September 24, 2003
Cheaters?, Part II
This probably happens to everyone. After I filed my column for Tuesday, I started to think I'd missed something, that there was one more thing I'd forgotten to look at. The next day, I had the same feeling, so when the column went up, I went back again and there it was, staring me in the face. I started this follow-up immediately, and considering the amount of e-mail I normally get I was stunned that I was able to dive into it a couple hours before a reader sent feedback that nailed the problem exactly:
I found your latest Breaking Balls ("Cheaters", in case it takes a week for you to get to this e-mail) quite interesting. But I also find your conclusion a little odd, especially considering the Red Sox splits at home with RISP. The reason for the drop-off is right in your article:
"Some teams have supposedly gone to always using more complicated signs usually reserved for runner-on-second situations when facing the Sox."
Since RISP usually means a runner on second, teams will switch to the more complex signs. Anyone stealing signs would be more likely to screw up and relay the wrong pitch, or be unable to relay any information at all, either way one would expect a decrease in the hitter's effectiveness. In fact, one could argue that Boston's poor performance in those situations is evidence that they rely heavily on stealing signs. Not that I blame them, it's not cheating after all.
If I had written a note to myself from today to myself earlier this week, I couldn't have made it any clearer. The flaw in my logic, implied in the article but never outright stated, was this: The Red Sox are a smart team, and if they decide to steal signs, they're going to be able to do it almost all the time.
Oh ho, not so fast. What if the Red Sox, like a particularly bad product manager, "pick the low-hanging fruit" instead of coming up with a complicated system where they have two, three college kids in the stadium decoding the system?
It's harder to pick off signs when the other team tries to conceal them. Catchers may put down one finger and that means the next sign, two for the one after that, or they might go first sign, second sign. But given time and effort, it's possible to work out almost all the simple concealment measures. Well, Jamie Moyer is said to have worked out a super-complicated system where shaking off pitches means different things depending on which sign is flashed and when, but Jamie Moyer's such a preparation freak he probably does this kind of thing in his head while charting pitches on his off day.
My point being this: Say Theo Epstein and the rest of the brain trust were sitting around thinking of ways to squeeze wins out of the team, and someone said: 'Hey, let's steal signs--it's easy, it's effective, and once they catch on, no big deal.' But Epstein and everyone else were busy, so they gave it to an intern or something, and he decided to stick a television in center field and have the bullpen signal the pitches in. The bullpen guys turned out to be OK at stealing the signs when they were obvious--one finger fastball, two fingers curve, three fingers change--but they lost interest when it got complicated.
What did the stats from Tuesday's column really show us? Of the regulars I could get two solid years of data on, four of them showed some improvement in their home performance, two of which (Nomar Garciaparra and Johnny Damon) saw big jumps. However, in looking at the OPS of these five guys, four of them saw OPS drops of 100-195 points when there were runners in scoring position (Jason Varitek jumped 190 points).
So! Teams go to sign concealment when there's a guy on second. Looking only at guys with 400 or more PAs, I, Mojo the Statistical Assistance Monkey, will delve even further into this topic for your reading pleasure.
Wildly Varying Sample Size Spotlight Normal Signs Harder signs Batter Empty 1st 3rd 1,3 2nd 1,2 2,3 Loaded Nomah .438 .347 .467 .467 .440 .407 .333 .000 Damon .366 .420 .214 .250 .308 .412 .636 .333 Walker .370 .464 .500 .200 .436 .240 .571 .167 Ramirez .451 .407 .667 .263 .591 .345 .800 .000 Millar .305 .476 .500 .500 .419 .405 .375 .273 Mueller .376 .483 .483 .417 .391 .500 .600 .375 Nixon .432 .314 .429 .385 .619 .455 .222 .083 Varitek .322 .381 .625 .375 .385 .522 .556 .313 Ortiz .359 .512 .500 .467 .414 .333 .500 .583 Avg .389 .422 .488 .373 .450 .403 .528 .276
I don't really expect you to get any useful information from that, it's just that I spent a lot of time on it. Don't worry, I'm not paid by the column-inch or anything. It's the Internet--I'm hardly paid at all. Here's some more useful information.
Batter OBP, normal OBP, harder Difference Nomah .415 .400 -.015 Damon .363 .399 .036 Walker .395 .344 -.051 Ramirez .445 .506 .061 Millar .356 .393 .037 Mueller .423 .446 .023 Nixon .406 .419 .013 Varitek .353 .433 .080 Ortiz .407 .419 .012
Now a normal hitter is a little better in the harder situations, with a guy on second base. By 20-30 points, though I unfortunately don't have league situational splits to be precise about that. Most of these guys are right around that, with Manny way on the high side and Nomah and Walker way under it. And to a lesser extent, Nixon and Ortiz. So if there's cheating going on, those are the four you should suspect.
At home, in situations in which there was no runner on second, and the Red Sox bullpen could steal signs straight, the OBP of these guys was about .401, and in situations where it was harder to cheat, they turned in an OBP of about .418. What's that, comparatively?
Normal Signs Harder signs Team Empty 1st 3rd 1,3 2nd 1,2 2,3 Loaded Yankees .355 .348 .463 .355 .364 .334 .453 .320 Boston (all) .314 .372 .464 .341 .393 .371 .460 .327
I picked the Yankees because overall, their offensive profile seemed the closest to the Red Sox, though I freely admit I didn't apply similarity scores or anything.
Yankees with normal signs: OBP of .356. Harder to cheat: .357. Red Sox overall with normal signs: .333, and with harder signs, .384ish. So over the whole season, all parks, the Red Sox see a +.050 OBP boost in situations with a guy at second, and at home, that boost appears to be much slimmer, only +.020. How do the hitters close the gap at home by .030 points of OBP when there isn't a runner on second?
Or, to put it another way, we'd expect a hypothetical Tigers team to put up an OBP of .100 at home with no one on second, and an OBP of .150 with a runner on second, and instead, the split is .130 with no one on second, and still .150 with a guy on second.
I'm reluctant to point to 30 points of aberrant OBP in a situational split like this as evidence of everything. Anyone who pays attention to park factors has seen some weird one-year numbers come out of that. And yet, that's over 6,000 plate appearances this year, and a difference of 20 points of OBP for a team as a home field advantage turns the Twins into the Cardinals, or the Tigers into the Indians.
There are a couple of problems, as other readers pointed out to me in e-mail. We don't know what to expect if hitters are tipped off. How much can it matter, and do certain hitters profit more from knowing what's coming? Does anyone know how the signals came in, if they did? For a batter to pick them up, they have to be pretty obvious, but I haven't read any speculation on how the mechanics might actually have worked. Are there any other teams who show this kind of weird split between the two situations?
And the answer is that I have no idea, and Mojo's tired.
When I wrote my last column, I was convinced that if the Red Sox were cheating, it was too subtle to be easily seen. After working over all these stats today, perhaps the runner/absence of same on second offers us the outline of a not-at-all subtle scheme.