Tuesday, April 28, 2009

A Tale of Two Ballgames

I haven't written about it much on this site, but my favorite subject for research and analysis is relief pitching. Thanks to David Smith of http://www.retrosheet.org/, I have extensive data for every major league relief appearance for the past 55 seasons, and when I get a chance I like to play with the numbers compile data, and try to figure things out. In particular, I try to track the trends in bullpen usage and success/failure in the "saves era," the now-four decades since the Save became an official statistic in 1969.

No aspect of baseball has changed as much in the last 40 years as relief pitching. In the 1960s, it was standard for a manager to hand the ball to his starting pitcher with the expectation that he would pitch well enough to complete the game. If he got in trouble, he was supposed to get out of it himself, unless it was very late in the game. The starter carried the load. Today it's just the opposite. When the game starts, the manager is hoping he'll be in a position to use his closer in the 9th inning to protect a lead. He also wants his main set-up guy to pitch the 8th inning, and he has two or three other relievers slotted for the 7th inning. He figures if his starter gives him six decent innings--the parlance is "keeps us in the game" and "gives us a chance to win," the bullpen will carry the load in the late innings. So instead of pushing the starter forward as long as possible, today's manager works from the end of the game backwards toward the start.

What my research has showed is that this ass-backward approach doesn't work. There are more blown leads than ever, and they don't occur in the 9th inning, when the team's best reliever is working. More leads are blown in the 8th inning than any other, followed by the 7th inning. That's because the inferior relievers are relied on to get through those "transition innings," the crucial innings between the starter and the closer. In the 1960s, the standard pitching staff had ten pitchers; today there are a dozen, and managers crow about how each pitcher "knows his role" and performs better because he comes to the ballpark every day knowing exactly what is expected of him.

In a word, that's bullshit, and I'll tell you why. It isn't because the pitchers don't know their roles today. It's because those roles are tailored for mediocrity. The less we expect of people, the less we'll get from them. By expecting less from each of the dozen pitchers on his staff, today's manager gets less--much less all the time. Two games that were played last week illustrate this point perfectly.

On April 21, the Giants hosted the Padres, and the Giants took a 6-2 lead after six innings. Their starter, Matt Cain, had thrown 102 pitches, about average for a starter these days, but with a four-run cushion could've gone another inning or two. Giants manager Bruce Bochy elected to replace him with Brandon Medders, a 29-year-old career reliever who has never started a game in the majors and has averaged a little over an inning a game. He's a typical "role" pitcher you see in all bullpens, good for getting you through the 6th or 7th inning. He got the first out and then allowed a single--and Bochy took him out! What message does this send to Medders? "I don't trust you to overcome one baserunner. Your role is to pitch until you put somebody on base, and then I'm getting you the hell out of there." I have news for Bruce Bochy: baserunners are not the enemy. Batters get on base all the time, but not that many score. Last year, just over one-third of all batters got on base. In the National League, there were 33,742 runners. Only one-third of them scored. It is rare to see a game where a team has more runs scored than runners left on base. Putting one runner on base with a four-run lead is not a red flag. The red flag is the manager yanking the pitcher, implicitly telling him that he isn't good enough to protect the lead.

Am I being too tough on Bruce Bochy? Stay with me as we go through the rest of the game. Bochy replaced Medders with Jeremy Affeldt, a more experienced reliever, albeit one who is now pitching for his fourth team in the last four seasons. Affeldt got through the 7th inning without allowing a run, reinforcing Bochy's belief that he had made the right decision, even though Medders might be wondering why he wasn't given a chance to do what Affeldt did, considering that he had only thrown eight pitches before exiting. The score was still 6-2 Giants when the 8th inning began, and Affeldt trotted out there to continue his work. He got the first batter, gave up a hit--and yes, there went Bochy with the hook. This wasn't as extreme as yanking Medders--Affeldt had retired three batters--but he had thrown only 20 pitches and still had a safe lead.

In came Bobby Howry, one of the most experienced and respected middle relievers in the majors. A 35-year-old in his twelfth season, Howry is approaching his 700th appearance, every one of them in relief, averaging a little over one inning per game. Sure enough, it took him only six pitches to get the two outs needed to polish off the 8th inning. In the bottom half, Bochy pinch-hit for Howry. I would say he didn't need to do that. One thing I've researched closely is what happens when a pitcher finishes the 8th inning without allowing a run, as Howry did. Until the mid-1980s, if a pitcher went unscathed in the 8th, he was permitted to start the 9th inning more than 90% of the time. That figure has plummeted steadily ever since, and in the last few years it happens only 10-15% of the time. Yet the number of leads blown in the 9th inning has remained constant. The message there is simple: when you bring in a pitcher who shows in the 8th inning that he has good stuff (as Howry surely did), there is no need to bring in a new pitcher (with uncertain stuff) to pitch the 9th. With a four-run lead, Bochy didn't need to do anything but let Howry finish the job.

