2019 Running Back Report

[Dynasty DeLorean]

What does the model base things on? Do you think Penny and Foreman will still be studs? Maybe buy low time.

So basically I looked at past studs to see what they had in common and then project forward. As for Foreman and Penny, I think so. Obviously how Foreman bounces back from his injury will be big. It all depends on how you define "stud" though, for example DMC got his 2nd 1k yard rushing season his 8th year in the league. Ryan Mathews career was kind of disappointing overall, but he still got the minimum of 2 1k yard rushing seasons. Derrick Henry is on his way to checking the 1k seasons benchmarks but he hasn't really been all that useful for the better part of 3 seasons now. But yes, I do think it's buy low time on both of them.

Thanks for the report! As someone else posted, would all purpose yards be a better measure-not sure if you can calculate for receptions/receiving yards?

Thanks, but don't think that would be possible. Not to the degree of accuracy i'm looking for at least.

[Flgatorguy87]

Armstead in the 4th and Barnes for free. Good report for me.

What about Mattison?

[Cult of Dionysus]

BTW, not sure if an earlier post of mine was deleted, but I think it is absolutely silly that you can add CMC, Lesean McCoy and Leveon Bell to your Studs list, years after they've entered the league, just by "tweaking" some of your report parameters. You panned CMC pretty damn hard back in 2017. To slip him with a mere aside, just doesn't sit well with me.

For those who don't remember, here's your 2017 take on CMC:

Christian McCaffery
Hesitant between the tackles when there is traffic, runs with little power. Good player in space. Ok, I’m just going to stop here because [Hot Take Incoming!] I’m not even convinced he’ll be a RB in the NFL. How many times have we heard “I think CMC was the best WR at the combine!” And you know what, maybe he was. Maybe he’s a WR. I don’t pay much attention to the bench press, but there does seem to be some minimum level of core strength that you need to have in order to be a successful RB in the NFL. I won’t go into all the details, but I will say this. CMC put up a laughably bad 10 bench reps at the combine. Out of a list of 370 players, CMC did worse than everyone except Trey Watts and Donnel Pumphrey. CMC is in the 0th percentile for bench press. For reference, the 13 bench rep mark is the first time you see an actual real live productive NFL RB appear. I can envision a scenario in which CMC is not even labeled as a RB in a few years. Bottom line is, I’m not even sure CMC is a RB so he’s essentially out of my jurisdiction.

Best Comp – Taiwan Jones

The dude average 5 ypc on 218 carries in 2018, for team without a lot of other weapons and without another meaningful RB.

At the end of the day, as I've noted before, you basically present a list of bona fide Tier 1 talent, without any metrics and without any misses, and state that your proprietary formula has generated that list. And through some "minor tweaks" you are able to expand the list of Tier 1 talent, without any of the existing talent falling off or generating any additional misses. There's an adage for this type of thing, it begins with "If its too good to be true..."

If your formula really did predict these talents, and had no misses as you submit, then you wouldn't be on here asking for Donations. You'd be sitting in the front office of an NFL team, any of which would give you a hefty contract for developing a model that can do what none of their scouting/analytics departments has been able to do.

The great thing about this is that, as the years go by, we will eventually get a clear picture of who is right and who is wrong. I won't make any conclusions at this point, but I am looking forward to 2021-2023... So I will shut up and wait for time to go by.

[dynastyninja]

To me the willingness to change and modify the "formula" adds credibility. Who gives a damn if he's right or wrong? Just keep trying to build the most accurate model possible as an alternative to the eye test, which is flawed and not that reliable even for NFL-level people.

DD could be stubborn and stick to the original formula for fear of backlash, or he could continue to tweak and modify the model to be more accurate and have to suffer through the idiots as a result. Props to him for doing the latter.

I've long considered myself a non-believer in this model, but I do think there can be something to it as a predictive tool. What I'd hate to see is a stubborn unwillingness to change it in any way for fear of being "wrong."

[Dynasty DeLorean]

BTW, not sure if an earlier post of mine was deleted, but I think it is absolutely silly that you can add CMC, Lesean McCoy and Leveon Bell to your Studs list, years after they've entered the league, just by "tweaking" some of your report parameters. You panned CMC pretty damn hard back in 2017. To slip him with a mere aside, just doesn't sit well with me.

