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

Part I, Intro

A little over a year ago I decided I’d spend some time trying to figure out what made players good, and how to predict talent. More specifically, I’d seen the QB prediction system in Football Outsiders 2006 and wanted to replicate it for RBs and WRs. I didn’t expect that I’d find something that performed as well, but wanted to try.

This article summarizes the results of that research for RBs. The article on WRs will follow.

It’s tempting to walk through all the dead ends and the various stages in the process, but I’m realizing that would be a 20 page report. And I’m just that lazy. So here are the findings…

1) Weight matters. After messing with this piece of the puzzle for months I’m confident that there’s a range of weights that makes little difference in the performance of a RB, and that range is roughly 205 to 235 or 240 pounds. Weighing more than 240 is pretty much a death knell for RBs, but the range 195-204 on the low side can be ok, if the other criteria are strong enough. For reasons I’ll cover at the very end, I think it’s worth breaking RBs into two groups based on size.

Luckily it doesn’t seem to matter much exactly where you draw the line between small and large backs, as long as it’s somewhere between 221 and 225. I picked 222 on the subjective basis that Cedric Benson (222) and Willis McGahee (223) run more like big backs and LT (221) ran more like a small (though powerful) back. Your mileage may vary.

It’s important to note that I did my best to use the weight that the back entered the league at. I used pre-draft scouting reports where I could find them and the Pro Football Database when I couldn’t because it consistently trended below a player’s current weight and I suspect that’s because it’s listing rookie weights.

2) Age matters. Guys who are under 23 years old on September 1st of their rookie season do better than guys who are older. And it’s a progression – the younger the better, on average. If you run a regression you get a positive but weak result. And it seems that age is more relevant for small backs.

3) The third thing that matters is the most obvious: draft position. Very few backs chosen after the middle of the 4th round (pick 105-110) are worth owning. And a surprising number of backs who enter the league young and with the right body type succeed despite being drafted in the 2nd and 3rd rounds.

But! But! Warrick Dunn?! Jerome Bettis?! Robert Edwards?! Willie Parker!? Terrell Davis?! Fair enough. There are exceptions to every rule. More on those in a minute, too.

Skipping through a whole bunch of trials and errors again, here’s the important part…

Those three items -- age, draft position (subjective evaluation of a running back’s talent by NFL teams) and weight -- combine dynamically to provide a stunning projection on whether the RB will be successful or not. As with any model it works especially well on the edges, but with a bit of work even the middle grouping can be subset in a way that predicts relative success among the sub-groups.

As is also true with any model, you can tweak this to your heart’s content. The categories below are just the ones that I landed on because I wasn’t seeing much improvement with additional effort. Plus I wanted my life back. The model treats Small Backs and Big Backs separately, which I’ll explain/hypothesize about below. The data set includes all backs taken in the 1st four rounds from 1996 to 2007.

Note that I made no effort at all to adjust for years played, injuries, quality of team or quality of the other backs on the roster. Partly because using PPG eliminates the problem of the games missed due to being a rookie backup or injury, and partly because the model worked damn well without making any of those adjustments. Plus the reasons I mentioned above: I’m lazy with details. And have a life. And couldn’t get anyone with a database to help me.

Part II, Small Running Backs

#1a: These are the cream of the crop for small backs – not just within the required ranges, but squarely on the good side of the range: backs between 205 and 221 pounds, under 22.5 years old, drafted in first 76 picks; and other back drafted in the first ~half of the first round under 22.5 years old. These backs are fantasy gold, with a career average of 16.2 PPG (PPR). They include the following RBs:

Clinton Portis, Edgerrin James, Maurice Jones-Drew, Marshawn Lynch, Reggie Bush (#2 pick), Ahman Green, Laurence Maroney, Brandon Jackson, Kevin Jones, Thomas Jones, Jay Graham, LaDanian Tomlinson, Frank Gore, Adrian Peterson

I couldn’t find much on Jay Graham, but it’s clear he was injured a lot. And it suggests that Thomas Jones would have been an elite back if he hadn’t been in horrible situations for most of his career. Notice that this elite grouping accurately includes the surprising backs Frank Gore, Maurice Jones-Drew and Ahman Green. And it won’t surprise me at all if Brandon Jackson turns into a star in the next couple years.