Bochy no doubt considered himself a genius when the pinch-hitter doubled, igniting a two-run rally which made the score 8-2. My problem with this is simple: the fewer pitchers you use today, the more you'll have available tomorrow. If you let your pitchers finish the job, you might find that you need only ten pitchers on your staff, maybe eleven, giving yourself more bench depth with that extra player or two. Here's the relevant perspective from my research. We'll look just at the National League for a moment. From 1969-1974 (the first six seasons after saves became an official stat), there were 17,914 relief appearances. Today there are one-third more teams (16 rather than 12), so at the same rate there would be 23,885 appearances in the past six seasons. How many have there been? Well, 45,667. Nearly twice as many! Do you think every bit of that increase is due to more offensive firepower? Of course not. It's strategic, it's managers feeling that because they have more relievers in the bullpen, they have to keep giving them work to keep them sharp (even though their performance is not demonstrably improved by this system).

Here's another relevant stat. From 1969-1974, 25.4% of those relief appearances occurred when the pitcher's team was winning. It makes sense. If you're winning, your pitchers are doing well enough, and managers back then didn't panic unless there was an imminent crisis. They didn't pull a reliever who gave up one hit with a four-run lead. What is the figure from the past five seasons? Are you ready? It's 43.4%. More pitchers are being used today when the team is winning than when the team is losing; from 1969-1974, more than twice as many were used when the team was losing. The game in San Francisco is a perfect example of why this is so.

To start the 9th inning, Bochy put in Justin Miller, a 30-year-old with about the same major league resume as Medders. He had a six-run lead. Bear in mind how often teams score six runs in an inning: almost never! In ridiculous blowouts where some outfielder volunteers to pitch the 9th inning, even that non-pitcher doesn't give up six runs. I can't even measure how much confidence I'd have that a pitcher who had allowed an average of four runs in nine innings over the past two seasons could get me through the 9th inning with a six-run lead. But I can tell you this: I'd have a lot more than Bochy does.

Miller faced exactly four batters. He got two of them out, walked one, and gave up a hit, leaving two runners on base with one out remaining to end the game. You know what happened next, don't you? Yes, Bochy took him out! How would like to be Miller's family, or his agent, or anybody who cares that he's a human being with an ego and a fluid commodity like confidence at stake? What is Bochy telling you? Ohmygod--two runners on base! Get him out of there. Let's trot another pitcher out there, let's give someone else some work. So in came Alex Hinshaw, a 26-year-old lefty embarking on his second season in the majors. In other words, the pitcher who has staked a claim to the lowest rung of the Giants' bullpen ladder. Yes, let's build his confidence--that is, if you consider protecting a six-run lead for one out a spur to confidence. At that, it didn't happen. The first batter Hinshaw faced singled in a run, so in that sense he failed. He got the next guy, ending the game.

Was anything gained by using five relievers to log three innings with a four-run leadm (and two of them with a six-run lead)? You tell me. Does the phrase "using a howitzer to kill a squirrel" ring true? The next day, in a tie game, Howry came in to pitch the 8th inning, got one out, gave up a hit, and heard the thundering footsteps of Bruce Bochy approaching the mound to replace him. Affeldt got through the inning, and the Giants used one more reliever before winning the game. Then they had a day off, so Howry and Affeldt could have finished off both games with no trouble or overtaxing effort, meaning the presence of Miller and Hinshaw was superfluous.

Let's contrast this with another recebt game from April 20, with the Reds visiting the Astros. In this one, Reds starter Bronson Arroyo went seven innings before leaving for a pinch-hitter with a 4-3 lead. Manager Dusty Baker brought in Arthur Rhodes to work the bottom of the 8th. Rhodes has been in the majors since 1991 and is 39 years old, but he missed all of 2007 after arm surgery, and in 2008 found a new role (with the Mariners and then the Marlins) as a lefty-lefty specialist. Every team has one or two of these guys, lefties who are tough on lefties but seldom trusted to pitch to right-handed batters. In 2008, Rhodes pitched in 61 games, and in 21 of them faced only one batter. Only 14 times did he pitch a whole inning; for the season he recorded 106 outs, an average of 1.74 per game. It was a successful season in which his ERA was 2.08, the second-best of his career.

Now he's with the Reds, and he began the 8th inning by yielding a double to a pinch-hitter. I happened to be following this game on my computer, and I predicted what would happen next. The Astros' leadoff hitter, Kaz Matsui, would bunt the potential run over to third base, and that would be it for Rhodes with a succession of right-handed hitters due up. Not just any righty hitters, but a cluster of formidable hitters at the heart of the Houston lineup: Miguel Tejada, Lance Berkman, and Carlos Lee. Berkman and Lee had already homered in the game.

Sure enough, Matsui bunted the runner over, and I waited for the pitching change to show up on the screen. But no, Rhodes stayed in to face Tejada. I've been surveying people about this move, and the consensus is that out of 30 major league managers, about 25 of them would've brought in a righty to face Tejada. Not Dusty Baker. Note that Rhodes had pitched a full inning the day before, throwing 19 pitches, so he wasn't exactly well-rested. Not surprisingly, he walked Tejada, and now he had to come out. I was pretty shocked when Baker left him in to face Berkman. Granted, Berkman, a switch-hitter, has a better track record against right-handed pitchers. I didn't think Baker would press his luck with Rhodes, and he had one of those lefty specialists available in the bullpen. But he also had some faith in Rhodes, and it was justified when Berkman popped out.

