The Statistical significance of a ‘Hot Streak’ or ‘Slump’ for Your Fantasy Picks
The concept of hot streaks and slumps are as entwined into the sport of Baseball as rightly-lefty matchups and intentional walk. They are viewed as essential and ever true philosophies for success by the conventional wisdom that has long governed decisions. However, with the injection of math and increased statistical analysis in the sport, the effectiveness of these theories is being questioned. While there are always exceptions to the rule, I believe these are all overused notions as they apply very less often than is believed.
A hot streak is when a player has managed to link together several successful games and therefore is predicted to continue to perform well for the foreseeable future. A slump is just the opposite. It is a player who has performed exceptionally poor lately, and needs a good game or a couple good at-bats to return to his normal performance level.
However, the argument for the side rejecting the validity of these concepts is that the occurrence of these streaks do not exceed their theoretical probability in the whole. This means that statistically, a player is bound to have strings of good and bad games and it is not explained by the fact that they are hot or cold, but simply coincidence.
Consider this simplified example. A player has a 50% chance of having a good game, and 50% chance of having a bad game. If he has three good games in a row, analysts will call that a hot streak. Same goes for three bad games and a slump. The probability of each situation is 50%*50%*50% = 12.5%. With Hundreds of players in the league and this high of a probability, there are bound to be plenty of players on “Hot” and “Cold” streaks at any given time. Those analysts who believe in these concepts may use these to predict how a player will perform in future games, but really, his chance of a good or bad game is still 50%.
Let’s look at this concept in a real-world application. For those statistics experts out there, I used a modified exponential distribution to model each player’s fantasy performances versus their season fantasy averages. Since season fantasy averages are the main factor in determining a player’s salary in Daily Fantasy Contests, a player who exceeds his season averages is bringing your lineups value.
I took 500 MLB players and compared their past 3 games fantasy value versus their season averages. From this, I determined a probability that that player hit his given fantasy value over the past 3 games, assuming he is playing no better or worse than he has all season. My cutoff was 10% in the positive direction, meaning there is only a 10% chance that the given player is playing as well as he has over the past three days without improving his level of gameplay.
Statistically, we should see approximately 50 of these players attaining performance levels if the concept of hot streaks is false. However, if it is true, we would see more players in this threshold, telling us that those players are actually playing the game at a level that is statistically higher than their season average.
What I found not only disproved those concepts but actually tells us that recent performances may be a poor predictor of future performances. Only 23 players exceeded this 10% threshold, with Justin Smoak, Miguel Montero, and Martin Maldonado leading the list.
What we can learn from this is that apart from a couple outliers, season averages are the best predictors of future performance. This does not necessarily help you in Fantasy Cash Drafts, but what you can do is avoid the perils of your competitors. When your competitor makes a mistake, you win. Let your competitors choose players who seem to be on a “Hot Streak” and abstain yourself. Fantasy value is derived from consistent gameplay and consistent performances, using recent performances is gambling. Daily Fantasy Tournaments are games of skill. Make your draft picks as such.