Talent, Opportunity, and Why Player Evaluation Matters in Fantasy Football
Let me start by mentioning that I just published a new book on daily fantasy sports strategy—another in my Fantasy Football for Smart People series. While I’m at it, why don’t you follow me on Twitter. Also, if you live in the Philly area, you think you could pick up my laundry tonight?
Okay, let’s get to it.
One of my favorite stats is that, since 2000, running backs drafted in the third round or later have averaged a higher yards-per-carry than those selected in the first two rounds. I posted that nugget in the original draft of this article and Fantasy Douche pointed out that YPC is pretty volatile and probably a poor way to judge running back talent, which is right. Plus, as he noted, mid and late-rounders probably benefit from a reduced workload and maybe getting the ball in beneficial game situations.
To try to account for that, I examined 2013 running back success rate. Success rate is a metric at Advanced Football Analytics that quantifies how often a play was successful, i.e. how often did it increase an offense’s expectation of scoring? A two-yard run on 3rd-and-1 is successful; a two-yard run on 1st-and-10 isn’t. The idea is that it should somewhat standardize running back touches based on situations.
Here are 2013 success rates broken down by draft round.
You can probably tell that there’s not too much of a relationship here, and the data points actually appear to be trending upward. The 2013 correlation between running back success rate and draft round was positive at 0.23, which means that backs drafted later generally outperformed those selected earlier, even after we account for game situations (although the effect was weak).
For the record, I also looked at expected points added per play (EPA/play), which is sort of similar to success rate but actually quantifies how much the running back helped his offense in terms of how many points they can be expected to score before and after plays in which he touched the ball. The correlation was again positive at 0.12.
So anyway, the point is that mid and late-round running backs probably aren’t very much worse than first and second-rounders. At the very least, it has been difficult for scouts to differentiate between them, and scouting running backs is an admittedly difficult process.
Scouting has basically become a “which-flavor-do-you-like?” endeavor through which scouts search for exceptions to the rule, as some of the smart RotoViz writers have pointed out. It’s a shame, because you have a group of evaluators who are presumably very good at doing what they do—study tape and get a sense for a prospect’s background and work ethic, among other things—yet for the most part, scouting isn’t standardized, quantified, or tested.
Even knowing that, though, we’ve seen far more dominance from early-round running backs than mid and late-rounders in terms of bulk stats. Yes, running back is a position at which you can find talent late because of the evaluation errors, but we’ve still seen the bulk of the success come from top picks; since 2000, 29 of the 37 rookie running backs to score at least 200 fantasy points in standard leagues were drafted in the first two rounds (h/t Jim Sannes). With the precipitous drop in running back draft slots that we’ve seen recently, that’s pretty amazing.
What’s going on here should be pretty obvious: the highly drafted running backs get more playing time. Since fantasy production is so dependent on touches, it doesn’t really matter that first and second-round running backs have largely disappointed in terms of efficiency; they still make up the majority of the league’s top running backs because of the coaches’ collective perception that they’re superior players, which if course leads to more touches. It’s a self-fulfilling prophecy: we’ve drafted this back highly because we think he’s good, and now he’s going to “become” good because we drafted him highly.
This is sort of a conundrum for us fantasy football idiot savants because one of the primary areas where we can potentially use advanced stats to generate an edge over the field is in talent evaluation, yet most of a player’s fantasy value comes via his workload. Predicting workload isn’t terribly complicated—just a matter of staying up to date on news and using some basic math, really.
However, I still think talent evaluation is important in fantasy football, and I think it can lead to championships if you use it in the proper way. I’ve come up with seven situations in which player evaluation helps fantasy football owners.
Production = Workload x Efficiency
That’s the general formula we need to decipher to win in fantasy football. It follows that if two players have nearly comparable projected workloads, we want the one who is going to be the most efficient. Things like a player’s teammates and scheme factor into that equation, but so does his own talent.
It might seem like siding with the more talented player in “tiebreakers” will lead to a small advantage, but don’t forget that pretty much every pick is a tiebreaker; there are multiple considerations with each selection, so we need to use something to differentiate those players.