That brought up Carlos Lee, a very tough righty looking at an invitingly short home-run porch in left field. No way would Dusty leave Rhodes in to face Lee. But he did! Another righty hitter, Hunter Pence, was on deck. A flock of righty relievers stood perched in the bullpen. They stayed there as Rhodes walked Lee to load the bases. Okay, now Rhodes had thrown 18 pitches, making 37 in two games, enough to sap the stamina of most relievers. We know that Bruce Bochy or Tony LaRussa would have been up to his second or third pitching change of the inning. Did Dusty Baker waver? Did he go out to the mound and tell Rhodes, "sorry, Arthur, I gave you a chance, but I'd better bring in a righty now"? Nope. He stayed with him. I was more than shocked and, as a Reds fan, fairly dismayed as well, with Rhodes now one hit away from being the losing pitcher.

Pence worked the count full, fouling off pitches and fouling off more pitches, taxing Rhodes' 39-year-old arm more and more and increasing his chance of getting a hit more with each pitch. On the tenth pitch of the at-bat, Rhodes fanned him. Amazing! Forget about 25 out of 30 managers not letting him pitch to Tejada. Perhaps only with Dusty Baker would Rhodes have still been out there three batters later with the bases loaded.

It's important to emphasize that this happened only a couple of weeks into the season. This wasn't like Jerry Manuel last September, trying to plug the gaping holes in the dike known as the Mets bullpen, where everyone has proved unreliable. April is the time to build up a team's identity and each player's sense of contributing. The manager challenges his players and finds out who can and cannot perform. That's what Baker did with Rhodes. I don't know what Bochy achieved apart from finding out that he still doesn't know which one of his middle relievers can pitch an entire inning.

Rhodes came through for his manager, and the Reds closer, Francisco Cordero, pitched the 9th inning to close out the 4-3 victory. Baker wound up using two pitchers to hold a one-run lead, compared to Bochy using five pitchers to hold a four-run lead. Bochy convinced his pitchers that if they waver a bit, give up a baserunner, he doesn't have faith that they can continue effectively, while Baker convinced Arthur Rhodes that he can get anybody out at any time, no matter how dangerous the situation. No wonder Baker developed a reputation--while managing the Giants--as a manager players loved playing for. If I were Arthur Rhodes, I'd love to be pitching for Baker. Come to think of it, if I were Brandon Medders, Bobby Howry, and Justin Miller, I'd be wishing I could pitch for Baker, too.

1 comment:

Ken said...

http://radicalbaseball.blogspot.com/

SATURDAY, MAY 2, 2009

Yanks provide more evidence that my pitching theory is correct.
Yankee manager Joe Girardi is sparking interest in my first talk radio program (see previous post) based on my blog entry here in February 2008: Thinking Man's Guide to Managing a Pitching Staff.

The basic idea is to get the optimum number of pitches from the best pitchers available in a particular game. The simplest and most conventional way of doing this to use your best relief pitcher first, not last. Girardi continues to show that he is no more innovative than his predecessor, Joe Torre, but without Torre's World Series championships.

Last night was May 1, 2009. It rained during the entire game, which could have ended early because of the rain. Even that did not inspire Girardi to use his relief pitchers in a different order. Yanks won 10-9 with four runs in the eighth and two runs in the bottom of the ninth, so Girardi lucked out. Here is the offending episode.

Top sixth, Yanks lead Angels 4-0.

Angels score 2.

Bases loaded, two out.

Starter Andy Pettitte is replaced by rookie Mark Melancon.

Gary Mathews hits first pitch for three run triple.

Abreu walks.

Melancon throws wild pitch allowing Mathews to score.

Hunter lines out to right.

Veras started the seventh and allowed three runs, retiring only one batter. 9-4 Angels after seven.

___________________________________________

Prior to last night's game Melancon had pitched three innings in his career: two innings April 26 and one inning April 28.

Melancon allowed all three inherited runners to score, plus one of his own. Four runs. One batter retired. Girardi's move was a disaster.

Girardi never considered using Mariano Rivera in the sixth inning with the game on the line. Who is more likely to get out of that bases loaded jam: Rivera or Melancon? Who is more likely to benefit by starting an inning: Rivera or Melancon?

Rivera might have retired that final batter and prevented those three inherited runners from scoring. He then could have pitched the seventh inning, maybe part of the eighth. Maybe the rest of the game would have been rained out.

Girardi never got to use Rivera because Girardi followed the rigid formula that all MLB managers follow: the best the relief pitcher, known as the closer, almost always pitches only the ninth inning, only with a lead. MLB managers maximize the number of pitches thrown by the team's worst pitchers, the middle inning relievers. MLB managers should bring in their best relief pitcher first, then the next best and so on. The idea is to not use the worst pitchers at all. Instead MLB managers start with the worst hoping to get an opportunity to use the best. MLB managers also put the worst pitchers into the most challenging situations. MLB managers should put the best pitchers into the most challenging situations. Starting the ninth inning is usually not the most challenging situation in most games.