Fantasy Performance of Top Running Backs

As we continue to develop our AI for ranking NFL players, I’ve been spending time diving deeper into running backs. In general, AI is a “black box” machine learning algorithm, we are also working to find signals that are more interpretable for fantasy managers and can be used to enhance their performance. I have listed some of my initial findings below.

Disclaimer: I will continue to post analysis similar to that shown below. The theories we test can have positive or negative results and we will write about both.

I looked at the top 12 RB since 2010 for this analysis

#1 There is higher weekly upside for the top tier RB

The graph below shows there is a higher standard deviation for the top 5 RBs and then it slowly falls off. This could mean a number of things. For the sake of this post, let’s just consider it’s impact on weekly points

Standard deviation will measure both upside and downside risk. Let’s explore this a little bit. Outside of the top rusher, there tends to be a pretty similar average yards per game (see graph below) for the highly ranked rushers. However, as you move down the rankings you have fewer outliers, or performances, that could carry your team in a given week. Not only do the highly ranked running backs have a higher upside risk but they also have a lower downside risk as well, which is represented by the skinnier part of the violin in the chart below.

Bonus: Here is graph for the top 32 running backs since 2010. This shows the same trend described above.

#2 Top Rushers Tend to Be on the Best Teams

The graph below is pretty chaotic. It contains the number top 12 running backs per win rate since 2010.

This one, not so much! This is looking at the % of teams that contain a top 12 running back in since 2010 by win rate.

The gist (aka my take): The top 5 rushers tend to be on winning teams.

This makes sense, good teams tend to have good players. In future drafts and rankings, I might question a potential break out candidate or “sleeper” who is on a team that is not projected to perform well. I would also be hesitant to draft a good running back that moves to a bad team.

For example: DeMarco Murray disastrous 2015 fantasy season. In 2014, DeMarco Murray was the top rusher in the league, running for 1,845 yards on a 12–4 Dallas team. That offseason, he moves to the Eagles who end up going 7–9 and only rushed for 702 yards. The Eagles massively under performed, as an Eagles fan, it was easy to see through the hype, I didn’t buy any stock in DeMarco Murray that year. DeMarco Murray was a consensus **top 10 RB in fantasy before the year started and ended up being complete shit.

**corrected text calling him a top 10 pick.

Bottom Line:

Top rushers have added value in the form of potential game breaking weeks and it is prudent to consider overall projected team performance when evaluating top rushers (based on rushing yards). Although the effects of the findings above aren’t large enough to warrant a complete overhaul of strategy, it is something I will be considering when drafting next year.