Daniel Murphy is on fire.
Coming into Sunday’s matchup with the Cubs, Murphy was riding a 7 game hitting streak during which time he is hitting .500 (14-28) with six extra base hits, five runs scored, and four RBIs. This recent offensive explosion has brought his triple slash to .301/.337/.455.
On Sunday, Terry Collins decided to move Murphy up to the leadoff spot in the lineup, presumably to take advantage of Murphy’s hot streak.
While that might seem like a good idea, it is a misguided move that is oft repeated by managers across Major League Baseball.
By adhering to streaks and making decisions based on who has the so-called “hot hand,” managers put too much stock into small sample sizes and instead ignore things like the talent of the hitter or pitcher.
All streaks are too small to matter.
Statistically speaking, Murphy has just as good of a chance of getting a hit Sunday as he did two weeks ago when he was “cold,” it’s just now, his BABIP is adjusting itself back to the mean, giving off the appearance of a hot streak.
One might argue that Murphy may have a mental edge when stepping into the box when he is on a hot streak, and that affects his approach at the plate and therefore, the outcome of the at bat.
But there is no evidence that this is the case.
For one, if it was the case, then you’d certainly see hitters get into more prolonged streaks, right? After all, by that argument, their mental edge at the plate would lead to them making better contact and getting more hits, leading to more of the same.
Also, the sample sizes that we’re talking about when we discuss hot or cold streaks are too small to be able to separate any perceived mental edge out from simple random variation.
The same goes for hitter vs. pitcher statistics.
Former catcher and current Miami Marlins manager Mike Redmond had a reputation for being Tom Glavine’s nemesis throughout his career. In 48 at bats against Glavine, Redmond hit .438 with a 1.075 OPS. Redmond’s career numbers in those categories were .287 and .700, respectively; pretty good for a guy who never was a regular starter, but nothing that suggests he should have owned a future Hall of Famer like that.
While it is entirely possible that Redmond saw the ball well out of Glavine’s hand, or that there was some sort of mental aspect at play, the sample size is insufficient to determine if Redmond truly “owned” Glavine or if the impressive stats are a result of random variation that would be corrected with a larger sample size.
Note that there is a distinction that needs to be made here between individual pitcher vs. hitter splits and lefty-righty platoon splits. The platoon splits are based on a much larger sample size and in most cases will stabilize within a season or two. They are therefore much more reliable when it comes to predicting the success of a matchup.
So what should a manager make of streaks?
Nothing. When a manager starts a player against a certain team because he’s had success in the past against that team or in that stadium, he his making a mistake based on the misuse of statistics. The manager should play a player based on whether or not he gives the team the best chance to win on that given day.
If Juan Lagares is 0-for-his-last-9, but has traditionally hit lefties well, then starting him against a lefty like Travis Wood is a good idea, while sitting him because he’s “cold” is not. The Mets won Sunday partially because Terry Collins ignored the fact that Lagares was cold and played him anyway.
 Although with the lineup Collins used Sunday, Murphy’s OBP skills can justify putting him in the leadoff spot, but this is not the reasoning that Collins used, which is what makes it misguided.