At the end of the day, as I've noted before, you basically present a list of bona fide Tier 1 talent, without any metrics and without any misses, and state that your proprietary formula has generated that list. And through some "minor tweaks" you are able to expand the list of Tier 1 talent, without any of the existing talent falling off or generating any additional misses. There's an adage for this type of thing, it begins with "If its too good to be true..."

If your formula really did predict these talents, and had no misses as you submit, then you wouldn't be on here asking for Donations. You'd be sitting in the front office of an NFL team, any of which would give you a hefty contract for developing a model that can do what none of their scouting/analytics departments has been able to do.

The great thing about this is that, as the years go by, we will eventually get a clear picture of who is right and who is wrong. I won't make any conclusions at this point, but I am looking forward to 2021-2023... So I will shut up and wait for time to go by.

The short answer is 5 years ago I started with a list of less than 200 players with a lot of incorrect data and now I have a database that includes 500 players with data that's all verified by myself. As for any other questions you may have, ask in a clear and concise manner and I'm always happy to answer.

To me the willingness to change and modify the "formula" adds credibility. Who gives a damn if he's right or wrong? Just keep trying to build the most accurate model possible as an alternative to the eye test, which is flawed and not that reliable even for NFL-level people.

DD could be stubborn and stick to the original formula for fear of backlash, or he could continue to tweak and modify the model to be more accurate and have to suffer through the idiots as a result. Props to him for doing the latter.

I've long considered myself a non-believer in this model, but I do think there can be something to it as a predictive tool. What I'd hate to see is a stubborn unwillingness to change it in any way for fear of being "wrong."

I appreciate the sentiment but I really want to stop this perception that I'm constantly changing things when i'm not. It actually would be a problem if i was totally overhauling things every year.

[lukkynumber13]

I don’t think it would be a bad idea to put a disclaimer next to the CMCs and McCoy’s of the world, something to the effect of “my original model did NOT predict success for this player, but I’ve since adapted my model slightly due to the massive success of player X”

Just my thoughts.

[Dynasty DeLorean]

I don’t think it would be a bad idea to put a disclaimer next to the CMCs and McCoy’s of the world, something to the effect of “my original model did NOT predict success for this player, but I’ve since adapted my model slightly due to the massive success of player X”

Just my thoughts.

That's not why CMC was included though.

[kamihamster]

So what were the change(s) you made that made your model better?

[Dynasty DeLorean]

So what were the change(s) you made that made your model better?

I raised the requirement for one criteria which made another criteria irrelevant. I was pretty psyched. Been puzzling over that problem for nearly 5 years now.

[petebusey]

@Dynasty DeLorean
I found it interesting that Warring was 1 of only 3 TEs on your “hit” list along with the highly touted Hock and Fant.

Can you speak any more on Warring?

[Dynasty DeLorean]

@Dynasty DeLorean
I found it interesting that Warring was 1 of only 3 TEs on your “hit” list along with the highly touted Hock and Fant.

Can you speak any more on Warring?

Former basketball player! Need I say more?

[lukkynumber13]

I don’t think it would be a bad idea to put a disclaimer next to the CMCs and McCoy’s of the world, something to the effect of “my original model did NOT predict success for this player, but I’ve since adapted my model slightly due to the massive success of player X”

Just my thoughts.

That's not why CMC was included though.

Sure, I get that, and that’s fine.

I’m just brainstorming. I feel like a disclaimer explaining that you’re not just adding former outliers (who turned out to be studs) willy-nilly cousin help bring credibility.

[Dynasty DeLorean]

I don’t think it would be a bad idea to put a disclaimer next to the CMCs and McCoy’s of the world, something to the effect of “my original model did NOT predict success for this player, but I’ve since adapted my model slightly due to the massive success of player X”

Just my thoughts.

That's not why CMC was included though.

Sure, I get that, and that’s fine.

I’m just brainstorming. I feel like a disclaimer explaining that you’re not just adding former outliers (who turned out to be studs) willy-nilly cousin help bring credibility.

I thought the rather large paragraph explaining everything was kind of the disclaimer?

[rubber_duck]

Thanks for posting this DD. In reviewing this year's class of RBs, I've gotten very similar results to yours.

This seems like the absolutely perfect year for a fantasy team to sell off the 1.01 or 1.02 rookie pick if they have them.

[kamihamster]

Sounds like you're doing a Multireg Method for Ad-Hoc Analysis. As the NFL game changes do you think that your model will become outdated? It would explain why changing one aspect of your analysis was able to explain guys you couldn't before. Are you using statistical software to do this, or you just using a spreadsheet?