Thursday, August 12, 2010

You Heard It Here First

Even though the Mets have entered the phone-it-in stage of the 2010 season, my wife and I found ourselves watching a couple of very good pitching duels between the Mets and Rockies the last two nights. First it was Mike Pelfrey outgunning Ubaldo Jimenez (17-2 coming into the game) to win 1-0. That game hinged on a dubious move by Rockies manager Jim Tracy, who apparently forgot to bring the scouting reports along on a Monday visit to the Hall of Fame. He didn't realize that Carlos Beltran (batting for Pelfrey in the 7th inning of a scoreless game) is the second-coldest hitter in the majors (second to teammate David Wright), gave him an intentional walk, and let the next batter win the game with a sacrifice fly.

It was a different story last night. Jonathan Niese pitched a beautiful game for seven innings, allowing a trio of harmless two-out hits and striking out seven over the first six innings. In the 7th, a couple of singles put him in trouble, two outs plated one run, and Niese finished off the inning with a 2-1 lead over Jeff Francis, who allowed just one hit after the 1st inning.

So we were surprised when Niese didn't come out to pitch the 8th inning. With the way the Mets "set-up" relievers have blown up lately, here was a pitcher who looked great out there on the mound, so why take him out? Did Jerry Manuel have a premonition that Niese might tell him later that his leg was a little sore? Did he think that just because two relievers had gotten the team safely through the final two innings the night before, it would automatically happen again?

Or was he simply subscribing to the supposedly un-second-guessable 21st-century manager's strategy of being satisfied with seven good innings and letting the bullpen handle it from there? Gee, he could say, Niese's pitch count was up to 106. He had given up a run in the 7th so maybe he was tiring. There were fresh arms ready in the bullpen. The new inhabitant of the 8th-inning role, Hisanori Takahashi, had breezed through the 8th inning the night before. How could you question the move?

Well, sports fans, I'm here to question it. I happen to have looked closely at this specific situation: 8th inning, your team is up by three runs or less, and your starting pitcher has gotten you through the first seven innings. Do you take him out? How far do you go with him? Does it matter? In the long run, which is the winning strategy?

How often does this situation come up? From 1950-2009, it occurred 35,809 times. That's a pretty good sample size. That's about 600 times a season, though it doesn't happen as often as it used to because more starters are being excused from class after the 6th inning. During the past decade, this situation has come up about 500 times a season. That's still enough to yield some conclusions that are--for my money--pretty damn conclusive.

The first key question is "how often does the manager let his starter continue into the 8th inning?" It used to be a no-brainer, in fact it was a no-brainer from 1950 all the way through the 1970s. If your starter was ahead after seven innings, he started the 8th inning more than 90% of the time. In the 1980s, that number dropped down to 80% or so. In the 1990s, the starter still got to work into the 8th inning two-thirds of the time. During the past decade, however, it has dropped below 50%, reaching a low of 40.7% in 2007.

More pitching changes are made in the 8th inning than in any other inning. Not surprisingly, more saves are blown in the 8th inning, too. In other words, by shoving their "best" reliever into the 9th-inning-only closer's slot, managers are trusting more games to their lesser relievers. Should they be shocked to discover that all that achieves is to blow more leads before the closer can get into the game and protect that save-worthy lead?

The second key question is: what happens when you let the starter work the 8th inning? Back in the old days (1950-1979), he finished the 8th inning more than 75% of the time. And guess what? If he finished the 8th, he got to start the 9th inning more than 90% of the time! Even as late as 1988, he started the 9th inning 83.2% of the time.

This figure has declined even more precipitously than the one about getting to start the 8th inning. That is, even if a manager lets his stud starter pitch the 8th inning, he is increasingly inclined to bring his closer in to get those last three outs. Only once in the last dozen years have these starters gone into the 9th inning more than half the time, and the overall percentage for the last decade is 42%.