#1b: As above, except backs between 22.5 and 23.02 years of age (basically the same players, but slightly older). This may be a fallacious differentiation that works only because Williams Green and J.J. Arrington are included in the group, and both of them landed in horrific situations for RBs. But subjectively the backs in group 1a do look stronger than those in 1b. I cheated a bit and bumped the ideal age range to 23.02 because it adds Alexander. Sue me. These backs have an average PPG (PPR) of 12.9 and include:

Tim Biakabutuka, J.J. Arrington, Warrick Dunn (#12 pick), Deuce McAllister, William Green, Travis Henry, Shaun Alexander, Caddy Williams (#5 pick), and Robert Edwards (#18 pick).

Now it gets a little trickier – we’re in the middle range of the model, between the ‘can’t miss’ and ‘can’t succeed’ players. This group can be meaningfully subset though, which I’ve tried to do below, but the 2s tend to be more hit or miss than the 1s or the 3s.

#2a: Backs who qualify on the ideal weight and age criteria, but were drafted a little later – those under 23.02 years old (again, I cheated by a few days to include Westbrook at 23.01), between 205 and 221 pounds, and picked between spots 77 and 109 (Ryan Moats to Marion Barber). These backs have an average PPG (PPR) of 10.4 and include:

Marion Barber, Chris Brown, Onterrio Smith, Reuben Droughns Domanick Williams, Darnell Autry, Ryan Moats, Karim Abdul-Jabbar.

Along with the 2nd and 3rd round guys in group 1a, these are the small backs that can be had for great value. Compare them to other backs taken from the middle 3rd to the middle 4th and their success rate is outstanding.

#2b: Next up are the players who qualified on draft position, pick 76 or earlier, but have a close miss on one of the other criteria – they’re either between 195 and 204 pounds with an ideal age; or are of an ideal weight (205-221), but are a bit too old – between 23.03 and 23.5 years old. The 12 backs in this group have an average PPG (PPR) of 9.1 and include:

Brian Calhoun, Maurice Morris, Byron Hanspard, Moe Williams, Tiki Barber, Ladell Betts, Michael Pittman, Julius Jones, Michael Bennett, Joseph Addai, DeAngelo Williams.

#2c: Finally, we have players who just miss on either weight or age – they’re between 195 and 204 or 23.03 and 23.5 years old - and were drafted a bit later, between picks 77 and 109. Those players have a career average of 6.1 PPG (PPR) and include:

Travis Minor, Amos Zeroeoue, Brian Westbrook, Doug Chapman, Rashaan Sheehee, Kevin Faulk, Jerious Norwood and Leon Johnson.

Unless Norwood turns out to be a stud here, Westbrook salvages the entire group and we should probably say that he's an anomaly. If so, the 2cs may as well be 3s and be avoided.

#3: Backs who were older than 23.5 when they started their career, weigh under 195 pounds or were drafted outside the top ~109. In a nutshell the only backs that can overcome a dramatic deficiency in terms of weight or age are those drafted in the first half of the 1st round: Cadillac Williams, Warrick Dunn and Robert Edwards. Otherwise, this group of players is the bust list, and has a career average PPG (PPR) of only 4.5. And it’s 20% of all RBs drafted in the first four rounds. If you did nothing but eliminate these players from your draft board you’d be way ahead in the game:

John Avery, Trung Candidate, Tatum Bell, Kenny Irons, Michael Cloud, Sedrick Shaw, Troy Davis, Travis Prentice, James Jackson, Lorenzo Booker, Vernand Morency, Winslow Oliver, Garrett Wolfe, Justin Fargas, Tavian Banks, Ricky Whittle, Antonio Pittman, Quentin Griffin, Ciatrick Fason, Lee Suggs, Leon Washington, Mewelde Moore, Travis Stephens, Chris Darkins, Olandis Gary, Alvin Pearman, Cedric Cobbs

You could subset this group of players into those that just barely missed the cutoff and those that were well beyond it if you wanted to. Those that are just outside do perform better. But why bother? Not one of the players in this list has a career PPG (PPR) of 10.0 or higher.