There are a lot worse methods of drafting than sorting players into tiers based on their workload, then organizing those tiers based on their projected efficiency/talent. So basically opportunities dictate draft range and talent determines which players you actually select (such that you’re always optimizing some combination of workload and talent).
Timeshare Situations/Ambiguous Roles
This is sort of related to “tiebreakers,” but any time two players are either in a timeshare situation or they have otherwise ambiguous roles, that’s a situation in which we should be emphasizing talent. Remember, all other things equal, we want really good football players.
The most obvious timeshare situations are at running back, but it doesn’t have to stop there. I ended up with Jimmy Graham in a lot of dynasty leagues years ago before he broke out because I emphasized his raw athleticism among other late-round tight ends with unclear roles. So we’re not just comparing teammates, either.
This sort of approach really helps hit on late-round players because 1) everyone either has an ambiguous role or simply isn’t slated to get much playing time and 2) we want upside anyway, which leads to the next point.
One of the reasons that I think a value-based drafting system can fail is because it assigns a singular number to a player and that’s supposed to represent what he can do in a given year. But things aren’t that black and white.
When we start thinking more probabilistically, players start to differentiate themselves. Yeah, maybe this ultra-athletic tight end has a very low median projection just because he probably won’t get playing time, but what if he does? What can he do when he gets the looks? How much upside does he have?
I’m a proponent of comparing floor projections to cost early in drafts and then pretty quickly transitioning to an emphasis on ceiling projections. When you look at the players who post not just good seasons—not just seasons that return “value”—but great seasons, you see that they’re typically highly athletic (outside of the running back position at times, which is basically 90 percent workload-dependent).
Value isn’t binary. There’s a range of it, and accurate player evaluation can help identify the players with 1) high ceilings and 2) a high probability of realizing that potential.
Talent — > Workload
Another reason to worry about talent is that, because projected workload is flexible, the best players (usually) rise to the top. This is basically a case of trying to identify the best player before coaches realize it, thus getting the player at a reduced cost (usually nothing) before others notice his potential. The danger is that you kind of have to count on teams properly evaluating talent, which probably doesn’t always happen.
We don’t really have time to wait it out in redraft leagues, so opportunity reigns supreme. In daily fantasy, it’s even worse; it’s basically a game of week-to-week workload projection.
In dynasty leagues, though, you of course have time to wait on players to develop and for coaches to finally come around on them. To me, there’s a sliding scale of how much we can and should emphasize player talent, with redraft leagues being at one end, keeper leagues somewhere in the middle, and full dynasty leagues at the opposite end of the spectrum.
There’s also a range of player evaluation usability by position, i.e. identifying talent is more important at some positions than others. Specifically, it’s most important when the potential deviation in efficiency is at its peak. Let me explain.
The reason that workload matters so much for every position is because it can differ quite a bit among players; No. 1 running backs are typically so much more valuable than No. 2 running backs not (necessarily) because they’re more talented, but because they see so many more touches.
Meanwhile, think about the deviation in rushing efficiency. Any back at 5.0 YPC is absolutely crushing it, yet that’s less than a 20 percent increase from the league average of around 4.2 YPC.
It follows that talent matters most for the positions at which efficiency varies the most on a per-touch basis. In order, those positions are wide receiver, tight end, quarterback, and running back. It should be no surprise, then, that I value talent far more at receiver and tight end than at running back. For starters, receiver opportunities aren’t as binary, and second, they have a larger potential deviation in per-opportunity efficiency.
As a side note, this is related to why running quarterbacks can offer so much value. There’s a huge deviation in both rushing workload and efficiency for quarterbacks, but it isn’t fully accounted for in fantasy drafts.
The king of fantasy football Shawn Siegele has an article “Why NFL Draft Analysis Wins Fantasy Titles” in which he basically sums up my thoughts on drafting rookies, so I’d definitely check that out. Rookies might not offer value as a whole, but many do simply because fantasy football drafts so frequently mirror the real draft, and we know the real draft is inefficient.
You can find value on rookies by beating the NFL market. That might sound really difficult, but trust me when I say it certainly isn’t an impossible puzzle to solve.