I tend to analyze a lot of crap that many people think is inconsequential – things like a player’s hand size or his broad jump or how he feels about eating cereal for dinner. Why?
In his book David and Goliath: Underdogs, Misfits, and the Art of Battling Giants, Malcolm Gladwell suggests that, for a variety of reasons, it’s superior to be a big fish in a small pond than vice versa. This idea is exactly why it’s preferable to focus on small, “minor,” underappreciated traits.
In effect, we’re trying to increase our market share of exploitable advantages by decreasing the number of people with which we need to compete. Something can’t be exploitable if it’s overvalued by the public, so it follows that it’s advantageous to focus on acquiring big pieces of smaller potential advantages – those of which people aren’t aware or aren’t valuing. Here’s a somewhat crude representation of this concept.
In this graph, the two stats in question possess nearly the same predictive value; Stat 1 is just slightly more predictive than Stat 2 (on this hypothetical and completely arbitrary scale of 100).
The key here isn’t only the predictive capabilities of each trait, however, but also how much they can be exploited. The exploitable value of Stat 1 is only 20 percent, compared to 50 percent for Stat 2. The latter has more actual value – that’s what I’ve labeled here as ‘exploitable value’ – because it isn’t priced into the market as heavily as Stat 1. Note that Stat 1 could be a lot more predictive of success than Stat 2, yet still possess less actionable, exploitable value.
Ultimately, we’re not looking for the most important traits in some philosophical sense, but just the ones that can lead to the biggest advantage over the field. That’s calculated by subtracting public perception – or how strongly a stat or measurable is factored into market value – from inherent value (or its predictive ability).
Thus, we get the big-fish-in-a-small-pond effect; by focusing on a bunch of “minor” traits that are highly exploitable, the aggregate advantage can be massive. Finally, note that some traits can have negative value. The short shuttle isn’t predictive for running backs, for example, but it’s still priced into their cost to some degree (in the actual NFL Draft). In that case, when we subtract the public perception of the short shuttle’s importance (some positive figure) from its inherent value (zero or close to it), we necessarily get a negative number, which informs us that it would be smart to target running backs who have other predictors of success but who actually struggle in the short shuttle.
WR Stat Correlations
With that said, we still need to know the importance of “the big stuff” to understand how to exploit marketplace inefficiencies. Below, I charted the correlation between a few wide receiver stats and fantasy points over the past three seasons.
I included the difference between standard and PPR scoring just so that you can get a sense of which traits might have added value on DraftKings – a full PPR site.
You can see that targets are the most important stat in both standard and PPR leagues, which isn’t surprising. Workload is the biggest predictor of fantasy success at every position; no matter how talented the player, he simply can’t produce without seeing enough opportunities. Also note that targets are slightly more important in PPR leagues with a ridiculous 0.92 r-value. That means that a wide receiver’s targets basically predict 92 percent of his fantasy production in PPR leagues.
Next up are touchdowns, which check in just behind targets. Considering that the majority of a wide receiver’s fantasy points come via receptions and yards – not touchdowns – these values are very strong. You can see that touchdowns matter slightly less in PPR formats than standard leagues, but the r-value is still 0.78, which is very strong. Just for fun, I checked on the correlations for catch rate and aDOT (average depth of target, developed by Pro Football Focus). The effects are weak; in general, we want receivers who catch a high percentage of their targets with those receptions coming as far downfield as possible, but again, the efficiency doesn’t matter much if the targets aren’t there.
Remember what I said about the difference between theoretical and usable value? Well, it could very well be possible that, despite a stronger correlation between targets and fantasy points than touchdowns and fantasy points, scoring could be more important to emphasize, even on a full PPR site like DraftKings.
I will be doing more research on this topic in the future, but it seems as though wide receiver salaries often mirror past stats pretty well. The wide receivers who have performed the best have the highest salaries, as you’d expect.
With touchdowns, though, there’s enough volatility that sometimes elite touchdown-scorers have extended scoreless droughts – see LeSean McCoy, Le’Veon Bell, Keenan Allen, and Andre Johnson (among many others) in 2014 – and that causes them to drop in price (when they probably shouldn’t).
We don’t see the same sort of trend with targets. It isn’t like a wide receiver can go a month or two without seeing targets or receptions; because those are relatively high-frequency, they quickly regress toward the mean. That adds up to wide receiver salaries on DraftKings often being an accurate representation of future targets, but not terribly accurate in terms of touchdown-scoring ability.
Since touchdowns are just barely less important than targets when it comes to wide receiver fantasy scoring and there’s potentially much greater salary inefficiencies when it comes to touchdowns, there’s arguably more usable value in trying to identify future touchdowns. There’s a lot of variance from week to week, but not nearly as much over the course of a season; the same types of receivers score again and again. In a nutshell, targets and receptions are fully “priced in” to a wide receiver’s salary. Touchdown-scoring ability is not.
Cash Games vs GPPs
Lastly, I want to quickly note that your strategy might change based on the league type in which you’re playing. In cash games, it’s smart to pay for predictability. Targets are far more predictable than touchdowns from game to game, which means those PPR monsters (think Julian Edelman) have a high floor and a lot more value in head-to-head and 50/50 leagues.
Some people extend that argument out to tournaments, targeting high-volume receivers who can’t consistently score. I think that’s a mistake. Opportunities are vital and of course you want as many targets as possible for your receivers, but upside comes in the end zone; wide receivers usually need to score to offer GPP value on a consistent basis.
As I’ve said in the past, the first thing I ask myself when choosing wide receivers for a tournament on DraftKings is “Can this player realistically score two touchdowns on a semi-consistent basis?”