Part III, Big Running Backs

Big running backs, those over 221 pounds, are much simpler. Eliminate backs taken after pick ~105 or so, and also those over 240 pounds or over 23.0 years old, unless they’re chosen in the first round. Then sort simply sort by draft position.

#4a: The can’t miss version for big backs: between 222 and 240 pounds taken in the first round of the draft. With an average career PPG of 15.0 they’re every bit the fantasy gold of their smaller brethren. These backs include:

Eddie George, Fred Taylor, Ricky Williams, Jamal Lewis, Larry Johnson, Steven Jackson, Willis McGahee, Ronnie Brown, Chris Perry, Cedric Benson.

#4b: Backs under 23 years old at the start of their rookie season who weigh between 222 and 240 pounds and are taken in the 2nd, 3rd or early 4th rounds of the draft represent value – especially those taken in the 3rd and 4th rounds since there doesn’t appear to be much difference between those taken in the 2nd round and those taken later. These backs have an average PPG (PPR) of 9.3 and include:

DeShaun Foster, Corey Dillon, LaMont Jordan, Chris Henry, Duce Staley, Musa Smith, Kevan Barlow, Jerald Moore, Tony Hunt, Rudi Johnson, Stephen Davis, Sedrick Irvin, Lendale White, Heath Evans.

Note that I eliminated Lawrence Phillips and Maurice Clarette from the data set, but it wouldn’t make a material difference if they were included.

#4c: All other big backs – the bust list for big backs. I haven’t bothered calculating a PPG for these guys (lazy, see?), but it’s clear these are not guys you want on your team. Brandon Jacobs owners, beware! Notice that being drafted in the 1st round isn’t enough to save you if you’re a big back that fails to qualify otherwise. This list includes:

Ron Dayne, Eric Shelton, T.J. Duckett, George Layne, Correll Buckhalter, Curtis Enis, Chad Levitt, Jonathan Wells, Brandon Jacobs, Reggie Brown, Michael Bush, Joe Montgomery, Kantroy Barber, Michael Robinson, Brian Leonard, Sean Bennett, Anthony Thomas, Curtis Keaton, Skip Hicks, Richard Huntley, Frank Moreau, Dwayne Wright, Jermaine Fazande, James Johnson, Antwoin Smith, Artose Pinner

To summarize there are 128 RBs included in this data set, and the CAREER PPG (PPR) for backs in each group is as follows:

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%)

If you did nothing but ignore the bottom two groups in my ratings system (Group 3 and Group 4c), you’d have eliminated more than 40% of all RBs drafted in the first four rounds, with virtually no chance you’d missed someone who was fantasy startable.

I’ll include my thoughts on why I believe these criteria measure talent (which is what I’ve come to believe), and also why they’re different for small and big backs in a subsequent post. Working with this has generated a bunch of ideas, and I’ll be sure to try and put them all down thread in time. I do believe this entire model is real-world explainable with a single big idea. And to the extent that fantasy starts are relevant to the NFL, I think there's something here for NFL talent evaluators as well.

Also, I’m happy to answer big-picture questions, but if you want to spend time arguing about where a back was placed, or why I should have included injuries, or any other thing that focuses on the trees and ignores the forest, I’m not interested. As I mentioned, I’m not much on details, and I’m sure there are some I’ve missed. But I’m very confident that this moves things forward for an objective evaluation of talent in RBs.

And stay tuned for my work on WRs. I’ll try to post it sometime in the next couple weeks. The model works as well as this one (and does a great job at predicting early 1st round busts), but hinges on completely different criteria.

Interesting.

What were the other variables you looked at (pre-draft 40 time, height, . . .)?

I decided that all those combine variables were likely encapsulated in draft position, so didn't include them. To be honest, I forget what all I looked at now since age, weight and draft position were the only ones that jumped off the page when I was starting this. There definitely could be some I either didn't have access to or missed.

I do remember that height was 100% irrelevant except as it correlates to weight.

Predictions for Class of 2007, based on the above. Note that I'm not altering these for my subjective opinion, and that the overall ranking is based on the likelihood the player will become fantasy startable.

