Today I’ve got a piece on a topic that’s been kicking around the Yankeesphere for the past 18 months, and one that’s been the subject of heated internal debate practically since the dawn of the Baseball Prospectus/ESPN partnership: should Joba Chamberlain be moved to the bullpen. Two pieces, in fact, since the ESPN version differs significantly from the BP version due to space and other considerations. Here’s a taste from the latter:

You can’t have too much pitching, as the old saw goes, and the weight of the evidence — a staff ERA of 4.88, 12th in a 14-team league — suggests that the Yankees don’t, despite their $200 million payroll. For the past year and a half, wags have pitched their solution: return Joba Chamberlain to the bullpen, clearing the rotation logjam while fortifying the bullpen with a top-flight setup man. Superficially, the move makes sense: once Chien-Ming Wang demonstrates full health and command, he and Phil Hughes can round out the rotation behind CC Sabathia, A.J. Burnett and Andy Pettitte, with Chamberlain resuming his late-2007 dominance as the bridge to Mariano Rivera, sans midges. But that notion rests on flawed assumptions.

That the bullpen is the Yankees’ bigger need is hardly clear. Though more expensive and better pedigreed than the staff’s other end, the Yanks’ rotation ranks seventh in the league in both SNLVAR and Fair Run Average. The bullpen is seventh in WXRL, and though their Fair Run Average is considerably higher than the starters (5.53 to 5.04)—second-to-last in the league, in fact—much of the damage has been confined to low-leverage situations; those 14 runs they yielded on April 18 once Wang departed trailing 8-2 meant little beyond mop-and-bucket duty for Nick Swisher. Though a few of manager Joe Girardi’s go-to guys — Jose Veras, Edwar Ramirez and Jonanthan Albaladejo — have been lousy, the skipper proved adept at weeding through his no-name relievers last year, and the team’s recent hot streak owed something to the emergence of similarly unheralded Phil Coke and Alfredo Aceves as viable late-inning options in the absence of top setup man Brian Bruney.

Furthermore, Chamberlain leads the rotation in strikeout rate (8.6 per nine), ranking second in ERA (3.71) and third in Support Neutral Winning Percentage (.532). Though he’s averaged just 5.3 innings per start (1.1 less than Burnett), that includes a line-drive-induced departure after just two-thirds of an inning. Excluding that jumps his average by half an inning while nudging the Yankees above the league average in the percentage of innings thrown by starters (65 percent), so it’s tough to argue he’s putting an undue burden on the pen. At worst, he’s been the team’s third-best starter, far better than Hughes (5.45 ERA, .496 SNWP and a shade under five innings per start), all while being paced to toss around 150 innings to avoid the so-called Verducci Effect.

The shift also assumes the setup role is a better match for Chamberlain’s abilities — and vulnerabilities. While he’s known no greater success in the majors than his 2007 relief stint, the role is beneath him. Although he’s struggled with his command at times this year, in part due to difficulties in pacing himself, Chamberlain is the rare possessor of three plus pitches. According to the Baseball America Prospect Handbook 2008, scouts grade his fastball, slider and curve at 70 or 80 on the 20-80 scouting scale, and consider his change-up solid-average as well. Rivera and his legendary cutter aside, most relievers survive on two pitches because they’ll only face each hitter once. With his deep arsenal, Chamberlain has proven his ability to retire hitters multiple times in one outing; those in their third or fourth plate appearance against him have batted a feeble .222/.306/.324, essentially equal to their first turn (.231/.299/.314), albeit in a smaller sample size.

…For a third point, the shift assumes that a top-flight setup man can be more valuable than a frontline starter. Our win expectancy-based pitching metrics (SNLVAR, WXRL and the associated leverage score) allow for direct comparison of each role’s impact in terms of wins above replacement level. In what we’ll call the Eckersley Era (1987 onward), 67 starters have finished the year with at least 8.0 SNLVAR, while just six relievers have reached 8.0 WXRL, all closers — and remember, Chamberlain won’t be closing. Lower the bar to 6.5 and the ratio is 225 to 30, with only two saving fewer than 29 games and thus working in lower-leverage situations: Rivera in 1996 and Rafael Betancourt in 2007. While the former is obviously the Jobaphiles’ model, the latter has been a disaster since that season, underscoring the risks of heavy relief usage. Expecting Chamberlain to live up to the best reliever in baseball history is a ridiculously tall order.

While the piece was in the pipeline — and while I was conducting the BP chat I somehow forgot to mention here — the Yankees announced that Wang would take Thursday’s start, while Hughes would move to the bullpen, a course of action I suggest further down in the piece, noting that Hughes has compiled a 5.22 ERA through 28 starts, and that the Yanks’ best course of action is to acclimate him to getting major league hitters out the first time rather than worrying if he can do it a third.

The ESPN version thus carries the following intro: “Shortly after Jay Jaffe filed this piece explaining why the Yankees should keep Joba Chamberlain in the rotation and move Phil Hughes to the bullpen, the team announced its plan to do just that. Coincidence? You decide.” Nice.

Choice chat cuts in the next post…