2019 Running Back Report

[alewilliam789]

I tgink with all these statistical models it’s more created to give you higher odds at hitting and also weeding out really clear “fakers/fake studs”.....Montgomery.....

But otherwise if you use the system for what it is, a tool to help you make decisions, and not the be-all-end-all sorta deal then you’ll be fine.

Also if you have a system that has shown success, and if there are revisions to it that help you identify more hits without effecting your hit rate, why wouldn’t you make that change?

I mean I’m working on a similar system for WRs and I had one change that helped me identify like 6+ more “stud” with late breakout ages. That includes Michael Thomas. But with systems like these which include noticing statistical patterns, you often find that there are more and more variables that become important/relevant as you continue to build it that you wouldn’t have even thought of before

[Cameron Giles]

I tgink with all these statistical models it’s more created to give you higher odds at hitting and also weeding out really clear “fakers/fake studs”.....Montgomery.....

But otherwise if you use the system for what it is, a tool to help you make decisions, and not the be-all-end-all sorta deal then you’ll be fine.

Also if you have a system that has shown success, and if there are revisions to it that help you identify more hits without effecting your hit rate, why wouldn’t you make that change?

I mean I’m working on a similar system for WRs and I had one change that helped me identify like 6+ more “stud” with late breakout ages. That includes Michael Thomas. But with systems like these which include noticing statistical patterns, you often find that there are more and more variables that become important/relevant as you continue to build it that you wouldn’t have even thought of before

I think a lot of these formulas include an incredible amount of hindsight and are revised to include more hindsight. I understand why that is, because new information is acquired, but it's hard to label that as "predictive."

It essentially implies that there's an underlying mathematical reason that Michael Thomas is really good and we need to find out what it is so we can recognize the next Michael Thomas (there isn't).

Rotoviz, which is (or used to be) a prominent website for advanced fantasy metrics consistently said that Michael Thomas would be a bust. They had formulas and everything to support it. I used to follow their Phenom Index as well, which was touted as their highest predictor of WR success and last I checked, it hasn't been very good at it. It would be odd now for them to turn around and create a WR formula based on Thomas. Just like it would be odd to create some UDFA WR formula based on Adam Thielen.

I'll always read what people put up in regards to this stuff, but especially for the WR position, I don't think you can boil things like route running, releases, leverages and separation vs. press and man to a predictive formula. Those things are a lot more predictive to success for me and a lot of that is what I saw when I watched Michael Thomas (even though I didn't think he'd be THIS good).

[gunfrees]

What about that guy on DynastyNerds and DynastyFF that's done a superscore for 4 years and picked out the the Michael Thomas or similar talent each time? I can't say if the formula is right but something must be working The post he made recently

[Cameron Giles]

What about that guy on DynastyNerds and DynastyFF that's done a superscore for 4 years and picked out the the Michael Thomas or similar talent each time? I can't say if the formula is right but something must be working The post he made recently

I really don't agree with using hand size and 40 time to predict a WRs career. These things ignore so much about what really matters at the position.

But, I get it. It's fantasy and there are only so many numbers to work with if you're creating a formula, even though they ignore a bulk of the key traits.

[Orenthal Shames]

I tgink with all these statistical models it’s more created to give you higher odds at hitting and also weeding out really clear “fakers/fake studs”.....Montgomery.....

But otherwise if you use the system for what it is, a tool to help you make decisions, and not the be-all-end-all sorta deal then you’ll be fine.

Also if you have a system that has shown success, and if there are revisions to it that help you identify more hits without effecting your hit rate, why wouldn’t you make that change?

I mean I’m working on a similar system for WRs and I had one change that helped me identify like 6+ more “stud” with late breakout ages. That includes Michael Thomas. But with systems like these which include noticing statistical patterns, you often find that there are more and more variables that become important/relevant as you continue to build it that you wouldn’t have even thought of before

I think a lot of these formulas include an incredible amount of hindsight and are revised to include more hindsight. I understand why that is, because new information is acquired, but it's hard to label that as "predictive."

It essentially implies that there's an underlying mathematical reason that Michael Thomas is really good and we need to find out what it is so we can recognize the next Michael Thomas (there isn't).

Rotoviz, which is (or used to be) a prominent website for advanced fantasy metrics consistently said that Michael Thomas would be a bust. They had formulas and everything to support it. I used to follow their Phenom Index as well, which was touted as their highest predictor of WR success and last I checked, it hasn't been very good at it. It would be odd now for them to turn around and create a WR formula based on Thomas. Just like it would be odd to create some UDFA WR formula based on Adam Thielen.