Marshawn Lynch, Brandon Jackson and Adrian Peterson

About as close to a sure thing as you can get barring injury or a multi-year collapse of the running game by the team that drafted them. No one entering the league at their age, at their weight and their draft position in the last ten years has failed for any other reason. Brandon Jackson is a great test case. As long as there are other teams that shared the Packers' assessment of his talent (i.e. they didn't overreach by more than a half round or so), the model says he should be a star.

Chris Henry and Tony Hunt

These are the best of the bigger backs in the 2007 draft, but both are 2nd tier. It's possible that either or both will be outstanding, but also possible that they will flop. Still after the big three above, they represent the best options among the 2007 rookie RBs. Note that even a full round reach by the Titans does not alter the prediction for Henry - it wouldn't change even if he'd been drafted in the early 4th. I don't like it, because I think he's a flop waiting to happen, but the model says maybe not.

Kenny Irons and Lorenzo Booker

Entering the league at nearly 24, and weighing substantially under 205 would be a likely death blow even for a first round pick. Taken at #49, Kenny Irons is all but ensured of fantasy irrelevancy over the course of his career. Booker will most likely share his fate -- he's slightly younger, but is also too light and taken even deeper in the draft, at #71.

Brian Leonard

Leonard plays at a good weight and was taken early enough in the draft to be relevant, but is too old. Approaching 24 as a rookie, he's likely to be a career backup -- with some value, but never a reliable fantasy starter.

Pittman, Antonio

Pittman enters the league very young, under 22, and at a weight that could be successful if he'd been drafted earlier. But at pick #107, it's far more likely than not that he'll never be relevant as a fantasy starter.

Wolfe, Garrett

177 pounds say it all -- no one at his weight taken anywhere near as late in the draft, #93, has gone on to be a reliable fantasy starter in the last ten years.

Dwayne Wright and Michael Bush

A special place in fantasy hell is reserved for later-drafted players over 24 years old, like Wright, and well over 240 pounds, like Bush. They're both so far away from the successful criteria that even being a first round selection would only salvage an Anthony Thomas or Ron Dayne type career. At picks #111 and #100, it's nearly certain you want neither of these players on your roster.

will revisit at some point.

[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.

Good read, lots to digest.

I think age matters for two reasons, but neither one is provable.

First, I think RBs probably develop up until, say, age 25 or so. And coming into the league at 21 or 22 gives them more time to do that at the NFL level. Maybe as much as 50% more time.

Second, and this relates the the idea I'll try and post when I have more time again, I'm hypothesizing that RBs who are small and young develop better vision early in their football careers to avoid being killed and make use of the holes. A guy that's older and bigger than his peers learns to run like a bully - he can run over people. Until he reaches the pros anyhow. Age appears to be a less relevant for bigger backs.

Brilliant assessment. And I love how you've taken the bold step of classifying the 2007 draft class, which looks to me like how it will play out over time.

I think one factor worth considering is that the young RBs that start in the NFL at 21-22 are there because they've got superior talent, have had great success in college, and enter the draft after their junior years. That is, the talent has always been there, thus they prove themselves early in college and enter the NFL as soon as they smell the payday. Not only do they get the opportunity to start their NFL development early, but one can argue they had the talent to get there early in the first place.

Most ienresting thing is that the bust list includes (if I remember correctly) a Rookie of the Year in A-Train. he's had a weird career, I wouldn't call him a bust, really, as I'd be very happy if I got that out him as a rookie, & 2nd year player, then he just bounced around.

What I do think is intersting, is that even on the bust lists you have guys who are useful, but you generally don't want them, unles you have another good reason:

Anthony Thomas started for the Bears as a rookie, put up decent numbers for a few years. (of course Enis was suppsoed to do that, too)

Brian Westbrook: slipped in draft due to 1AA school, & injuries, but was freakish in college.

You missed a group I think is interesting: UDFA's that make rosters. This includes a bunch of no-names, but also: Dominic Rhodes, Mike Anderson, Willie Parker (I think), etc. I actually would think that a UDFA would have a better shot than a late pick, since they've got to show a ton to beat out a pick, and hopeufully they've picked a good situation for themselves. thoughts?

Asinine said:

I think one factor worth considering is that the young RBs that start in the NFL at 21-22 are there because they've got superior talent, have had great success in college, and enter the draft after their junior years. That is, the talent has always been there, thus they prove themselves early in college and enter the NFL as soon as they smell the payday. Not only do they get the opportunity to start their NFL development early, but one can argue they had the talent to get there early in the first place.

