What classifies as a breakout?
How one classifies a breakout player is very subjective. Some say finishing in the top 24 at a given position is a breakout, others say it is outperforming ADP by a sizable number. Initially, I used a positive point differential from the one season to the next to classify a breakout. However, this seemed like a flawed process, as prior season points are not the only factors contributing to ADP the following season (age, opportunity, coaching changes to name a few other contributing factors.) I eventually came across JJ Zachariason’s series of articles on finding breakout players [1]. He plotted preseason ADP versus end-of-season fantasy points, and used a trendline to calculate expected fantasy points at a given ADP. From there, he calculated the difference of total end-of-season fantasy points minus expected-ADP fantasy points.
I ended up following the same process, pulling overall running back PPR ADP from MyFantasyLeague from 2013-2019, and plotting it against end-of-season PPR points.
Using the generated logarithmic trendline equation, I was able to calculate expected fantasy points by draft position, a point differential.
The Process
Next, I had to think about what variables I would include in my RB Breakout model. Like any logical person, I decided to pull all 101 different player stats, including combine metrics and prior season statistics. I then ran the correlations of their point differential against each of these variables (from a sample of 396 running backs). The highest correlation was prior season run snaps with an r-squared value of 15%, admittedly not staggering results.
Version 1
In my first model, I eliminated all variables that had an r-squared of less than 1%, leading to a pool of 35 variables. Next, I created a formula that weighted each of these variables by strength of correlation, producing a score that predicts how much a player will outscore expected points at their ADP, with 10 being a perfect score. A 10 means the given player is projected to outscore their expected ADP point by a larger margin than a player with a score of 9. The results are below.
Although interesting results, this failed to predict any true breakouts. Looking at the top 10 players, they have all broken out. These results go to show that prior season opportunity is most predictive of outperforming expected ADP points.
Version 2 – The Real Breakouts
In the second model, I only took variables that were on a per game, per carry, or rate basis (e.g. carries per game, light front yards per carry, or stuffed run rate). That way, the model would not heavily favor those who didn’t play a complete season or were on the field fewer snaps. Again, using the 1% r-squared threshold, leading to a formula that took 20 different variables into account.
This led to some large score jumps from a few players, most notably, Mike Boone, Derrius Guice, and Rashaad Penny.
Final Results (3rd time’s a charm)
Taking it a step further, I eliminated all running backs with a 5th round or higher ADP, leading to the results below.
Important Notes
It is important to know that these models do NOT take situational changes into account. That is to say that free agent signings, rookie running back additions, or coaching changes are not factored in. That is why you need to discount players’ scores like Mark Ingram, who now has to compete with second-round rookie J.K. Dobbins.
Additionally, I ended up removing rookies from the model because all combine metrics had too poor of correlations, so the only variable left that applied to rookies was years played.
Furthermore, this model is not the end all be all. The correlations between point differential and each of the variables were poor at best. This should be used as one piece of information to determine who you will believe will break out. It should not be used as the only piece of information you use to make those judgements.
Thanks for reading! If you’re interested in more content like this, you can follow me on Twitter at @LukeNeuendorf.
Sources:
[1] Zachariason, JJ. “How to Find Breakout Running Backs in Fantasy Football.” NumberFire, 2 June 2020, www.numberfire.com/nfl/news/32021/how-to-find-breakout-running-backs-in-fantasy-football.
