Understanding consistency in weekly fantasy football
DraftKings Jonathan Bales explains what owners need to know about understanding consistency in weekly fantasy football.
By Jonathan Bales DraftKings
For the most part, weekly fantasy players don’t pay much for kickers. Amateurs and pros alike understand that it’s usually senseless to pay top-dollar for a position that’s not consistent from week to week. It doesn’t matter how many points a player scores and it doesn’t matter how scarce those points are if you can’t predict his performance.
We all seem to intuitively know that we shouldn’t pay for kickers, but few people extend this argument to the other positions. So even with DraftKings removing Kickers from their roster formats this season, you can still use lessons learned from them in developing a lineup. In leagues in which safety is the name of the game, there should be a strong positive correlation between the percentage of cap space you’re willing to spend on a player and your ability to accurately project his performance.
It’s not like any of the skill positions are unpredictable in the same way as kickers, but there’s still varying degrees of predictability. Those should undoubtedly have an influence on your decision-making. All other things equal, you could maximize your team’s long-term floor by allocating a higher percentage of the cap to the safest players.
I charted the number of “startable” weeks for the players at each position. A “startable” week was defined as finishing in the top 33 percent at the position (among the top 30 quarterbacks, tight ends, defenses, and kickers and the top 75 running backs and wide receivers).
A “startable” week was defined as finishing in the top 33 percent at the position.
You can see that running backs have been the most consistent position, with the best of the bunch giving you a top 10 performance 67.0 percent of the time. Quarterbacks aren’t far behind at 61.1 percent, but no other position is close.
When you think about it, that shouldn’t be a surprise. Consider the number of opportunities each position has per game. For quarterbacks, it might be 35 attempts. For top running backs, it’s in the range of 15 to 25 touches.
Meanwhile, wide receivers and tight ends might be lucky to see 10 targets in a game, and it’s often much fewer. Just based on those numbers alone, you’d expect quarterbacks and running backs to be more consistent, and thus more predictable. It’s like asking if a baseball player will come closer to hitting at his career average after five games or after 20 games. . .there’s just no contest.
Taking it a step further, I analyzed the percentage of “top-tier” weeks turned in by each position. I defined “top-tier” as a top two finish for quarterbacks, tight ends, kickers, and defenses or a top five finish for running backs and wide receivers (the top 6.7 percent for each position).
"Top-tier” is defined as a top two finish for quarterbacks, tight ends, kickers, and defenses or a top five finish for running backs and wide receivers
Again, no contest. Quarterbacks and running backs are just far more consistent on a week-to-week basis than all other positions. When you’re paying for reliability, you should start with the quarterback and running back positions.
The position consistency data is certainly useful in all league types, but we can cut up the data a little more to obtain even better accuracy. Specifically, we can look at subsections of each position to see which types of players are the most consistent, and thus worthy of the majority of our cap space in head-to-head leagues.
Why look at player types? Well, even though we’d expect certain individual players to be more consistent than others, it’s really difficult to separate that data—which necessarily comes with a small sample—from noise. Maybe Player X is safe, but maybe he’s just gotten lucky; we really can’t tell. Would we ever create enduring narratives for baseball players through eight games? Never, so why do it for NFL players?
I really believe weekly fantasy players as a whole place too much stock in past game-to-game consistency on the individual level. Again, it’s not that game-to-game consistency doesn’t exist, but just that it’s going to be really, really difficult for us to separate it from randomness.
It’s the same reasoning behind my typically bullish stance on injury-prone players. Are some players more likely to get injured than others? Probably, but that doesn’t mean we can turn that idea into actionable information.
Injuries are relatively low-frequency events controlled heavily by randomness, and humans aren’t built to properly deal with randomness. We perceive all sorts of signals that aren’t really there because it’s not all that evolutionarily beneficial to say “I don’t know.”
But in weekly fantasy sports, saying “I don’t know” is a great thing; by factoring your own fallibility into your decisions—a choice that’s reflected in your stance on week-to-week consistency and injury-proneness alike—you’ll be able to acquire value where others are overlooking it.
The bottom line is that the majority of what most think they see as consistent play is noise. It would take years of NFL data to establish individual player consistency to the point that we can trust what we’re seeing isn’t just randomness. By that time, it’s too late.
The crux of my individual-player-consistency-is-kind-of-overhyped-but-maybe-not-completely argument is that a small sample size hinders our ability to obtain meaningful results. Thus, we’re better off analyzing certain player “types”—big wide receivers, mobile quarterbacks, and so on—instead of the past consistency of individual players.