In the first half of this analysis, I showed that in all three series, the Sox were amazingly better with RISP and either 0 or 2 outs than in all other PA. Here's what happened. R, 0/2 is short for RISP with 0 or 2 outs.
Yankees PA BA OBP SA EqA
Other 131 .246 .313 .381 .253
R, 0/2 31 .455 .548 .682 .417
Astros PA BA OBP SA EqA
Other 166 .196 .283 .291 .217
R, 0/2 30 .458 .567 .958 .479
Dodgers PA BA OBP SA EqA
Other 181 .193 .260 .337 .218
R, 0/2 30 .435 .567 .739 .436
Total PA BA OBP SA EqA
Other 477 .208 .282 .333 .227
R, 0/2 91 .449 .560 .797 .445
In the "special situation," they had 31, 30, and 30 PA. They hit .455, then .458, then .435. The had OBP's of .548, .567, and then .567 again.
You might recall that in the extra innings of Game 3, their offense in general disappeared, as they were playing while their bodies were literally asleep. They went 1/4, IBB in the special situation (and the hit was Nunez's infield hit that produced the error and the run). That series also, of course, includes pitchers hitting. So for the purposes of the next comparison, here's the WS results, counting only PA by non-pitchers in regulation play:
Dodgers PA BA OBP SA EqA
Other 146 .210 .253 .377 .225
R, 0/2 25 .474 .600 .842 .468
So, how do you explain all this?
(I should first note that the entire hypothesis that follows was created when I just knew about the general RISP splits. When I crunched the actual data and discovered that RISP with 1 out didn't follow the pattern, it made instant sense to me.)
The Astros staff was historically great, and many people were convinced that they could shut our offense down. The Dodgers staff was populated by lefty flyball pitchers, essentially our Kryptonite, and many people thought they'd be a serious challenge. There was a lot less enthusiasm for the Yankee staff against us.
So here's your first clue: if you rank these staffs the way we just did, by their warranted confidence they could succeed against us, you get a .262 EqA improvement in the special situation, then .243, then .164. IOW, the more confidence these teams could deservedly have that they could shut down our offense, the better we hit them in the special situation. (Again, this is something I discovered after the general explanation was devised, and it fit the theory perfectly).
So what's so special about these situations? There is a subjective sense, in the post-season and only in the post-season, that the game and hence the series is on the line.
When you have RISP and 0 outs, you are staring in the face of a potential big inning, and one big inning decides a majority of baseball games. You have to get the next guy out, or you've already allowed a run, in all likelihood, and you still have no outs and runners on. Think of the emphasis put on the next batter after a leadoff double in any inning of a post-season game. That PA is always perceived as crucial. The batter has to at least get the runner over, and a hit is a back-breaker. A crooked number looms.
The same thing is obviously just as true of RISP and 2 outs. Get the batter at the plate, and you're out of the entire jam. Give something up, and you've probably allowed runs to score. Again, the emphasis placed by announcers on these PA is impossible to miss.
RISP with 1 out? Not anywhere as much. A leadoff double is a crisis. A 2-out double leads immediately to the thought that you can't give up another hit, because you had 2 outs and no one on and then gave up a run. But a 1-out double seems like a much less big deal (the actual leverage indexes would tell a different story, of course). I've never heard an announcer stress the importance of getting the next guy out because of that situation. It's obviously important, but it's not hyped.
I believe that opposing pitcher psychology has to be the same.
And what do pitchers say to themselves when they think the game is on the line? All together now: "I'm not getting beat throwing my [name of any pitch except their best]."
That's my hypothesis: the higher the perceived leverage (in the post-season, especially), the more pitchers lean on what works best for them. They become more predictable. More subject to analysis.
Furthermore, I believe that the best pitchers -- the sorts of guys who are likelier to appear in the post-season -- are more prone to this, and that in those cases it's often a matter of warranted ego.
You probably remember the national TV game where Pedro faced Roger, and Trot Nixon went yard in the 9th to break a scoreless tie (I believe the screenshot of that swing used to be my image, before I changed it to Wright's double-breaking knuckler). What you may not know is that Trot Nixon owned Roger Clemens, and always did, because Nixon was a tremendous fastball hitter and Roger tried again and again to prove to himself that he could consistently throw his fastball by him. Even though he plainly couldn't.
So, pitchers in general narrow their repertoires as game leverage gets higher. They are more likely to go with what works best for them. In the post-season, this phenomena is multiplied dramatically, for two separate reasons; the leverage is much higher, and the pitchers are more confident (and perhaps have some ego investment) in the idea that their best stuff can get anyone out.
I actually think that's a good trait for a great pitcher to have; exhibit A, Pedro Martinez. But if you become downright predictable in key situations, in terms of pitch choice or (more likely) favorite pitch sequences, you are playing right into the hands of ... an analytically brilliant team.
The Sox were unexpectedly dominant this post-season because they had an exceptionally good idea what was coming in what everyone regards as the highest-leverage game situations. It's as simple as that.
And imagine how hard it is for an opposing pitcher to alter their patterns even if they suspect the other team has analyzed them. It's just human psychology. You're asking them to reject the approach they are most confident in, in favor of some approach that they feel is less effective.
Finally, there's a huge extra factor here for the hitters which is entirely psychological.
