October 17, 2012
Should McRae Have Considered Brett For Miller in the Seventh?
Q. Did you consider Brett for Miller in the seventh?
... Why the f---- would I hit Brett for f------ Miller? Miller's starting the f------ game, he's playing against left-handed f------ pitchers, Brett is not playing against left-handed pitchers, why in the f--- would I bat for Miller? You think I'm a goddamn fool? Tired of all these stupid-ass questions every night. Asking me that stupid-ass s--- every motherf------ night.
The Royals entered their game on April 26, 1993 with a 7-11 record, six games behind the first-place Angels and a half game ahead of the last-place A's. It was Hal McRae's second season as the Royals manager, and you could appreciate the frustration of starting so slowly. The Royals had won just 72 games the year before, but they were seen as a bubble contender in the AL West. In this preview of the season, the writer Jack Logic summed up the Royals' position thusly:
In other words, McRae had a team that wasn't all that great, but that could still feel an urgency about losing. The Royals trailed by four going into the seventh inning on April 26, 1993 and made two quick outs. But the next two hitters singled, and then there was a walk, and the Royals had the tying run at the plate. The scheduled batter was Keith Miller, a right-handed third baseman who hit a lot of singles and got hit by a lot of pitches. On the mound was Dave Haas, a 27-year-old right-hander in his third season. He had started and he had relieved, and at the moment he was relieving, having replaced Tom Bolton, who was making his first start of the year and was pulled after 66 pitches.* George Brett, 40, a left-handed hitter, was on the bench. Miller was allowed to bat, and, on an 0-2 pitch, he fouled out to the third baseman. The Royals' chances of winning dropped from 11 percent to 4 percent. Now, apologies if this is a stupid-ass f------ question**, but ...
Should McRae have considered Brett for Miller in the seventh?
First, to get the semantics out of the way, everybody should at least consider everything. Here, I'm considering eating a sock. OK I've decided not to eat a sock. No harm done! I'm sure McRae considered Brett for Miller in the seventh, unless he was distracted because he was in the clubhouse watching The Pelican Brief (topical! It's 1993!). The question is whether he should have moved from consideration to action.
The issue almost certainly came up because Miller had started the season slow, and when a player starts the season slow, the numbers stand out. Through three weeks or so, Miller was hitting .150/.190/.150. Brett wasn't tearing things up, but his .234/.261/.297 line was an improvement. When we talk about whether players look worse than usual, we (I) mostly have to defer to the managers and coaches and then decide whether we accept the weight those managers and coaches place on appearances. But in this case, McRae is actually taking the position that his player, Miller, was doing fine. And Miller had, in those three weeks, struck out just once. Between McRae's show of confidence and our own healthy suspicions of small samples and streaks, we should ignore the extremely recent histories.
Expanding our timeframe a bit, Miller had been a slightly better hitter than Brett the previous season, with a 107 OPS+ to Brett's 102, and with more of his value tied up (as we would wish) in his OBP. As a part-timer with the Mets in 1991, Miller had a 113 OPS+, while Brett's with the Royals had been 101. One might argue that Brett was likely even worse, because of his age, but both players were probably in the downslope portion of their career parabolas, so the age difference in this case isn't instructive. They were similar hitters, who were both in decline, but Miller seemed to be slightly better.
Ah, but the platoon advantage. Always the platoon advantage. Miller was a right-handed batter, Haas was a right-handed pitcher, and Brett was a left-handed batter. In 1993, the league-average platoon differential was about 10 percent of OPS for left-handed hitters; four percent for right-handed hitters; and seven percent for right-handed pitchers. Haas, Brett, and Miller had all shown greater splits over the previous two seasons:
If those two seasons represented true skill***, then Brett would be a .279/.335/.422 hitter against the right-handed Haas; Miller would be a .256/.314/.357 hitter. The edge, in this matchup, clearly goes to Brett. And, crucially, McRae's actions indicated he agreed with this: the previous day, against a right-hander, he had started Brett but not Miller. The day before that, both had started against a right-hander; and the day before that, he started Miller but not Brett against a left-hander. He was platooning these guys. And, in the seventh inning, the platoon should have switched.
