Working with streaks a useful tool in fantasy leagues

There's no doubt that Major League players go on streaks. They get hot and go on long hitting streaks. They get cold and seem barely able to hit the ball. The question is whether those streaks are meaningful. In other words, are they the result of something that will continue? Whether it's "seeing the ball well" or a bump in confidence, does it mean to expect the good results consistently? Or are they just a byproduct of randomness? With so many players competing in so many games every single day, we're bound to see some streaky play just from chance alone, right?

When you start to dig into this issue deeper, you realize that it's easy to identify past streaks, but difficult to turn that information into something actionable to make better predictions. Sometimes hot streaks continue, other times, they don't. However, our concern in the Official Mini Fantasy Game of MLB.com from DraftKings should not be the philosophical discussion regarding streaky play and randomness, but rather the potential predictive nature of streaks.

The overwhelming evidence suggests that it is difficult to predict the continuation of a streak. Even if non-random streaky play exists, we'd still expect random streaky play to exist, too. Again, it's just a numbers game; you can flip a quarter 10,000 times and see all kinds of patterns in the data, but it won't change the fact that you have a 50 percent chance to get heads on the next flip.

While we don't know if non-random streaky play exists or how prevalent it might be, we do know some streaks are just the result of chance. As a result, we can expect "hot" batters to hit worse in the future than they have in the recent past (and vice versa for "cold" hitters). It's not that hitting the ball well guarentees hitting it worse in the future, but just that we'd expect stats to regress toward the mean across the long haul.

When a pitcher opens the season averaging 15 strikeouts per nine innings with a 0.50 WHIP, we can be pretty certain that his production is going to regress, right? Even if he's indeed "hot" in a non-random way, we still know his future stats are more likely to resemble his career numbers than those from his other-worldly start to the year. In that way, it can be smart to perhaps avoid "hot" players and target "cold" ones if others are on or off of them for reasons that aren't likely to endure.

Another reason to target players who appear to be slumping and bypass those who are on a tear is that daily fantasy baseball is very much a marketplace. It's not a static entity, but rather a market with ever-changing player costs, i.e. salaries. When players get hot, the Official Mini Fantasy Game of MLB.com accounts for it by bumping up their salaries. When players are cold, they dip in price.

But if all, or even most, streaky play is due to variance, the changing prices aren't reflective of real-world value. In other words, if salaries fluctuate to represent expected changes in production based on the very recent past, then there's probably value in going against the grain. This means that skipping over hot players whose salaries have soared and targeting those who aren't performing well and have seen a drop in price is a solid strategy. Note that this is true even if some form of non-random streaky play exists; if the Official Mini Fantasy Game of MLB.com overcompensates for streaks, there's value to be had by doing the unconventional thing.