Click to expand...

Great stuff. I'd read everything over at FootballOutsiders.com regarding predicting QB success, and was also wondering what sort of data and outcomes could be produced from looking at the other skill positions.

I'm curious though, did you look at games started in college (like FO.com did with QBs), or even college stats of any sort, as part of your trial and error process to narrow down the relevant criteria? If so, did you find anything useful from that?

I'm guessing that games started is far more important in the QBs, since experience plays a bigger role in the position typically.

Looking forward to the WR data!

Re: UDFAs -- no model's going to catch everyone, and it's super clear if you line the data up right that after the first part of the 4th round you might as well forget about it. There are always going to be Willie Parkers and Terrell Davises that come out of nowhere. So I only looked at RBs taken in the 1st four rounds.

One other thing I didn't highlight enough above... if I'd had more data I'd have tried to quantify the benefit of being taken in the first part of the first round. The fact that a player as small as Dunn and as old as Robert Edwards had success speaks volumes. But that early 1st round benefit gets muddled because of the inefficiency in the draft for players in my #1a group - who should have been first rounders.

btw... there's a catch-22 here. If NFL talent evaluators ever quit believing their lying eyes and adopted something like this or the FB Outsider's QB system, the system would stop working as well because they'd start drafting guys that fit the criteria earlier.

Really looking forward to seeing where Brandon Jackson in a couple years. He's what Phillip Rivers was to the FO QB model for this system.

wdcrob said:

I think age matters for two reasons, but neither one is provable.First, I think RBs probably develop up until, say, age 25 or so. And coming into the league at 21 or 22 gives them more time to do that at the NFL level. Maybe as much as 50% more time.Second, and this relates the the idea I'll try and post when I have more time again, I'm hypothesizing that RBs who are small and young develop better vision early in their football careers to avoid being killed and make use of the holes. A guy that's older and bigger than his peers learns to run like a bully - he can run over people. Until he reaches the pros anyhow. Age appears to be a less relevant for bigger backs.

Click to expand...

These are very interesting thoughts. The second one especially makes a lot of sense. Excellent job taking the time and effort required to do this, and coming up with some unique thoughts about the RBs.

It's interesting...and I certainly can't argue with what seem to be strong correlations. But I'm disturbed by any analysis or predictor of success that relies so little on past production. I suppose production factors into draft position...but I'd want something more explicit. Something that will tell me RIGHT NOW whether Chris Henry, for example, is a stud or a bust. He'll be an interesting "test case" of this system, to be sure -- totally unproductive in college, yet has the "measurables" that make him a promising pick by this system.

When I evaluate rookie RBs, I look at three main factors:

1. Production in college

2. Combine numbers

3. Draft position

It's relatively rare for a first round RB with great college production and combine numbers to become a bust. It happens, but it's rare.

I also like my RBs to be 5'9"-6'1" and 200-235 pounds. Most of the top RBs in the past fifteen years have fit this physical mold.

I don't think age is relevant. As a previous poster mentioned, the reason age might appear to be related to success is probably because almost all of the top RB prospects leave college after their junior season.

wdcrob said:

Part II, Small Running Backs

#1a: These are the cream of the crop for small backs – not just within the required ranges, but squarely on the good side of the range: backs between 205 and 221 pounds, under 22.5 years old, drafted in first 76 picks; and other back drafted in the first ~half of the first round under 22.5 years old. These backs are fantasy gold, with a career average of 16.2 PPG (PPR). They include the following RBs:

Clinton Portis, Edgerrin James, Maurice Jones-Drew, Marshawn Lynch, Reggie Bush (#2 pick), Ahman Green, Laurence Maroney, Brandon Jackson, Kevin Jones, Thomas Jones, Jay Graham, LaDanian Tomlinson, Frank Gore, Adrian Peterson

Click to expand...

Not sure what the bolded part is implying. Is this actually two different sets of criteria? For instance, are you saying that small running backs are in tier 1a if they are either:(a) 205-221 pounds, under 22.5 years old, and drafted in the first 76 picks

or

(b) less than 221 pounds, under 22.5 years old, and drafted in the first half of round 1 (basically higher drafted but potentially lighter)

Interesting stuff.