Note that the 42% figure represents the number of starters who finish the 8th inning and get to start the 9th. But fewer and fewer are getting to start the 8th, much less the 9th. If you look at the number of starters who get through the 7th inning and are still out there in the 9th, the seismic shift in strategy becomes even more apparent. Through the 1970s, 69.1% of them lasted into the 9th inning. From 2000-2009, only 12.9% did so. We've gone from more than two-thirds to roughly one out of eight.

That brings up the third key question, and the most important: has this shift translated into more wins? Good question. When I started looking into this, I assumed that it must. After all, why would modern managers, who have an infinity of statistics available to them, universally adopt this strategy if it didn't produce more wins? Although it makes life easier for them to do things that produce plausible answers for the press, the bottom line is winning. If deploying an array of relievers didn't produce more wins, why were they all doing it?

That's what I thought. But no. It turns out that this universal strategy is producing fewer wins. The most accurate way of measuring this is to look at the number of wins recorded by the starting pitcher. I also tracked team wins, but that is less relevant. After all, when Jerry Manuel removed Niese last night, he wasn't thinking, "okay, even if my bullpen blows the lead we still have a chance to win." He had to be thinking, "my bullpen has a better chance of holding this lead than Niese does." Whether it was because the relievers were better pitchers or Niese was tired or the batters due up had good records against him or whatever, he had some compelling internal logic telling him that his relievers were more likely to hold the lead.

Too bad the numbers tell a different story. Let me attempt a simple chart to explain what I found. It is arranged by decade. First comes the number of times when the starting pitcher got through seven innings with a lead of three runs or less. Then I give you the number of times that starter got the win--whether he finished the game himself or had two or three or five or a busload of relievers hold that lead. Finally, there's the percentage.

1950-1959 4,825 3,746 77.6
1960-1969 6,108 4,784 78.3
1970-1979 7,401 5,749 77.7
1980-1989 6,420 4,896 76.3
1990-1999 5,816 4,408 75.8
2000-2009 5,239 3,988 76.1

For the first 30 years of this study, the starters won 77.9% of the time. For the past 30 years, they have won 76.1% of the time. That's a noticeable difference in a very large sample. In the world of formal statistical study, that is considered well within the range of normal deviation. In the world of baseball history, it is, I'll granted, slightly skewed by the pitcher-friendly decade of the 1960s. But since managers began reinventing pitching-staff deployment in the mid-1980s, the percentage has failed to improve. That's the bottom line.

Looking at team wins (including all the extra-inning games that wear down a pitching staff after a lead is blown), the percentage has stayed the same. From 1950-1952, when the starter in this situation got a complete-game win more than 60% of the time, the team winning percentages were 82.7, 84.4, and 83.5. From 2007-2009, when the complete-game figure plummeted to a mere 8%, team wins were 83.1, 84.7, and 82.8. The best percentage I found was during the 1970s, the worst during the 1990s; as with the starter wins, the numbers fell within a 2% range.

What's the point? Relying more and more on relief pitching is not a winning strategy for managers. Tying up your roster with 12 or 13 pitchers does not produce more wins. It doesn't produce more effective pitching, and it deprives the manager of extra options off the bench. The solution, in theory, is simple: train your starters to go further. They're your best pitchers, and getting them to carry more of the load will take the strain off your inferior pitchers, allow you to jettison the two or three worst arms, and give you that extra pinch-hitter when you really need one.

Of course, theory and reality are two different things. The modern managerial geniuses who have led their generation of leaders into this quagmire might protest that it is simply too hard for starters to work as many innings as they used to. If that's the case, it's because they aren't being trained to go further.

The discussion of pitch counts and innings restrictions, the coddling of young starters because of the delusion that arm injuries can somehow be prevented, and the myriad factors which caused baseball offenses to explode in the 1990s, will remain beyond the scope of this article. They will be covered in the book I'll eventually write, titled Why Today's Pitching Sucks. For now, just remember the bottom line: even though managers make way more pitching changes than they used to, all that maneuvering does not win more games. Treading water may help them survive and keep their jobs, but it won't get them to the shore where the championships lie.

No comments: