You’ll see it from time to time — ‘this guy is hot, you have to pick him up,’ or ‘this guy is hot, this team needs to move him up in the batting order.’ Brandon Belt was terrible last week, now he should be the number two hitter. Jon Jay is hitting .444 over the past two weeks, maybe you even heard someone tout him specifically.
Except that, if you read The Book: Playing the Percentages in Baseball, you know that a player’s work over the past week or two has little predictive bearing when it comes to his next plate appearance. There’s a recap of the study here if you need a refresher. It makes intuitive sense — if you’ve gathered information on a player for four minor league seasons, and then two-plus major league seasons, why would Jay’s last two weeks stack up to that? We’ve learned all about sample size, and this is in the SSS wheelhouse.
There is a preponderance of evidence, some even more recent than Tango’s work. Ben Lindbergh found just this year that despite Albert Pujols’ contention that his home runs would come in bunches and he wasn’t worried, and even despite the subsequent home runs that came in bunches for him, home runs do not, in fact, come in bunches. Sky Andrecheck looked at 30-game hitting streaks — and he found that a player’s batting average was LOWER than expected during the streak. Derek Carty replicated the work found in the book for pitchers and hitters and found that, while cold streaks are slightly more reliable, you’re not going to learn too much from a hot or cold streak.
Perhaps you want to give a little bit of credence to a cold streak. There’s a tiny sliver of evidence that it can mean something, and it’s probably because there are other things going on. How many cold streaks ended with a DL stint? How many cold streaks came during a divorce or similar personal problem? Players are human, and negative feedback seems to beget more negativity. But let’s not become amateur psychiatrists here.
Let’s just emphasize that there’s little to be learned from a hot streak.
At most, you might be able to cobble together some trade value with a less-saber-savvy fantasy owner, or maybe a manager will be duped into giving a player a bigger role based on a streak. Considering the likelihood that the hot streak isn’t real, those are not sturdy pegs upon which you should rest your hat.
Case closed? Not quite. It turns out that there is some evidence that there are more volatile seasons than others. A player might be more streaky in a certain season — and all things equal, we want our position players to be more consistent than streaky in real life, and especially in head-to-head leagues, where a bad week at the wrong time can sink the best juggernaut.
Bill Petti, writing for Beyond the Box Score, did a five-part series on hitter volatility last season. He started by showing rolling ten-day weighted on-base-percentage graphs. Check out Evan Longoria and Scott Rolen versus David Wright in 2011. David Wright, by the way, showed up on Seth Samuel’s similar work on FanGraphs as a streaky player as well.
You can see that Evan Longoria was the clear own in head-to-head fantasy baseball last season. There were weeks where David Wright was nigh un-ownable.
The lessons from his reasearch are not completely clear for fantasy baseball, since year-to-year correlation is still beyond Petti’s reach. His calculator can find you a streaky player this year, but won’t tell you if the same guy is going to be streaky next year. And since this was designed for real-life baseball, it favors players that can take walks and stay productive while the hits weren’t falling. Unless you’re in an OBP league, that doesn’t have a ton of use even in a head-to-head league, other than a few more runs and stolen bases, perhaps.
But there are some general takeaways that work in fantasy baseball. The equation uses strikeouts as an input, which points once more to the negative aspects of the whiff. Players that strike out don’t put the ball in play enough to show good batting averages, and since their batted ball samples are smaller, they can end up at the extremes in that category as well. Then again, Adam Dunn’s .245 BABIP and Carlos Pena’s .252 BABIP also have a lot to do with their batted ball mixes, so let’s just stick with the impact strikeouts have on batting average and volatility.
If you are considering two players with about-equal talent levels for your weekly lineup head-to-head league, you might want to use Petti’s volatility calculator to find out which player is more streaky. Maybe you’ve got a choice between Josh Reddick and Corey Hart for your final outfielder slot — you’ll find that, despite almost identical wOBAs, Reddick is a tiny bit less volatile. He hasn’t had a month as bad as Hart this season, and he also hasn’t hit .300 in a month like Hart, too.
But they’re pretty similar. For a more stark difference, you’ll have to go to something like Brandon Phillips vs Michael Cuddyer. They both have double-digit home runs, some stolen bases, and decent batting averages — and a .342 wOBA. Now you’ll get a volatility score that’s 20% higher for Cuddyer than Phillips. If you need a steady eddy, you’d take Phillips. If you’re an underdog that would need everything to go right to win, you might pick Cuddyer. Actually, right now, with Cuddyer out with an oblique strain, the decision would be easy, but it’s still too good of an example to pass up.
And there you have it. Hot streaks tell you nothing, cold streaks tell you barely anything, and a player’s streakiness in general can help you split a few hairs down the stretch. The easiest answer is to just ignore that whole aspect and get the best players on your team, but who said anything about fantasy baseball being easy.