Except that, as McRae seems to suggest in his muddled explanation, Haas probably wouldn't have been left in to face Brett. "Miller's starting the game, he's playing against left-handers, Bretty is not playing against left-handed pitchers, why in the f--- would I bat for Miller?" I think, if I'm correctly parsing the words I can actually hear, that McRae is saying Brett didn't start the game because he can't be trusted against left-handers; so, if a left-hander is likely to enter the game to face him, why would McRae send him up there?
Indeed, after Haas retired Miller, it was a lefty—Bob MacDonald—who started the next inning. So now McRae was faced with the decision of whether it was better to let Miller face a righty or let Brett face a lefty. It was a close call! It was worth considering! But, unless he wanted to burn another pinch-hitter then and there, it was more an issue of roster management than in-the-moment strategy.
Because, as he also seems to explain, "I didn't want to get myself into a situation where all I had to go after him was all right." I'm speculating a bit here, but I think he meant that the Tigers would bring closer Mike Henneman into the game at some point, and he wanted his left-handers on the bench to be available for that. The benefit there is that he knew the Tigers wouldn't replace Henneman, so he would be sure to get the platoon advantage. And, sure enough, the Tigers did bring in Henneman in the ninth, and McRae did go to Brett (and also lefty pinch-hitter Chris Gwynn), and Brett did double, and the Royals did score some runs. So, success!
Maybe. The value of Brett's double by that point was quite small, because the bases weren't loaded and the Royals weren't going to win anyway. It moved their win probability from 5 percent to 12 percent; a double against Haas (or MacDonald) would have been about a 23 percent gain. Of course, a double against MacDonald would have been less likely.
Would a double against MacDonald in the seventh have been likely if the Royals had responded to a pitching change with another pinch-hitter, for Brett? The new pinch-hitter likely would have been Hubie Brooks, who had batted just .216/.247/.337 in 320 plate appearances for the Angels the year before. (Switch-hitting Felix Jose had already been used as a pinch-hitter earlier in the inning, and Gwynn and back-up catcher Brent Mayne were both lefties. Brooks did pinch-hit in the eighth, with two on and trailing by four.) Brooks' 1992 stats were a career-low, so some bounceback could have been expected, but it's understandable that McRae wouldn't want to burn two pinch-hitters—his best one (Brett) and his only right-handed one (Brooks) just to get the old and tired Brooks up against a lefty.
So, to summarize, McRae had three options:
What it all comes down to is not Miller, or Brett, but Brooks. If you think Brooks is an average or better hitter, as he had been from 1984 to 1991, then you orchestrate the Brooks-against-MacDonald at-bat. Simply, there's likely not going to be a bigger moment in the game, and there's little point saving Brooks (and Brett) for lower-leverage opportunities. But if you think that career history is less important than the recent past, and that Brooks is just overpaid, washed up, and helpless against a good fastball, then you don't bother. You take your chances with Miller.
The Royals would finish in third place, 10 games back. Alan Eskew, the reporter who was bleeding as he left the office, was treated in the trainer's room. His vision was blurry for a few hours, and he went to a hospital to get a tetanus shot, but he was not seriously hurt.
*A week later, Bolton got the first out of the fifth inning against the Twins, then unraveled. He gave up a walk and a single, then a runner reached on an error, then two more hits. In his next start, he gave up five hits and didn't retire a batter. The next outing, in relief: two hits and a walk, no batters retired. That's 13 batters in a row reaching base. Twelve runs scored in that stretch, and his ERA went from 1.35 to 8.10. He was out of the majors by the end of the next season, which is true of a lot of the people in this piece: Brett retired in 1993; Haas was out of the league after 1993; Brooks after 1994; and McRae was fired after 1994, though he got to manage 196 Rays losses a few years later. Miller hung on through 1995, but got just 32 plate appearances in his final two seasons.
**It's possible McRae's anger was inspired not by the Miller/Brett question, but by the previous one, asked by the same radio reporter: Should Brian McRae, batting with the tying runs on and nobody out in the ninth inning, have bunted instead of swinging away. Brian McRae lined into a double play. Good question!
*** Of course, two years of splits aren't nearly enough to demonstrate a trend. Regressing those to the league averages (based on the work in The Book), we get smaller expected differentials:
Haas: 7.4 percent