EBF said:

When I evaluate rookie RBs, I look at three main factors:

1. Production in college

2. Combine numbers

3. Draft position

It's relatively rare for a first round RB with great college production and combine numbers to become a bust. It happens, but it's rare.

Click to expand...

wouldn't there be a correlation between combine numbers & college production with draft position? Seems to me that draft position is a function of the other two; I think that's what the OP is accounting for - draft position is a catch-all type variable.

EBF said:

I also like my RBs to be 5'9"-6'1" and 200-235 pounds. Most of the top RBs in the past fifteen years have fit this physical mold.

I don't think age is relevant. As a previous poster mentioned, the reason age might appear to be related to success is probably because almost all of the top RB prospects leave college after their junior season.

Click to expand...

Don't you think that the bolded section is significant on it's own?ETA: to wdcrob. one of the best posts in the spool in years; looking forward to more.

moleculo said:

wouldn't there be a correlation between combine numbers & college production with draft position? Seems to me that draft position is a function of the other two; I think that's what the OP is accounting for - draft position is a catch-all type variable.

Click to expand...

Yes and no. Do good combine numbers help a prospect's draft stock? Yes.Do good statistics help a prospect's draft stock? Yes. But not every top RB prospect is great in these categories. Cedric Benson and Frank Gore bombed in workouts. Ronnie Brown, Joseph Addai, and Chris Henry had marginal production in college. So you can definitely find examples of high draft picks who didn't have great combine numbers and/or great college production. The reason I look for prospects who are elite in all three areas (draft position, combine, and college production) is because they're as close to bulletproof as you can get. They have the physical ability to play on Sundays. They dominated in college. And they were good enough to convince a pro franchise to use an early pick on them. You can't ask for much more than that.

No doubt part of the age function is the guys who come out early. But I'm 90% sure that there's still an effect for guys that play all four years. I'll try and gather up the underclassmen that declared for the draft and take them out of the analysis to see. Again, I'm pretty sure that 22 is always better than 24, regardless of how many years of college ball were played. But I haven't checked that specifically yet.

Q-Bert said:

Not sure what the bolded part is implying. Is this actually two different sets of criteria? For instance, are you saying that small running backs are in tier 1a if they are either:(a) 205-221 pounds, under 22.5 years old, and drafted in the first 76 picksor(b) less than 221 pounds, under 22.5 years old, and drafted in the first half of round 1 (basically higher drafted but potentially lighter)

Click to expand...

That's right. As of right now (b) only adds Reggie Bush though. Bush and Dunn make the point that you can be lighter if your college production and combine #s (which I do indeed maintain are captured by draft position) are strong enough to get you into the top part of the 1st round. This likely has relevance for Steve Slaton next year.There are several places where I wasn't clear and things need to be cleaned up. I'll try and edit soon.

Any significance to the order that you listed the RBs? For example, in both the original post and in the '07 rookie prediction post you had the three main backs listed as Lynch, Jackson and Peterson, in that order. Found it interesting.

Obviously, this isn't causing me to re-think my rankings or anything like that, but I was just curious if the listing was something other than random, like Lynch's measurables are better "on paper" than Peterson's or Jackson's.

No, I just grabbed them out of my spreadsheet in whatever order they happened to be sorted at that moment. No rhyme or reason to it.

EBF said:

moleculo said:

wouldn't there be a correlation between combine numbers & college production with draft position? Seems to me that draft position is a function of the other two; I think that's what the OP is accounting for - draft position is a catch-all type variable.

Click to expand...

Yes and no. Do good combine numbers help a prospect's draft stock? Yes.Do good statistics help a prospect's draft stock? Yes. But not every top RB prospect is great in these categories. Cedric Benson and Frank Gore bombed in workouts. Ronnie Brown, Joseph Addai, and Chris Henry had marginal production in college. So you can definitely find examples of high draft picks who didn't have great combine numbers and/or great college production. The reason I look for prospects who are elite in all three areas (draft position, combine, and college production) is because they're as close to bulletproof as you can get. They have the physical ability to play on Sundays. They dominated in college. And they were good enough to convince a pro franchise to use an early pick on them. You can't ask for much more than that.

