Share this post:Tweet about this on TwitterShare on FacebookShare on Google+Email this to someone

Trent Richardson

The RB Similarity App had me curious…if I accept the app as an accurate way of projecting 2013’s production, what would this app have to say about my first round selections?

If you’re unfamiliar with the app, go ahead and give it a little test drive here, or read this primer page.

How accurate is it the app?  Well, in a world where usage never drastically increased or decreased and offenses never changed… I’d think it could be very accurate! Basically, the app tries to predict the future (year N+1) using the historical data we have about similar running backs.  Once we understand what similar backs have done in year N+1, we can then understand the most likely outcomes for a player this coming year. Sure, this information needs to be coupled with our knowledge of the NFL: Some backs get traded to new teams and are in different systems for example. This opens doors for backs on the other team, etc…But one thing at a time.

For the sake of being interesting and fun, let’s see what the app has to say about what the top 10 PPR running back rankings should look like. The scoring projections come from the RB Similarity App, with no modification.

If this app was ‘gospel,’ this is how we might rank the top running backs:

Rk

Player Low Median High

1

Doug Martin

16.5

20

21.3

2

Ray Rice

13.4

17.5

20.1

3

Trent Richardson

12.8

14.8

20.6

4

Arian Foster

12.2

14.3

17.4

5

CJ Spiller

9.9

15.6

18.3

6

Marshawn Lynch

11.6

14.8

16.4

7

Alfred Morris

9.8

13.9

16.5

8

Matt Forte

10.6

12.5

15

9

Adrian Peterson

9.4

13.4

15.4

10

Jamaal Charles

9.5

11.7

14

 

Very interesting stuff!  Here are a few takeaways I have from this exercise:

  • Doug Martin fares VERY WELL in this analysis. Historically speaking, backs with similar years to his last year, do very well in year N+1. With no new or major competition behind him on the roster or significant reason to not like the guy this year, I’m VERY high on him.
  • Matt Forte may continue to be underrated on draft day.
  • Ray Rice and Trent Richardson are frequently being drafted in the second half of round one, yet both fare very well here. Perhaps you’d draft other guys ahead of them still, but this app would seem to reassure anyone nervous about drafting them.
  • I’m sure you all noted, as I did, how low the off-season darling Jamaal Charles was there. (Oh yeah, and some guy named Adrian Peterson didn’t fare too well either. Maybe you’ve heard of him?)
  • Again, the app doesn’t factor in all the information you and I have. It doesn’t know both Charles and Peterson were recovering from major injuries last year and were playing “less than 100%.”  It doesn’t know Charles or Forte have new offenses this year.

However, the cool thing about this app is that we can customize some of the inputs to affect the data we get out of it. For example, Peterson was returning from an ACL and MCL tear. He played much better in the second half of the season than the first.  We’re actually able to manipulate the app, telling it to factor in only the second half of the season. We can do the same for Jamaal Charles.  Maybe you’re one of those people who believe that Martin is overrated due to a few of his highest offensive outputs. So we can actually ask the app to give us similar RBs to Martin without those games. We can also use the app to somewhat predict usage changes. We can tell the app to give us data for Rice using only the games where Pierce got X number of carries, and for games with Spiller where he got at least Y number of carries.

 

Player Low Median High Data Points

A

Adrian Peterson

12.8

14.5

16.6

2nd Half Only

B

Jamaal Charles

9.7

12.2

13.5

2nd Half Only

C

Doug Martin

16.1

19.4

22

Taking out Biggest Games

D

Ray Rice

9.7

15.9

18.7

Pierce at 8+ carries

D2

Ray Rice

12.2

16.5

19.1

Pierce at < 8 carries

E

CJ Spiller

12.2

16.5

19.1

FJax < 15 carries

From this new set of data, we learn that AP projects out much better this year when we only consider his second half of last year. Given his continued healing, it’s not tough to use this app to project him for another great season this year.  Jamaal Charles, interestingly though, has no real change in data points in this workup.  The app doesn’t know about the improved offense and quarterback play Charles will enjoy this year, but it does seem to offer us Charles believers solace in noting that his floor is still as a (later) first round running back.  You’ll note that Martin still fares extremely well, even missing a couple of his biggest games. Make no mistake about it, this app just loves Doug Martin.

Rice projects similarly in the games where Pierce carried the ball a reasonable amount than when he didn’t. That’s in part because Pierce got a lot of work in some games where Rice ALSO got a lot of work…It’s not perfect, but it’s a start. Perhaps we shouldn’t freak out about Pierce’s effect on Rice just yet.  Spiller, however, earns a significant bump in projection when we consider when Fred Jackson worked on limited carries.

So what does all of this mean?

The app may not be a perfect predictor of future success or failure….But I think there is plenty to learn from playing around with it. As a fun exercise, play around with these ten backs (or more) and let me know what ranking the app tells you they warrant. Besides being an interesting exercise, I wonder what else you could learn while doing it.

After making all the adjustments listed above, here is my “rotoviz RB app top 10.”  What’s yours?

1)      Doug Martin

2)      Trent Richardson

3)      CJ Spiller

4)      Arian Foster

5)      Adrian Peterson

6)      Marshawn Lynch

7)      Ray Rice

8)      Alfred Morris

9)      Matt Forte

10)   Jamaal Charles

Play with the RB Similarity app yourself, here.

Subscribe for a constant stream of league-beating articles available only with a Premium Pass.

Share this post:Tweet about this on TwitterShare on FacebookShare on Google+Email this to someone
3 comments
FFAvenger
FFAvenger

Hmm... I'd have to take a real skeptical look at any 'RB TOP 10" list without LeSean McCoy on it.  Probably a hole somewhere in your app given his down 2012, but I'd take him over Morris, Forte, and Lynch.

FFRittle
FFRittle

@FFAvenger Thanks for the feedback James. Worth noting that this isn't necessarily a "RB TOP 10" list exactly. It's merely meant to offer us an objective way to look at some top RBs. In that light, I probably should've counted McCoy and done #11 RBs. However, the app does generally prefer these 10 backs over McCoy and "ten" is such a slicker number than "11."  For what it's worth, the app offers us quite the range of possibilities for McCoy.  It includes a "median" ppr score of 10.9 but a "high" score of 19.1. It would not be a stretch to use some subjective thought and project McCoy for hitting the top half of that range.


We all use subjective thoughts in rankings. This article (and the app) is just meant as a tool to offer us some objective evaluation as well. I'd love to discuss more if you have other questions

rotoviz
rotoviz moderator

@FFAvenger Probably not a bug of the app, it's a feature. It forecasts based on the most recent season. That doesn't mean that McCoy couldn't be better in 2013, it's just that the app will be blind to it. I always recommend using the full range of the information available in order to get to your final position on a player.