MLB Daily Games Strategy: The Truth About Hot Streaks

The Truth About Hitting Streaks

Did you see the link I tweeted last week to this article a colleague of mine at the University of Rochester wrote about the hot hand phenomenon? It provided a few opportunities to compare DFS players to monkeys, but there is so much more to it. Dr. Hayden's lab is overall interested in how people make optimal decisions, using the probability of receiving a reward as motivation. In other words, he wants to understand how our brains enable us to make the decisions that yield the best possible outcomes. If you read me ever, I bet you can already guess why I find this so fascinating and relevant to daily fantasy sports.

The hot hand is akin to the streak. You've heard the term in casinos no doubt as it gets thrown around any successful gambling situation. The new research by Dr. Hayden's group used monkeys that were trained to respond to a binary choice in the way they felt best predicted a juice reward. (Monkeys are famous for loving bananas, but grape juice is the treat they'll really work hard to earn). The experiment was to vary the likelihood that the next reward was correlated with the previous rewarded choice. Using conditions of low correlation so that the previously rewarded choice was almost never rewarded on the follow up trial to very high correlation (the previously rewarded choice was usually rewarded again on the follow up). They found that monkeys believed in the streak, or hot hand even when actual rewards were not correlated, whether intentionally or in a truly random trial.

This is important because it means that our own, human tendencies to believe in streaks, particularly in sports, are not the result of some sports culture, or something we learned growing up, or anything environmental. The tendency to believe in streaks has apparently been around since we were monkeys! By studying this phenomenon in monkeys the Hayden lab hopes to discover the exact way that the brain decides to favor the rewarded choice, even in the face of evidence that it may not be rewarded in the future.

Hitting streaks in baseball are the best example of the hot hand in sports. People do reference the hot hand in basketball shooting, but this doesn't really present an opportunity for DFSers to take advantage, since the hot shooting performance is typically confined to a single game in basketball. There has been a LOT written about this topic. As early as 1993, C. Albright concluded that while hitting streaks exist in baseball, they don't exist at a higher frequency than would be predicted by a model of randomness. In other words, if you know how many hits each player got over how many at bats (he used data spanning several years) and modeled it assuming randomness, you'd see hitting streaks emerge. Imagine flipping a coin 400 times in a row. You'll get some fairly long runs of heads and tails. Streaks.

You can do a quick google search of "baseball's streakiest players" and get all kinds of results. Whether it's the number of games in a row that a player hits safely in or the way his batting average fluctuates week to week or month to month, some players get a streaky reputation. As a DFS players, does it make sense to target players in the midst of a hot streak, or avoid those in the midst of a slump? The data all says no. The most relevant study may be this one from 2011 in which Derek Carty compares the ability of a 7, 14, or 30 day hot streak (defined as a player averaging 6+ FanDuel points over the stretch) to predict the next day's fantasy performance to that of a weighted three year projection system, Marcel. The projection was more accurate than the streak by 2-3 FD pts in each case, though longer streaks did better than shorter streaks in predicting the next game.

With all the evidence pointing to hitting streaks being nothing but our biased interpretations of random events, and not useful for our daily fantasy purposes, why do we still believe? Why has the tendency for people to see streaks where mathematics sees nothing evolved over millions of years to still rule our decisions?

For better (most of the time) or worse, we have profited from identifying patterns in our environment. For one thing, going back to simple times of hunting and gathering, returning to "lucky" spots resulted in life over death, literally selecting for people that exhibited this ability. The brain operates at its most efficient when it can simplify complex information into known, reliable patterns. Sometimes, patterns are created when they don't actually exist, thus the hot hand phenomenon.

There is an emotional side to this too. The difference between predictable stress and unpredictable stress is one of the biggest disparities we can face. When we can create a framework in which to understand the events we face, we feel prepared and equipped to make good decisions that benefit us. When we feel we can predict the future, even when it isn't positive, the chemistry in our brains is measurably different (and healthier) than when we're faced with uncertainty. This provides a powerful motivation to be able to find patterns in random events. It's really good for us.

So we know that streaks happen, even in a completely random sequence of events. We know that there are good reasons for why we tend to and want to ascribe them special importance and predictive value. In sports, their occurrence may even be more than just random. Success breeds confidence which can breed more success. Teammates may change their behavior in order to keep the streak alive. While we as fantasy players may hit upon a player in a streak and reap the benefits of that streak for a while, we have to know it won't last. When my research all points to a guy being a good play and he happens to be in the midst of a hot streak, I do let it make me feel more confident for using him...even though I know it doesn't matter. In the end, I guess we're all just slightly less hairy monkeys when it comes to daily fantasy sports.