Click to expand...

Now all we need is a way to figure out which ones are whacked out in the head...call it the Phillips/Williams test...

Hobbes said:

EBF said:

moleculo said:

wouldn't there be a correlation between combine numbers & college production with draft position? Seems to me that draft position is a function of the other two; I think that's what the OP is accounting for - draft position is a catch-all type variable.

Click to expand...

Yes and no. Do good combine numbers help a prospect's draft stock? Yes.Do good statistics help a prospect's draft stock? Yes. But not every top RB prospect is great in these categories. Cedric Benson and Frank Gore bombed in workouts. Ronnie Brown, Joseph Addai, and Chris Henry had marginal production in college. So you can definitely find examples of high draft picks who didn't have great combine numbers and/or great college production. The reason I look for prospects who are elite in all three areas (draft position, combine, and college production) is because they're as close to bulletproof as you can get. They have the physical ability to play on Sundays. They dominated in college. And they were good enough to convince a pro franchise to use an early pick on them. You can't ask for much more than that.

Click to expand...

Now all we need is a way to figure out which ones are whacked out in the head...call it the Phillips/Williams test...

Click to expand...

I'm gonna go out on a limb here and say that dragging your girlfriend down a flight of stairs by her hair isn't typically a sign of outstanding moral fiber.

Bush and Dunn make the point that you can be lighter if your college production and combine #s (which I do indeed maintain are captured by draft position) are strong enough to get you into the top part of the 1st round.

Click to expand...

I don't think you can lump combine performance and draft position together into one variable. They're two very different things.Reggie Bush and Ronnie Brown were both chosen at 1.02. Bush's combine numbers blow Brown's out of the water.Adrian Peterson was chosen within three picks of where Cedric Benson and Cadillac Williams went. His numbers blow theirs out of the water.I'll agree with the general statement that players who do well at the combine tend to get drafted higher and vice versa, but it remains true that there's a lot of variance in the workout numbers of the RBs chosen in the same round.

You definitely could be right EBF -- and it's also possible that age and weight are factored into draft position as well, especially weight. I'm hoping someone out there with a great database, SQL skill and some time can take this and run with it.

Basically turn age, weight and draft position into binary variables using the ranges I've put out there and play with it. I'm very happy with the initial findings here and the categories seem to work well, but I know there's more out there to find.

Fascinating stuff! Thanks.

I'm curious where SD's Michael Turner falls in your grids.

Good question...I don't remember Turner in the data set. Was he drafted in the 1st four rounds? If not, he wouldn't be in here. If so, I just missed him somehow. I'm guessing he was a later pick since I was in and out of the draft data a hundred times.

Also, a general note -- remember these are DYNASTY predictions and my PPG totals are career averages. There's no reason some of these guys couldn't have a good season or two. What's interesting though is how many of the borderline guys like Betts or Michael Bennett seem to tease, but somehow never quite break through.

btw...it's a shame Ryan Moats got hurt. I'd have bet money he eventually got a chance somewhere and did well. The model likes him a lot.

And while Addai's in a perfect situation, he probably isn't a very talented back. It's hard to imagine he'll ever be bumped, but he's definitely replaceable if the Colts wanted to let him go rather than pay him big money in a couple years. And if the model's right, on another team he might not be very effective.

wdcrob said:

Good question...I don't remember Turner in the data set. Was he drafted in the 1st four rounds? If not, he wouldn't be in here. If so, I just missed him somehow. I'm guessing he was a later pick since I was in and out of the draft data a hundred times.Also, a general note -- remember these are DYNASTY predictions and my PPG totals are career averages. There's no reason some of these guys couldn't have a good season or two. What's interesting though is how many of the borderline guys like Betts or Michael Bennett seem to tease, but somehow never quite break through. btw...it's a shame Ryan Moats got hurt. I'd have bet money he eventually got a chance somewhere and did well. The model likes him a lot.And while Addai's in a perfect situation, he probably isn't a very talented back. It's hard to imagine he'll ever be bumped, but he's definitely replaceable if the Colts wanted to let him go rather than pay him big money in a couple years. And if the model's right, on another team he might not be very effective.

Click to expand...