I'll always read what people put up in regards to this stuff, but especially for the WR position, I don't think you can boil things like route running, releases, leverages and separation vs. press and man to a predictive formula. Those things are a lot more predictive to success for me and a lot of that is what I saw when I watched Michael Thomas (even though I didn't think he'd be THIS good).

Wait, so there are no magic formulas?

[FantasyFreak]

I’d take Sanders over Jacobs. He fits today’s nfl better, plays on a better team in a better offense. Plus he has killed it the last 4 weeks. Could see him moving out of the 50/50 timeshare and getting more like 70% of the work.

There's nothing that Jacobs doesn't fit in with today's NFL. Jacobs is the better, more instinctive runner, and he was an excellent receiver at Bama, the big question is will Gruden use him more that way? Sanders is good, but I feel he has a bigger chance to regress in the passing game next year. Wentz likes to push the ball down the field, and the use of RB's more heavily in the passing game was more due to necessity with all the WR injuries. I like Sanders, but I'd take Jacobs over him every time.

[Cameron Giles]

Wait, so there are no magic formulas?

I don't believe there is, but I think this report can help you get a feel for the type of RBs that do not translate to the NFL long-term. It was popular opinion that this is a weak RB class.

It's fantasy football and we are amateur scouts at best when it comes to watching games. So, the demand for shortcuts is high so we can just look at a single number or formula for a verdict. I just don't think those things are predictive long-term as much as they are reactive.

[Vcize]

What about that guy on DynastyNerds and DynastyFF that's done a superscore for 4 years and picked out the the Michael Thomas or similar talent each time? I can't say if the formula is right but something must be working The post he made recently

With a sample size of 4, with really only half of them looking like true hits so far (and we'll see if JuJu actually ends up being a hit in the long run).

1. Draft Pick
2. Height
3. 40 Time
4. Hand size

You really think it's that easy? Based on that tiny sample size with 2 hits?

If you take the highest drafted player whose last name starts with an 'S' over the last 4 years you get Sterling Shepard, JuJu Smith-Schuster, Courtland Sutton, and Deebo Samuel. Code cracked?

[Dynasty DeLorean]

I think a lot of these formulas include an incredible amount of hindsight and are revised to include more hindsight. I understand why that is, because new information is acquired, but it's hard to label that as "predictive."

First, a formula is never going to be as accurate as what I do here. You simply catch too many false-positives. Secondly, I can't speak for other people's work but for myself the biggest problem was the small sample size. Unfortunately all the college production metrics and combine data simply wasn't tracked well up until recently. I don't think it's that it doesn't work, there was just so little accurate data at the start. Even playerprofiler doesn't have all the data correctly, no single place does. It took a lot of work to go double check pretty much all of it from different sources and news articles, something I'm pretty sure nobody else on the planet has done. I started off with somewhere between 100-200 players that only went back 5-10 years at best (and that didn't include every RB that entered the league, only the ones good enough to stick around), entering thousands of data points by hand. Now i'm up to 500 players that better represents the league in it's entirety, which is a huge improvement. As for my revisions, much of that is from fixing incorrect data. I've been using the same criteria for 5 years, although that's not what the trolls would like you to believe.

[Cameron Giles]

I think a lot of these formulas include an incredible amount of hindsight and are revised to include more hindsight. I understand why that is, because new information is acquired, but it's hard to label that as "predictive."

I can't speak for other people's work but for myself the biggest problem was the small sample size. Unfortunately all the college production and combine data simply wasn't tracked well up until recently. I don't think it's that it doesn't work, there just wasn't enough accurate data at the start. Even playerprofiler doesn't have all the data correctly, no single place does. It took a lot of work to go double check pretty much all of it from different sources and news articles, something I'm pretty sure nobody else on the planet has done...

Absolutely. I think your report is much more sophisticated than what's typically out there and you put in hard work to do it.

I still wonder what the next step for you is as the game evolves. More and more we're seeing these players like Kamara, Ekler, White, etc. put up RB1 numbers without being the classic 1,000 yard between the tackles runner.

[Dynasty DeLorean]

I still wonder what the next step for you is as the game evolves. More and more we're seeing these players like Kamara, Ekler, White, etc. put up RB1 numbers without being the classic 1,000 yard between the tackles runner.

