It has been a few weeks since I wrote a blog, and I hope you've enjoyed my absence. I've been doing research lately on pitching, my favorite subject, and doing the research in my favorite way, which is to look at box scores on Retrosheet. This site, http://www.retrosheet.org/, has every box score and (for 99% of the games) batter-by-batter results from the present to 1953 (with more seasons added every year) invites long visits. It's like a chocolate factory without the demented proprietor, since David Smith, who created the site, has the altruistic motive of making the result of every batter in baseball history available free of charge.
Is it possible to have too much of a good thing? In the case of Retrosheet, no. I've spent more hours than I can count perusing the box scores (and other copious material) there, and it never ceases to fascinate me. There's a pattern to game scores, line scores, and box scores [that is, the final score, the inning-by-inning scoring of runs (or not scoring), and the details of which batters were responsible for the scoring (and the non-scoring). Even though I'm focusing mostly on pitching, I also notice and marvel at the offensive feats, the big scores and big comebacks.
Last week I found a box score which seems to be Exhibit A in the case against modern managers for overusing their bullpens. One of the truisms of baseball is that substitutions are risky. Even though there may be a perfectly logical reason for making a substitution, it does not follow that every supportable substitution should be made. Unless the player being replaced has been injured or performed so horribly that his continued presence on the field could produce only disaster, the new player is an unknown quantity. The best historical example is the 1951 National League playoff, where the Giants were rallying in the bottom of the ninth inning of the decisive game. Dodgers manager Chuck Dressen called down to the bullpen to ask which of two pitchers looked better warming up. Just then, the bullpen coach saw Carl Erskine bounce a curve to the bullpen catcher. He advised Dressen to put in the other guy, Ralph Branca. In came Branca, and two pitches later out went the rocket off Bobby Thomson's bat which made both himself and Branca famous. Who knows if Erskine would have done better. Maybe Thomson would've hit his first pitch into immortality.
A few years ago, I did a presentation at the SABR convention on some changes in recent decades in how bullpens are used. Here's one important thing my research uncovered: from the 1950s all the way through the mid-1980s, if a reliever entered the game in the eighth inning and got out of the inning without allowing any runs to score, he came back to start the ninth inning more than 90% of the time. That's just how it was done, and it explains why Hall of Fame relievers Rollie Fingers, Bruce Sutter, and Goose Gossage got all those two-inning saves. The two-inning save is a rare event today, because managers choose to divide the duty, using one pitcher for the eighth inning and, no matter how great he pitches, bringing in another guy to pitch the ninth. This strategy has become so widespread that the roles have coined new terms: "set-up man" for the eighth-inning specialist and "closer" for the ninth-inning finisher.
Let me repeat that main point. As recently as 25 years ago, if you got through the eighth inning looking strong, you stayed out there for the ninth inning. Period. If you got in trouble then, someone else would come in. But the manager saw enough of your stuff in the eighth inning to like your chance in the ninth. Did this strategy work better than today's specialization? No, not better. But the same. The percentage of saves and blown saves has remained roughly the same. It doesn't matter whether you use one or two pitchers to hold a lead in those last two innings. So why take up a roster spot for a pitcher whose role is redundant and unnecessary?
To put it another way: if it ain't broke, don't fix it.
It's time to present Exhibit A, a game played on May 25, 2001. The Tigers took a 2-0 lead in the bottom of the first and expanded the lead to 4-0 after four innings. Their pitcher, a 29-year-old righty named Chris Holt, had a no-hitter going. He took the no-no to the top of the sixth, when he allowed a pair of runs on a triple, walk, single, and sacrifice fly. Through six innings, he had a two-hitter with seven strikeouts, and a 4-2 lead.
His manager, Phil Garner, took him out. Of course, since we weren't there, we don't know whether there were special circumstances which compelled Garner to remove him. Maybe he was developing a blister or had twisted his knee or had a touch of the flu. Perhaps Garner looked at Holt's track record, a decidedly losing record in four seasons with Houston before joining the Tigers in 2001, gaudy ERAs, and only two starting efforts longer than six innings so far this season. The interesting thing is that Holt was in exactly the same position a month earlier, on April 26. Also at home, he took a 4-0 lead to the sixth inning and promptly gave up two runs on a single, a triple, and a passed ball followed by a single. What did Garner do that time? He let Holt start the seventh inning. No runs. He pitched the eighth inning, too. Three up, three down. By this time, the lead was 8-2, and out he went for the ninth inning as well. A pair of one-out might have alarmed Garner, but he left Holt in and was rewarded with two outs to finish off the complete game, the last of Holt's four complete games in 112 career starts.
