.... or Top 16 under 22.5 age)

The Ideal Guys -

McFadden

Mendenhall

4A - 15.0 PPG - (RD1 & 222-240 lbs)

Stewart

1B - 12.9 PPG - (Same as 1A but age 22.5 - 23.02)

James Davis

Marlon Lucky

... both JR's yet to declare. would greatly improve the draft class if the come out.

2A - 10.4 PPG - (Same as 1A, but drafted later (77-109))

Ryan Torain

Let's assume he's drafted between 3.14 - 4.14. Bump him up to 1B is he goes before pick 3.13

4B - 9.3 PPG - (Big Back, under 23 age, 222-240 lbs., drafted after 1st)

Matt Forte

2B - 9.1 PPG - (Drafted before 3.13, EITHER "195-204 lbs and <23.02 age" OR "205-221 and 23-23.5 age")

Jamaal Charles

Felix Jones

Ray Rice

Steve Slaton

.... only way out of this category is to be drafted Top16 (bump to 1A)... Maybe but unlikely

2C - 6.1 PPG - (Drafted between 3.14 & 4.14; either 195-204 lbs or 23-23.5 age)

Mike Hart

Ian Johnson

Chris Johnson

Yvenson Bernard

.... Get drafted lower and off the map for these guys

3 - 4.5 PPG - (Older rookies, >23.5 years old)

Tashard Choice

Dantrell Savage



ETA - forgot to put Charles in the first time around....

 
Last edited by a moderator:
Love this thread!! :thumbsup:

It's basically telling me that if somebody wants to pay a high price for Charles, Jones, Rice or Slaton...then if I believe this model, I should willingly sell my mid-to-late 1st rd draft picks to them. Fantastic stuff.

Looking forward to the inevitable bump after the draft.

Oh, a few thoughts on the model:

- While I see the difficulty due to limited sample size, SOS is spot on with the suggestion to withhold part of the sample and use it to test the model. Maybe something to try in a year or two?

- Will you track the model's effectiveness going forward?

- Somebody (AZ Prof?) mentioned incorporating 40-yd times. My hunch - based on no data - is that 40 times are implicitly taken into account as they likely correlate with the round a player is drafted. Probably particularly true for both really small and really big RBs

Again, great stuff.

Edit to add: is it possible to see a list of every RB included in each category, along with their avg PPG? Curious to see what the performance variance is for each category.

 
Last edited by a moderator:
ConstruxBoy said:
We don't have their draft positions yet obviously, but here are their other two measurements, knowing that weight may not be 100% accurate.

Name Wt Age on 9/1/08Darren McFadden 205 21.03Jonathan Stewart 230 21.47Rashard Mendenhall 210 21.22Felix Jones 200 21.33Ray Rice 200 21.62Steve Slaton 195 22.67Jamaal Charles 200 21.70James Davis 205 22.68Chris Johnson 200 22.96Tashard Choice 205 23.80Mike Hart 196 22.41Matt Forte 223 22.74Marlon Lucky 210 22.52Ian Johnson 194 22.95Dantrell Savage 195 23.56Ryan Torain 213 22.08Yvenson Bernard 202 22.87
Click to expand...
Mendenhall is listed at 225, and is as ripped as Thomas Jones, if that means anything at all...I've noticed that the added 10-15 pounds has really helped him this season...but it is probably something that can't be measured with numbers.

http://fightingillini.cstv.com/sports/m-fo..._rashard00.html

 
Last edited by a moderator:
428143 said:
ConstruxBoy said:
We don't have their draft positions yet obviously, but here are their other two measurements, knowing that weight may not be 100% accurate.

Name Wt Age on 9/1/08Darren McFadden 205 21.03Jonathan Stewart 230 21.47Rashard Mendenhall 210 21.22Felix Jones 200 21.33Ray Rice 200 21.62Steve Slaton 195 22.67Jamaal Charles 200 21.70James Davis 205 22.68Chris Johnson 200 22.96Tashard Choice 205 23.80Mike Hart 196 22.41Matt Forte 223 22.74Marlon Lucky 210 22.52Ian Johnson 194 22.95Dantrell Savage 195 23.56Ryan Torain 213 22.08Yvenson Bernard 202 22.87
Click to expand...
Mendenhall is listed at 225, and is as ripped as Thomas Jones, if that means anything at all...I've noticed that the added 10-15 pounds has really helped him this season...but it is probably something that can't be measured with numbers.

http://fightingillini.cstv.com/sports/m-fo..._rashard00.html
Click to expand...
Yeah, the weights I used were from Tony Nowak's pre-season guide, I believe. They should be updated with combine/pro-day weights when they happen.
 
I'm not sure I've ever been this eager to see a follow-up post.

If you have it put together at all right now, I'd love to see it even before the draft.

That WR split is incredible. Where did Rashaun Woods fall?

 
wdcrob said:
So while you're right that some of what I'm calling a 'system' is probably random and/or biased, I'm pretty sure there's something here. The results may be overstated, but they are useful even allowing for that IMO.
Click to expand...
Hey, I've Sehorn'd the thread, haven't I? You'll get no arguments here. :lmao:
 
You must log in or register to reply here.

Similar threads

rockaction
PSA - NFL.COM's Rushing Yards Over Expectation Is Not A Comparative Statistic
Replies
4
Views
173
rockaction
rockaction
Judge Smails
**Official 5th Pick - 1/2 point PPR strategy thread
Replies
3
Views
603
Judge Smails
Judge Smails
Terpman22
If Jalen Hurts Gets 100+ Rushing Touchdowns = Hall of Fame?
5 6 7
Replies
330
Views
10K
Uruk-Hai
Uruk-Hai

Users who are viewing this thread

Total: 1 (members: 0, guests: 1)
Back
Top