Joe Girardi made a series of bold moves in Game 3: Starting Brett Gardner, Eric Chavez, and star prospect Eduardo Nunez; benching Alex Rodriguez and Nick Swisher; batting Curtis Granderson 8th and allowing Raul Ibanez to face a lefty down a run with 2 outs in the 9th. There is a thread running throughout this list: making decisions based on extremely small statistical sample sizes.

Forget for a moment the outcome of each move. These roster choices reveal that the core of Girardi’s decision-making process relies on completely unsound methodology — which means he’s likely to continue making suboptimal decisions. This NJ.com article provides a trove of quotations about Girardi’s thought process, and it’s not a pretty picture.

“Gardy played well at this time of year last year, so we’re going to run him out there.”

Before you criticize us for being stat facists, let’s be clear here. We don’t have a problem with using scouting in a small sample size. Having the backing of reliable, robust stats is ideal. Yet, sometimes in a pinch, having coaches or scouts who can spot mechanical or health issues can be vital. But, that’s not what we’re talking about here. Girardi is ostensibly going on a statistical basis — it’s just a really, really bad one.

There is precious little evidence that a player’s stats in a given month is predictive. The fact that BMG had good results in 18 plate appearances over 5 games in October of 2011 has no bearing on how he should be expected to perform on October 17th, 2012.

We love BMG. We like players who put up 5 and 6 WAR over the course of full seasons. But, Girardi had relegated BMG to the bench for the start of every playoff game until Game 3, telling reporters he didn’t think Gardner’s bat was ready. What changed his mind? Five practice at bats? With Ivan Nova and Freddy Garcia tossing? Apparently so:

“We had him get four or five live at bats yesterday and he swung the bat well …I stuck with the guys who had played the whole year and had a lot of at bats, but I just felt like now, I’m going to give him a shot.”

And now moving on to the sensitive Nick Swisher…

“[Swisher]‘s one of the guys who’s struggling.”

Yes, Swisher hasn’t gotten the job done. Maybe his overdone eyeblack is weighing him down. But he’s only played in 7 games. One week. One week, mind you, where his walk and strikeout rates aren’t wildly afield from his career norms, but where he’s been battered with a .211 BABIP. He’s had plenty of bad one weeks stretches. He had a whole month back in May of this year where he was just putrid. Heck, Swisher had a whole year where he struggled with BABIP issues. That was exactly what allowed Brian Cashman to steal him from the White Sox, knowing full well that his regressing to career averages was the best bet.

“Granderson’s struggling, too…But we also looked at who was pitching for us, too. When you look at [Swisher] and Granderson, their numbers are pretty similar off Verlander, so we wanted the defense out there.”

While Granderson played in Game 3, we highlight this quote because Girardi often cites pitcher vs. batter matchups. Forget about the fact that Curtis Granderson has hit for a .379 wOBA against right handed pitchers over 1126 regular season games in his career. Instead, let’s look at what Granderson did against Verlander in 18 total PAs.

Girardi’s reasoning isn’t even internally consistent. If he’s basing his decisions on batter vs pitcher stats, guess who should have been in there…Alex Rodriguez. ARod has gotten a look at Verlander 28 times in his career and has posted a 1.198 OPS.

Here’s something else about Arod. Despite his continued decline, he absolutely murdered lefties this year to the tune of .308/.410/.514 (.398 wOBA). That’s a skillset to have handy for pinch-hitting duties when you’re down a run in the 9th with two on and two out, and Phil Coke is on the mound. Southpaw Coke holds the line (!) against lefties (3.38 career xFIP), but has had his share of troubles solving the average righty (4.85 xFIP).

Rather then using someone who kills lefties, Girardi left Raul Ibanez in. We don’t need a quote to know why he did it. Ibanez has had 5 immensely clutch hits recently in very similar game situations:

– a game-tying homer and walk-off single against the Red Sox October 2nd
– a game-tying homer and a walk-off homer in Game 3 of the ALDS
– a game-tying home run off Valverde in Game 1 of the ALCS.

All of these came in Yankee Stadium, where the right field porch gives lefties a big helping hand. However, beyond the fact that this game was in Comerica, the bigger issue is that Girardi let Ibanez hit against the lefty on the mound. Ibanez has a .262/.315./.420 (.320 wOBA) career line versus lefties, and in recent years, he’s getting worse: .230 wOBA this year and .255 wOBA in 2011. ARod would have been the clearly superior choice against Coke. If you’re worried about Leyland countering with Benoit, you can send up the switch hitting Nick Swisher.

Bottom lime, Girardi let the narrative bias in the small sample of recent games overwhelm the decision-making process. He has not given the Yankees their best chance to win, and based on his own words, we shouldn’t expect him to start doing so anytime soon.