I've always argued that clutch hitting is what psychology calls a state variable rather than a trait variable. Clutch hitting happens and is real when it does, but it doesn't happen because of an innate ability to rise to the occasion. It happens when a player or team has confidence that they can produce in clutch situations. That confidence calms the whole nervous system down. The adrenaline boost that happens in pressure situations evolved to help us flee from predators. That it incidentally impairs fine motor skills didn't matter; we never evolved any mechanism to block that impairment. Any time you have some confidence that you can succeed especially well under pressure, it creates a positive feedback loop. Your adrenaline level goes down, your fine motor skill execution improves, you have more success, and your confidence increases.
Sometimes that confidence is the result of nothing more than past success. In fact, one PA can start the feedback loop. When the Sox were looking at J.T. Snow, I looked at his late-and-close split. The split had a ridiculous sub-split, which had a vanishingly small chance of being random:
Years PA BA OBP SLG
1992-2000 666 .230 .329 .359
2001-2005 328 .332 .451 .487
I know there's another such example ... oh, yeah. A guy named David Ortiz hit .786 / .850 / 2.286 in 20 potential walk-off PA from the 2004 post-season to late in 2006; the Sox won all 14 games, with Ortiz (I think they call him "Papi") getting the walk-off in 11 games and being on base after a walk when they won the other 3. The three times he was retired, he had a walk-off later in extra innings. As soon as he failed a couple of times, he went back to being absolutely ordinary in the clutch for a long stretch. (I think that this performance is so extreme -- I estimated the odds of it happening in a random simulation as a billion to a trillion to one -- that you need to include the likelihood that the opposing pitcher believed you were clutch.)
So those are two examples of how rising to the clutch can start a positive feedback loop. You can see, however, that knowing you have an actual edge, because you have a really good idea about what's coming, would start the feedback loop before you had some success.
So, how did our guys do in the normal situation versus the predictable one? Keep in mind that posting a .400 EqA is, by design, as difficult and rare as hitting .400.
Name nPA nEqA pPA+ pEqA+ Diff
Jackie Bradley 42 .210 8 .717 .507
Brock Holt 27 .244 5 .673 .429
Rafael Devers 33 .212 4 .633 .421
Eduardo Nunez 23 .161 5 .537 .376
Ian Kinsler 30 .142 5 .460 .318
And. Benintendi 50 .203 11 .449 .246
Mookie Betts 60 .208 10 .383 .174
Mitch Moreland 14 .273 5 .431 .158
Steve Pearce 38 .341 9 .461 .120
J.D. Martinez 49 .316 13 .356 .040
Xander Bogaerts 56 .226 10 .261 .036
Christ. Vazquez 33 .212 5 .156 -.056
Sandy Leon 13 .233 0 n/a
Even though the sample sizes in the predictable situations were tiny -- half the guys had 5 PA or less -- every hitter but one hit better in them. The first EqA column reveals how guys were actually swinging the bat, in fact. Pearce was the one guy who was locked in and out of his skull -- and having just one of those guys in a post-season is a typical number. JDM was JDM, and his reputation is that he's doing more analysis of opposing pitching patterns than anyone else, to begin with. Moreland's 14 PA were probably chosen against pitchers he could hit especially well. Most everyone else hit poorly in regular situations, as is often the case against elite post-season pitching. Amusingly, Sandy Leon was the 5th best hitter in the post-season in these normal PA's, which actually fits with the way he looked.
Xander was the only guy who seemed unable to exploit the data, but there's an interesting split in his performance in predictable situations. He started out 0/4, BB, with nothing hit hard and nothing out of the infield (0/3, BB in the DS, and 0/1 in game 3 of the CS). Starting with game 4 of the CS he was 2/3, IBB (plus 0/1 in extra innings), the hits being the game-tying hard GB single up the middle off of James that scored Benny from 2nd with 2 outs in the 5th of game 4 of the CS; and the ripped liner to left off of Maeda that scored the last run of the 5-run 9th of WS game 4.
He also called his father, whom he hadn't seen since he abandoned Xander and his family when Xander was 3, after the Astros series and invited him to the WS, where he saw him for the first time in 23 years. In 2013 his Dad flew in to watch him play and Xander elected not to meet him. I know I'm notorious for inventing narratives to explain data anomalies, but if you were a scriptwriter and were trying to create a reason why one guy on the team was mentally distracted and would join the party late, you couldn't do any better than that.
So what about Christian Vazquez? He was 1 for 5 when predictable, and the hit was the infield hit that Severino got a piece of that scored Devers from 3rd with 2 outs in the 2nd of game 3, starting the rout. He made no hard contact at all.
But if indeed they had a series of hitter meetings where they went over opposing pitchers and what they expected them to do with the game on the line ... wouldn't CV have been at a different meeting were they were going over the opposing hitters? Or at least sitting in the hitters' meeting and not paying attention as he went over his pitch-calling game plan in his head. I laughed out loud when I saw who the one exception to this pattern was. Again, it took no effort to explain.
This post will disappear soon (please don't quote it; everyone will know what you're talking about, and you'll save acres of screen space), and for obvious reasons I've decided not to write a guest column for BP entitled "The Red Sox Just Proved Clutch Hitting Was Real." If you want look at this again, you might grab a copy.