I’m clearly fascinated with game theory’s role in daily fantasy sports tournament success, and I think that’s really the next frontier of research. I think it’s natural to dismiss game theory or a contrarian style of play and say, “Well you just need to pick the best players.” That’s a naïve view, in my opinion, because it totally ignores probabilities and the potential benefits of actually selecting the most high-value players.
I always like to use extreme examples when attempting to demonstrate a point, and the hypothetical of a player or stack being 100 percent owned does a nice job of showing the absurdity of ignoring player utilization. If the Rockies were such a good play at home one night that literally everyone played the same stack, then no one would benefit or be hurt by whatever happened to those players; they could collectively hit 25 home runs or strike out 25 times, and it just wouldn’t matter.
If you were the sole player who faded that stack, you’d be in a sensational position if the Rockies tank; if you were in a 100-man GPP and 99 users rostered the Rockies, you’d basically win the thing if they fail to score, for example (or even if your stack outscores them and you don’t mess up your pitching).
In that scenario, you’d only need to be “right”—you’d only need for the chalk to disappoint—more than one percent of the time for you to benefit. If Colorado had just one bad game in 10, for example, you’d still be in a wonderful spot given that you’d surge past a field of 99 users in 10 percent of games. Anyone else interested in winning 10 percent of GPPs?
At the other end of the spectrum, imagine that 99 daily fantasy players had their heads up their asses and failed to roster the Rockies in a quality matchup at home. If you were the only user to play a Colorado stack, you’d benefit immensely from rostering high-value players without a heavy level of ownership. Whereas Colorado going off in the first scenario would put Rockies owners ahead of just one percent of the field, the same in the second scenario would put the lone Rockies owner ahead of all 99 other users.
It follows that player utilization matters—a lot—and that there’s a balance that we need to strike between finding value and still getting away from high-usage situations when it’s appropriate. That’s particularly true in baseball because, due to stacking, there’s a ton of lineup overlap.
The point at which value and usage “balance out” depends on how accurately we can predict the future. If we could predict the future with 100 percent accuracy, it would never make sense to go against the grain. But we can’t, so we should.
BUT, the frequency with which we should be contrarian (and the level of “contrarianism” we’re willing to take on) depends on the level of certainty we can have in game outcomes and player results. If the Rockies theoretically finished as the top fantasy offense 90 percent of the time when they’re projected above, say, 5.5 runs, that would make it much more challenging to fade them in such a situation than if they’re the top fantasy offense just 10 percent of the time, for example.
So instead of working with just pure value in a vacuum, there are really three factors at play here: value, the consistency with which we can properly identify that value, and the benefits we’ll receive if we’re correct in identifying that value.
As the ability to make accurate predictions decreases, the value of fading the chalk increases dramatically. And, as we’ve seen, baseball is a highly volatile sport from night to night, which is again the main impetus for my stance on going against the grain.