Projected Running Back Success Score: 2021 RBs



Last year, I decided to dive into some predictive running back metrics to determine how off my film grades were (checks and balances are important, folks). I ended up spending hours creating a database to study more than 5 years of running back production and its relation to their college output. I released the first article here.


An incredibly important piece of utilizing the Projected Running Back Success (PRBS) Score as a tool is understanding that it is just that - it is not how I rank and it is not the Bible. It is simply giving an indication of guys who could outperform the perceived expectation if given the opportunity. Before we get into this year’s class, let me explain the PRBS Score.


Rather than separating analytics and film entirely, this model aims to incorporate both. Predictive metrics, draft capital, and my personal film grade on the player are all components of this model. What is the model used for? Identifying the most likely players from each class to hit, of course. A hit is a top-24 finish within the first two years of a players’ career. As aforementioned, the data that this model derives from is a 5+ year sample ranging from 2015-2020. “The sample isn’t large enough.” I utilized a smaller sample because the game has changed in drastic ways in the last ten years, and around 2015 is when we started to see that evolution truly impact the game. Some of the metrics include college target share, burst score, and college yards after contact. These metrics aren’t ultra correlative as individual values, but when combined to create a composite, they have an r-value of 0.53. Without further adieu, here is what the PRBS says for the 2021 NFL Draft class:



A couple of likely names at the top, of course. First round capital is important when discussing running backs. Does that mean that they are bulletproof? No, absolutely not. Clyde Edwards-Helaire did not get the top spot in the 2020 version of the PRBS Model. But it helps. Travis Etienne gets the largest bump in my rankings in terms of relevant players, according to the PRBS; that’s with my film grade having him at RB3. Here, he is the clear RB2. Najee checked a couple of boxes, and his capital gave him a shot to be relevant. Why wouldn’t he have been if he didn’t get first round capital? Well, because the average PRBS Score of a top-24 back over the last 6 years is 7.08. If Najee somehow fell to the third, I would have a much different outlook on him.


The important note here is that PRBS does not like Javonte Williams as much as I do; he didn’t hit on many of the predictive metrics, and despite being my RB1, he falls below some of the top guys. His PRBS is above average, though, and I still feel confident about his film and would draft him at RB1 or 2.


Gainwell and Chuba coming in just under the top dogs was surprising, but various metrics favored the two of them. Interestingly enough, Kenneth Gainwell has been compared to Nyheim Hines by his new coach in the last week, and there’s a really good chance that Gainwell becomes an RB2 or better according to this score.


One that is going to upset people is Michael Carter. Carter’s Day 3 capital mixed with his lack of success in the predictive metrics would indicate that he is a long shot to finish inside the top-24. I know a lot of analytics people who are still in on Michael Carter, and he was my RB4 in film, so this is one I would push back on. You’ll notice that some drafted guys aren’t on here; I’m still working on exactly how to incorporate those that I didn’t study into the PRBS. Going to find a median film grade and give them that (or just be better about immediately studying the players after they’re drafted).


Remember: use this as a tool and not as the gospel. If you use this tool and say “that’s a guy I like more now, I’m going to get him,” it’s critical to understand the perception around said player. If you want Gerrid Doaks, for example, you can wait forever; even if you want him three rounds before.


I'm always refining my models to get the most correlative sample. I will continue doing so, and I have some plans to incorporate a similar score for each of the fantasy-relevant positions next season.


Until next time.