2017 Running Back Report

[lukkynumber13]

I've read through this entire conversation, and really appreciate DD's work, and all the effort he has put into this. Reading all the comments/criticisms, I wonder if people are talking past each other a bit, so I will ask Dynasty Delorian this question:
Does your analysis lead you to believe that CMac is unlikely to be an effective RB for Fantasy Football, PPR purposes (top 15 PPR as a threshhold), or are you only looking at his likelihood of being an effective "workhorse/traditional" running back?

Excellent question!

[clarion contrarion]

so is the model affected by the fact that most of hunt's production came as a 233 lb pounder as opposed to to 209 or 214 lb that he weighed at the combine ? jersey scouting always gets killed here but his YPC dipped from 8.0 down to 5.6 from his soph year to his senior year.
david fluellen put up similar # from the same school and has done zilch in the league !
hunt 5'10" 216 4.62(40) 119" broad 36.5 vert 4945 career rush yards 6.3 YPC 3548 total yards 27 tds in 2 best years
fluellen 5'11" 224 4.72(40) 120" broad 36.5 vert 3361 career rush yards 5.9 YPC 3087 total yards 23 tds in 2 best years


even former rocket adonis thomas had 2 1000 yard seasons with over 6.0 YPC at toledo 2168 yards of total offense and 23tds in 2010 & 2011 - hunt screams system back from an inferior conference regardless of what the model projects .
just saying !

[juke05]

This is good stuff, thanks for doing this DD! Just curious - if someone is very close to one of the top categories (like stud, fake bellcow, or flash in the pan) but maybe missing one metric or something, would that make them a better prospect than someone that actually qualifies for one of the lower categories (like dart throw)? Like Perine for example - you say he was close to "flash in the pan", wouldn't he kinda rank ahead of say kareem hunt who is a "dart throw"? Or is that not how your system works? I guess it just seems odd to put jones ahead of cook/CMC when I'm sure cook/CMC were closer to the top categories than the dart throws, etc., but I guess if you're ranking in terms of value, then I suppose it makes sense. Either way, I ended up w/ Fournette and Foreman, so hoping you're right!

[Dynasty DeLorean]

Does your analysis lead you to believe that CMac is unlikely to be an effective RB for Fantasy Football, PPR purposes (top 15 PPR as a threshhold), or are you only looking at his likelihood of being an effective "workhorse/traditional" running back?

He isn't identified as either in my work here, but the bulk of my analysis is geared towards ID'ing "workhorse/traditional" running backs unless stated otherwise. That doesn't mean he can't be either, just that there isn't any measurable way to predict he will be. I suppose the simplest way I can put it is that he is a "bad bet" for his price point. Statistically speaking you would be better off taking Foreman at 1.3 instead of CMC there. From a more general standpoint, 1st round rbs are typically more talented, will get more opportunities and be more productive than their later round counterparts. So he does have draft pedigree in his favor which can lead to a high level of production even if sometimes it's not warranted.

I linked this earlier but another poster here does good work and he ID's CMC as being productive.

if someone is very close to one of the top categories (like stud, fake bellcow, or flash in the pan) but maybe missing one metric or something, would that make them a better prospect than someone that actually qualifies for one of the lower categories (like dart throw)? Like Perine for example - you say he was close to "flash in the pan", wouldn't he kinda rank ahead of say kareem hunt who is a "dart throw"? Or is that not how your system works?

Yeah that's not really how it works. The only thing I'd say is maybe it makes a guy like Perine a little more interesting than a guy of similar value who wasn't close to any category.

I guess it just seems odd to put jones ahead of cook/CMC when I'm sure cook/CMC were closer to the top categories than the dart throws, etc., but I guess if you're ranking in terms of value, then I suppose it makes sense. Either way, I ended up w/ Fournette and Foreman, so hoping you're right!

Not understanding the "jones ahead of cook/CMC" question. You mean Aaron Jones? I don't believe I said that.

so is the model affected by the fact that most of hunt's production came as a 233 lb pounder as opposed to to 209 or 214 lb that he weighed at the combine ? jersey scouting always gets killed here but his YPC dipped from 8.0 down to 5.6 from his soph year to his senior year.
david fluellen put up similar # from the same school and has done zilch in the league !
hunt 5'10" 216 4.62(40) 119" broad 36.5 vert 4945 career rush yards 6.3 YPC 3548 total yards 27 tds in 2 best years
fluellen 5'11" 224 4.72(40) 120" broad 36.5 vert 3361 career rush yards 5.9 YPC 3087 total yards 23 tds in 2 best years


even former rocket adonis thomas had 2 1000 yard seasons with over 6.0 YPC at toledo 2168 yards of total offense and 23tds in 2010 & 2011 - hunt screams system back from an inferior conference regardless of what the model projects .
just saying !

The best response I can give you is that's not really how any of this works.

As for Hunt specifically, I'm not super high on him or anything. All I said is that statistically speaking he's more likely than not to rush for at least one 1,000 yard season. He does look like a solid well-rounded RB. In my opinion he'd be a great investment for contending Ware owners, and probably a better bet than most of the Rb's being drafted in the late-1st or 2nd round of rookie drafts. If there were to be a Jordan Howard or Thomas Rawls this year it would probably be Kareem Hunt.

[juke05]

I guess it just seems odd to put jones ahead of cook/CMC when I'm sure cook/CMC were closer to the top categories than the dart throws, etc., but I guess if you're ranking in terms of value, then I suppose it makes sense. Either way, I ended up w/ Fournette and Foreman, so hoping you're right!