Now that we're getting a bigger sample size I probably will go back and check at some point, but you will get quite a few false positives. The quick way is to sort by College Target Share (from high to low on the list below) and then take a look at all the highly drafted RB's (let's say round 5 or better).

Ty Montgomery
Jacquizz Rodgers
Jaylen Samuels
Pumphrey
Dexter McCluster
Dri Archer
Chris Rainey
Duke Johnson
Tyler Ervin
Duke Johnson
Mikel Leshoure
Devontae Booker
Gio
DeMarco Murray
Kamara
Aaron Jones
Hines
Ito SMith
Isaiah Pead
McCoy
Ajayi
James White
Vereen
Daniel Thomas
Ballage
Charles Sims
Stepfan Taylor
Buck Allen
Prosise

I left out the more obvious ones (like Saquon, Fournette, etc). There's just a lot of false positives there, i'm sure I could come up with something better but I think there still will be a lot of false positives, more than I'd feel comfortable with anyway. Unfortunately, on paper there's a lot of RB's that "look" like Kamara, Ekeler, White, etc...

[hoos89]

I’d take Sanders over Jacobs. He fits today’s nfl better, plays on a better team in a better offense. Plus he has killed it the last 4 weeks. Could see him moving out of the 50/50 timeshare and getting more like 70% of the work.

Jacobs played half the season with a broken shoulder. I expect we'll see his receptions increase next season.

[blemly]

Not that my opinion matters to anyone else, but I’m prepared to call Sanders a “hit” as of today. Slow start to the season and certainly has question marks that remain (mainly vision), but I think he has even improved in that area as the season has progressed. For a raw rookie, he had had a very nice season and it’s worth mentioning that his fantasy numbers aren’t buoyed by TDs.

170-766-3 and 47-510-3 in 15 games is by all accounts a very promising season.

[ericanadian]

I tgink with all these statistical models it’s more created to give you higher odds at hitting and also weeding out really clear “fakers/fake studs”.....Montgomery.....

But otherwise if you use the system for what it is, a tool to help you make decisions, and not the be-all-end-all sorta deal then you’ll be fine.

Also if you have a system that has shown success, and if there are revisions to it that help you identify more hits without effecting your hit rate, why wouldn’t you make that change?

I mean I’m working on a similar system for WRs and I had one change that helped me identify like 6+ more “stud” with late breakout ages. That includes Michael Thomas. But with systems like these which include noticing statistical patterns, you often find that there are more and more variables that become important/relevant as you continue to build it that you wouldn’t have even thought of before

I think a lot of these formulas include an incredible amount of hindsight and are revised to include more hindsight. I understand why that is, because new information is acquired, but it's hard to label that as "predictive."

It essentially implies that there's an underlying mathematical reason that Michael Thomas is really good and we need to find out what it is so we can recognize the next Michael Thomas (there isn't).

Rotoviz, which is (or used to be) a prominent website for advanced fantasy metrics consistently said that Michael Thomas would be a bust. They had formulas and everything to support it. I used to follow their Phenom Index as well, which was touted as their highest predictor of WR success and last I checked, it hasn't been very good at it. It would be odd now for them to turn around and create a WR formula based on Thomas. Just like it would be odd to create some UDFA WR formula based on Adam Thielen.

I'll always read what people put up in regards to this stuff, but especially for the WR position, I don't think you can boil things like route running, releases, leverages and separation vs. press and man to a predictive formula. Those things are a lot more predictive to success for me and a lot of that is what I saw when I watched Michael Thomas (even though I didn't think he'd be THIS good).

Wait, so there are no magic formulas?

I don’t get this attitude. You must use some sort of thought process in deciding who to draft. If you use the eye test, fine, but to act like your eyes are some sort of magic formula is no less ridiculous than the stat geeks.

[AussieMate]

Not that my opinion matters to anyone else, but I’m prepared to call Sanders a “hit” as of today. Slow start to the season and certainly has question marks that remain (mainly vision), but I think he has even improved in that area as the season has progressed. For a raw rookie, he had had a very nice season and it’s worth mentioning that his fantasy numbers aren’t buoyed by TDs.

170-766-3 and 47-510-3 in 15 games is by all accounts a very promising season.

Yeah everyone jumps on him for his vision but it has slowly improved as the season has progressed combined with the fact that his vision seemed fine in college makes me believe he will only trend up from here.