Turner was drafted in Rd. 5 in 2004

wdcrob said:

You definitely could be right EBF -- and it's also possible that age and weight are factored into draft position as well, especially weight. I'm hoping someone out there with a great database, SQL skill and some time can take this and run with it. Basically turn age, weight and draft position into binary variables using the ranges I've put out there and play with it. I'm very happy with the initial findings here and the categories seem to work well, but I know there's more out there to find.

Click to expand...

do you know what you are looking to do on the db side? I might be able to put in a little time and help out.what do you have data for? how far back? what format is it in?assuming from your posts that you haveplayers ff pointsdraft positiondraft height/weightanything else?

Asinine said:

Brilliant assessment. And I love how you've taken the bold step of classifying the 2007 draft class, which looks to me like how it will play out over time.I think one factor worth considering is that the young RBs that start in the NFL at 21-22 are there because they've got superior talent, have had great success in college, and enter the draft after their junior years. That is, the talent has always been there, thus they prove themselves early in college and enter the NFL as soon as they smell the payday. Not only do they get the opportunity to start their NFL development early, but one can argue they had the talent to get there early in the first place.

Click to expand...

Ok, this makes a lot of sense to me.

Nookular Power said:

It's interesting...and I certainly can't argue with what seem to be strong correlations. But I'm disturbed by any analysis or predictor of success that relies so little on past production. I suppose production factors into draft position...but I'd want something more explicit. Something that will tell me RIGHT NOW whether Chris Henry, for example, is a stud or a bust. He'll be an interesting "test case" of this system, to be sure -- totally unproductive in college, yet has the "measurables" that make him a promising pick by this system.

Click to expand...

I agree that so little regard for college performance scares me also and seems counter-intuitive. But, when you think about it, how objective is college performance? Think of how different the competition can be. Think about how different the offenses can be and how different the college offenses in general are from the pros. Think about opportunity (a guy like Caddy and Brown sharing carries for example). And, in some indirect way college performance is a factor because it likely affects how early a guy comes out and how high he is drafted.

Nice.

Wish I had read something like this last year before I got all hyped on DeAngelo over Maroney...

:JasonSehorn:

Good job. Fun read and food for thought. Appreciate the time and effort.

Great stuff Rob! Always wanted to look into this more, just didn't have the time.

One thing I can add is that I wrote an article for NFLDraftGuys last season that looked at Fantasy Point Distribution by NFL Draft Round (here) and found the same conclusion in terms of draft round: 1st round is best, 2nd and 3rd are about even, and things really dry up by 4th round, other than some outliers that were undrafted like Holmes and Parker.

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.

wdcrob said:

No, I just grabbed them out of my spreadsheet in whatever order they happened to be sorted at that moment. No rhyme or reason to it.

Click to expand...

Great analysis. Lots of thought and I am sure time invested in trying to come up with a viable way to rate rookies. I love this because there does appear to be a pattern in this article. This is the first time I have thought along these lines but yuo can be sure that I am going to follow these predictions through the next few years and might just use it to help my dynasty league roster. Thanks for the effort.

Interesting analysis.

I think durability should also be considered here. Although predicting injury is problematic at best, I have observed that certain body types tend to handle the punishment of the NFL better than others.

Runners with a body mass distribution having a higher ratio of mass in the lower body (hips/thighs) seem to be more durable than other backs with a more evenly distributed body mass. This however is just an observation and I have no statistics to back it up.

What's the ETA on the WR article? Can't wait to check that out and see the data.

Big pool of promising rookie receivers this year, and I'm really curious to find out which ones the model predicts will be successful.

out-freakin'-standing post.

One of the best threads I've read here in ages.

I anxiously look forward to the WR thread.

thanks for the effort. Your hard work shines through!

Since a couple people have asked...I'll pull my data together into a single place and clean it up for you before I forward. But that may be a week or so. And the WR stuff is basically done, so I'll try and get that up as soon as I can too.

.

I'll analyze this one later.

wdcrob said:

Since a couple people have asked...I'll pull my data together into a single place and clean it up for you before I forward. But that may be a week or so. And the WR stuff is basically done, so I'll try and get that up as soon as I can too.

Click to expand...

Thanks, much appreciated.And, very much looking forward to the WR post.