Let's talk about it because lots of people are throwing it around relative to Michel Hernandez as a reason that he should be in the lineup more often. It sounds logical. If a player is hitting everything like he's Barry Bonds, you want him in the lineup as often as possible. The problem is, hot (and cold) streaks hold as much predictive value as the player's eye color.
If you haven't purchased your copy of The Book yet, please do so. The publication is a brilliant collection of analyses on a range of subjects that we discuss often; bullpen usage, lineup ordering, platoon advantages, etc. and most relevantly, hot streaks. Tom Tango and company found the hottest and coldest hitters (on five and seven game streaks) in 2003 and compared their wOBAs after the streak to their expected wOBA based on the previous three years of data.
Essentially, they found hitters who hit better than projected for five and/or seven games at a time and compared what they did post-streak to what we expected prior to the hot/cold streak. The 10 hottest hitters flashed a combined .712 wOBA during the seven-game hot streaks, but in the three games after hit .341, below their expected .370 mark. The 10 coldest hitters had a pitiful .076 wOBA through seven games, then turned around and replicated their expected .324 during the next three games.
Entering Sunday's game, Hernandez had a .454 wOBA, CHONE had him pegged for a .300 wOBA and ZiPS said .280. Hernandez wOBAs in the minors the last three years are .307/.326/.312, all in Triple-A. Average those out, assume for park, assume for an increase in competition, and so on, and it's going to tell you one thing; Michel Hernandez is the worst hitter on the team.
Yeah, but we should play Hernandez until the hot streak runs out.
Yeah, but we shouldn't. Forget about those four games and evaluate Hernandez as a hitter. Odds are you would say he's bad, right? Okay, would you want that in your lineup every day? Of course you wouldn't. Plus The Book already showed us that the game after a five-game hot streak batters hit about 0.006 wOBA points better than expected on average. That's a marginal difference for anyone, but for an already awful hitter it makes about no difference.
So if we know that Hernandez is not a good hitter and that hitting streaks are relatively meaningless, that tells us that we shouldn't alter our lineup one bit for him.
Sometimes there's more than just some random variation and fluctuation going on, but the odds of Hernandez and his .450+ BABIP being legitimate are about as good as me sprouting wings and flying off to Atlantis (Yes, I realize Atlantis is underwater.)
[Do note: this really has nothing to do with Dioner Navarro. In fact, I didn't mention him. Let's try and keep the comments away from Navarro and focus on streaks and Hernandez. We've all had the Navarro debate and there's nothing new to discuss anyways.]