Nice article from Baseball Prospectus on closers:
Closers are an aberration in baseball's history, a massive misallocation of resources, and eventually will go the way of the dinosaurs Carl Everett doesn't believe in.
A pure closer is a reliever who only comes in to protect a one- to three-run lead, only in the ninth. The worst pitcher in baseball stands a great chance of pitching the ninth inning without giving up three runs. With no outs, a team with an average offense against an average pitcher can expect to score half a run. The best offense in baseball last year, the Red Sox, averaged about .65 runs/half inning over the course of the season. The worst reliever in the major leagues last year was Jaret Wright, who gave up 46 runs in just over 56 innings of work--.82 runs an inning. Given a three-run lead in the ninth, pitching against the Red Sox, Wright could reasonably be expected to give up an average of a run each appearance, and if he did it all season, he'd rack up 20 saves, be anointed a proven closer, and sign with the Mets for $4 million a year.
This makes no sense. Keith Woolner did a really cool study on how many runs get scored in an inning, and between 1980 and 1998, 94.6% of all innings had less than three runs scored; 87.8% had less than two, and 73% of all innings played saw no runs scored.
It's obvious then that having a team's best pitcher in with a three-run lead doesn't make much sense, because almost anyone could do that job. Why do teams insist on doing this?
Blame the save. It's a counting statistic that was intended to measure an event--whether that's an event worth counting we can debate another time. But the measurement of that event led to the development of a role: the closer, who pitches only when his team has a lead of one to three runs, and in so doing becomes eligible to pick up a save. This role, once established, then picked up baggage of its own: the guy has to throw really fast, be mean, a real gun-slinger with attitude...even though Jeff Montgomery was a fine closer with none of those qualities.
A pure closer on an average team will probably only see 40, 50 chances to come into a game and pitch the ninth. Woolner's study also found that a perfect closer--one who never gives up a run pitching the ninth every time his team is ahead--only gives his team four more games in the standings. By contrast, a good but not quite elite offensive player or starting pitcher is worth four or five games; Barry Bonds was worth 11.5 wins last season. Eric Gagne, who was about as perfect as a closer can be in 2002, was worth only three games last year over what you'd predict for the Dodgers if they'd replaced him with innings from normal pitchers--and Gagne actually pitched more than one inning more than 10 times last year.
Where should a team's best reliever come in, then, if not to protect ninth-inning leads? Our Rany Jazayerli has examined this issue, and concluded that in a neutral ballpark with essentially equal teams, below are the situations where a reliever makes the most impact on a team's win percentage. The impact is in parentheses, so if the team could be expected to win .600 of their games when up by one but .700 if they use their closer, then you'd see (.100).
1) Top 9th, lead by 1 (.170) 1) Bottom 9th, lead by 1 (.223)
2) Top 9th, tied (.160) 2) Bottom 8th, lead by 1 (.158)
3) Top 8th, lead by 1 (.123) 3) Bottom 9th, tied (.155)
4) Top 8th, tied (.115) 4) Bottom 8th, tied (.122)
5) Top 7th, lead by 1 (.096) 5) Bottom 9th, lead by 2 (.113)
6) Top 7th, tied (.092) 6) Bottom 7th, lead by 1 (.111)
7) Top 9th, lead by 2 (.080) 7) Bottom 8th, lead by 2 (.108)
This reaches back to Ye Good Old Days, when teams didn't have closers, they had 'stoppers', the firemen who came in not just when the game was close, but when the starter was faltering. These brave souls would come in not with a lead and no one on, but with the game tied and two runners on--situations where being dominant and getting the outs has a clear and significant impact on the game's final score. These were situations where Gagne might strike out three guys to get out of the jam unscathed, while Jaret Wright's predictable single scores two runs and costs the ballgame. The tighter the situation, the more important it is to have your best pitcher on the mound. We understand this instinctually--you want your best hitter at the most important point in the game--but for some reason the modern mystique around the ninth inning has clouded the judgement of managers across the league.
This is why modern bullpen usage is inefficient. It's like saving your best pinch-hitter for when you're behind by three runs, or only starting your best option at shortstop on days when there's a full moon because that's when things get crazy. Resources should always be deployed where they can do the most good, and modern closers as blood-lusting Gods of War, along with their Phobos/Deimos setup men (one lefty, one righty), are a bad use of resources.