With that performance fresh in Garner's mind, there might well have been some extenuating circumstance which made him not even think twice before removing Holt on May 25. He brought in lefty Heath Murray to face lefty Jeff Liefer--and struck him out. Despite that auspicious beginning, Garner lifted Murray from the game. Does this mean that Murray had precisely enough stuff to retire a seldom-used outfielder with a forgettable career, but the strikeout didn't suggest that he had enough stuff to retire the next batter just because he was a decent platoon player who happened to bat righty?
In came Matt Anderson, a righty, who quickly fanned Herbert Perry and disposed of Sandy Alomar, Jr. on a ground out. That was impressive, getting those two hitters with little effort. You'd think that would qualify him to start the eighth inning--remember this is the American League, where a manager can use his pitchers exactly how he wishes because he doesn't face the dilemma National League managers when they might have to take out a hot pitcher for a pinch hitter. Can you think of any reason why Anderson didn't deserve to continue after getting a strikeout and a little ground ball? I can't.
Phil Garner could. Two of the next three White Sox due up were switch-hitters, but the middle batter was a lefty, and that was all the excuse Garner needed to bring in a fresh lefty, C. J. Nitkowski. Never mind seeing if Anderson could continue his good work. There's one lefty in the next three hitters, so let's bring in the lefty. Well, Garner's move worked, sort of. The first switch-hitter, Jose Valentin, batted right-handed and struck out. The lefty, Chris Singleton, worked Nitkowski for a walk, but he came back to whiff the other switch-hitter, Ray Durham. That brought up Magglio Ordonez, the cleanup hitter, a righty. Here we've got Nitkowski, a lefty who just struck out the two right-handed hitters he faced. And here we've got Phil Garner making another walk to the mound and wave to the bullpen.
He brought in a righty, Danny Patterson, to face Ordonez. I don't understand it. Nitkowski just fanned two righties--oh, but they weren't real righties, they were switch-hitters, while Ordonez was a real righty. That's much, much different. In came Patterson, and he got Ordonez to ground out.
Let's review the situation heading to the ninth inning, with the Tigers still ahead 4-2. In the past two innings, they have used four pitchers who collectively faced seven batters, striking out four, walking one, and getting two ground outs. Garner had to feel pretty proud of himself, navigating his staff through those two tricky innings which formed the bridge from his starter to his closer. Eight innings worked by five pitchers who allowed two hits. All had pitched well, and he had dodged those bullets of uncertainty, found four relievers who came right in and did the job without any fuss.
Now it was the ninth inning and time for his closer, Todd Jones, who had blown a save in the ninth inning two days earlier, but was now--to use a favorite announcers' phrase--being asked to get right back up on the horse. Sure enough, Jones started like his teammates had, striking out the first batter he faced, the toughest out in the lineup, Harold Baines. The next two hitters singled, but Jones got Perry to fly to center and get the White Sox down to their final out.
Jones never got that out. Carlos Lee pinch-hit for Alomar and singled in a run to make it 4-3. Valentin singled in the tying run, landing Garner in a true predicament. After squandering four relievers who showed that they had the stuff to get people out, now his presumably best reliever, his closer, was proving that he couldn't get anybody out. But there was nobody to replace him except the two pitchers at the bottom of the barrel. So Jones stayed in.
The next batter was safe on an error--by Jones. I wasn't there, but I'm sure it was a hideous miscue, one so inexplicable that it rocked Jones off what was left of his moorings. With the bases loaded, two outs, and the score still tied, Jones served up a fat pitch to Ray Durham, who drilled a double to score all three runners. What more did Garner need to see? Well, he needed to see one more double, by Ordonez, before he got Jones out of there in favor of Kevin Tolar, whose major league career consisted of 20 games and a 6.62 ERA. Tolar got Baines to foul out, and the nightmare inning was over. So were the Tigers' chances. They went meekly in the bottom of the ninth and lost 8-4.
There you have it. Phil Garner found five pitchers who pitched well, and found reasons to take them out before they got in trouble. He kept taking pitchers out until he found one who got in trouble, stayed in trouble, and made things worse. That's the guy he left in. His first four relievers faced seven batters and got six out. Jones faced nine batters and gave up six hits. But he was the "closer" so there he remained to take his drubbing. If it's any consolation, he kept giving up runs over the next few weeks until Garner saw the light and replaced him in the "closer" role--with Anderson, the pitcher who retired the only two batters he faced in this game before being sent to the sidelines to watch his good work undone.
That, in a nutshell, is what is wrong with the way managers use their bullpens today. They do not trust their own eyes when they see a reliever pitch effectively. They place their trust in the "splits" which show the statistical tendencies of hitters against certain types of pitchers. As long as they have a good statistical reason to use this guy instead of that guy, they feel justified. My point is that if Garner had gotten another inning out of Holt, or had used just two relievers on the bridge to his closer, he would have had the other two (effective-on-this-night) relievers available to bail out Jones when he got in trouble and save the game instead of throwing it away.