Not understanding the "jones ahead of cook/CMC" question. You mean Aaron Jones? I don't believe I said that.

Yeah that was my bad...for some reason I thought you had said at the very end that you'd rank the RBs as 1) Fournette 2) Foreman 3) Mixon 4) Aaron Jones, but you actually had Hunt at #4.

[ImaRounder]

Correct me if I'm wrong , but I thought somewhere in here that the price of each RB came into some equation for DD's list. If that's the case, I can totally understand your value based rankings and I think it's clear up this entire mcaffrey debacle.

*If it has nothing to do with their price, I'll go f*** off and apologize for wasting your time

[Dynasty DeLorean]

Correct me if I'm wrong , but I thought somewhere in here that the price of each RB came into some equation for DD's list. If that's the case, I can totally understand your value based rankings and I think it's clear up this entire mcaffrey debacle.

1) I'm not touching CMC for his price, so where I would have him officially ranked among all rookie Rb's is completely pointless.
2) Assuming I was being forced to acquire a RB and the acquisition cost is equal for each RB, I would rather select Fournette, Foreman, and Mixon in that order over CMC.

[Concept Coop]

However from this report, we know the odds of Fournette and/or Foreman hitting (being very good/great) are at least 80%, with a bust rate close to 0%.

No we don't. Do you realize the magnitiude of the claims you're making? The NFL, with all of its resources--including billions of dollars and the best minds that that money can buy--let D'onta Foreman slide to the 3rd round, where long-term starters account for 16% of the backs selected. Yet you think you've developed a system that promises an 80% hit rate? And you can't tell us anything substantial about it or represent your findings numerically?

[lukkynumber13]

However from this report, we know the odds of Fournette and/or Foreman hitting (being very good/great) are at least 80%, with a bust rate close to 0%.

No we don't. Do you realize the magnitiude of the claims you're making? The NFL, with all of its resources--including billions of dollars and the best minds that that money can buy--let D'onta Foreman slide to the 3rd round, where long-term starters account for 16% of the backs selected. Yet you think you've developed a system that promises an 80% hit rate? And you can't tell us anything substantial about it or represent your findings numerically?

I gotta admit, claiming that Foreman is basically guaranteed to be even an above average RB, seems pretty foolhardy.

[Dynasty DeLorean]

This is exactly why I don't provide details.

[AussieMate]

I think people are looking at this the wrong way. DD has worked a formula. Foreman landed near the top. DD didn't just put him there because he felt like it, he also didn't remove him because of his draft pedigree. He may fail in the NFL and that will alter the %of hit rate in future and not the other way round. I don't think DD is personally claiming there is a 0% chance of foreman busting, only that he landed in that zone in the formula. he has done the maths and the players landed where they landed, now we wait and see. Cmac also fell out because he doesn't fit the formula well and not because DD is a cmac hater. This isn't personal it's statistics.

[AussieMate]

Correct me if I'm wrong though, this is just how I interpret it.

[Jason3123]

Foreman has now landed in the top tier of DD's formula, Rotoviz workhorse model, UTH's model. If he does bust, he will be an outlier.

[Friction]

I will prob take a stab on him in at least one league where I was not planning on it. My drafts are not until July unfortunately and a lot can happen, including who others pick and at what slot, but I will have some faith. I ignored DJ a couple years ago and that bit me (drafted Ajayi instead), but I also nabbed Henry at 1.07 instead of my original plan of trading out last year. Don't remember exact details, I just know I feel a lot better about rostering him than option"B".

[ericanadian]

However from this report, we know the odds of Fournette and/or Foreman hitting (being very good/great) are at least 80%, with a bust rate close to 0%.

No we don't. Do you realize the magnitiude of the claims you're making? The NFL, with all of its resources--including billions of dollars and the best minds that that money can buy--let D'onta Foreman slide to the 3rd round, where long-term starters account for 16% of the backs selected. Yet you think you've developed a system that promises an 80% hit rate? And you can't tell us anything substantial about it or represent your findings numerically?

How many of those billions do you think are being spent on developing statistical models? The NFL strikes me as being in the same situation as baseball was before the statistical revolution and the reaction to the hiring of Paul DePodesta seems to back that up. From a stats standpoint, the obvious issue with the data is that the sample size is rather small. DD mentions 370 (or something around that number) backs being used, but how many of those are missing combine numbers, or other info, that are integral to the system? If you're throwing around contracts worth millions of dollars, you're going to trust the scouts and football guys you hired and know and trust over a statistical model with such glaring holes.

For the fantasy footballer, we have no real way of determining which scouting departments are effective and which ones are not. We don't know them personally and the sample sizes are too small to be relevant because we don't have their full board to work with, only the picks they actually make. The draft as a whole shows significant correlation between round drafted and NFL success, so obviously as a whole, NFL scouts are significantly better than throwing darts. This is often used as justification to use draft round as the best KPI for fantasy football production and it probably is your best baseline. However, after that, you're left with a bunch of statistically insignificant options or running the "eye test" and scouting yourself.

Most people don't have the time to thoroughly scout a class and would suck at it even if they tried. Hence, some of us look to small sample statistical models to try to gain an edge. RotoViz is pumping these out en masse. Why people get so bent out of shape over this is beyond me. Don't like a model? Don't use it. I've seen DDs original post where he walked everyone through the math involved. It makes sense to me. My guess is he stopped posting that detail because people feel entitled to answers like he's submitted a scientific thesis for peer review. Such arguments turn this from a fun statistical analysis into a never ending argument.