The Wild Card game starter: Balfour, Doolittle or Cook?

The talented Dave Cameron over at FanGraphs brought up a thought-provoking point: why waste a top starting pitcher in a Wild Card game?

Often, a team's most effective pitcher doesn't start the game, he pitches the 7th, 8th or 9th inning. Why risk having your best pitcher not come into the game until late? It's an idea just crazy enough to work, but Bud Selig's Blue Ribbon Committee will decide soup or salad before a MLB manager tries it in this era.

Cameron's guinea pig for this experiment was Atlanta Braves, who have clinched a Wild Card spot. He assumes that manager Fredi Gonzalez might want the emerging Kris Medlen or the veteran Tim Hudson to start a potential NLDS Game 1. So why waste that first start in a play-in game, when Atlanta has a great bullpen featuring the best reliever in baseball (according to FanGraphs): Craig Kimbrel?

It's pretty well known at this point that relief pitchers perform better than starting pitchers, as the ability to air it out for 15-20 pitches leads to increased velocity and better stuff for most pitchers. Additionally, relievers more often face same-handed hitters, so they get a larger advantage from platoon splits, and they hardly ever face the same batter more than once in a game. These advantages add up in a hurry, which is why quality relief pitchers can sustain performances that even the best starters can't come close to.

Except for in a few cases - Justin Verlander in Detroit probably the most notable one - the team's most effective pitcher is hardly ever a member of the starting rotation. In reality, the more innings you can allocate to good relief pitchers, the less likely the opposing team is to score runs. Starting pitchers are essentially a necessity born out of the need to play many days in a row without a day off, but that restraint doesn't exist in the play-in game.

Cameron suggests that the Braves start off with Kimbrel as the starting pitcher, giving him a max of two innings, depending on how high-stress the end up being. From then on, Gonzalez can mix and match for matchups. For the latter innings, he could turn to starters Medlen (if Hudson is planned to start NLDS Game 1), Paul Maholm or Tommy Hanson. This has the added benefit of not allowing the opposing team to set up a lineup optimized by the handedness of the starting pitcher.

So, let's assume a not-so-worst-case, completely hypothetical scenario: Texas takes the next two from Anaheim and ends up clinching the division at the Coliseum (which I'd hate to see). Oakland hangs on to one of the Wild Card spots, but they'd know that the play-in game is their destiny by Tuesday. This gives them Wednesday to throw Tyson Ross, Jesse Chavez, and other pitchers who probably won't make the postseason roster. The A's could withhold their most effective relief pitchers, leaving them fresh and rested for Friday's Wild Card game.

Here's the A's pitching lineup through the rest of the season:

• Saturday vs. SEA: Dan Straily
• Sunday vs. SEA: Tommy Milone
• Monday vs. TEX: Jarrod Parker
• Tuesday vs. TEX: Travis Blackley
• Wednesday vs. TEX: A.J. Griffin (if Oakland knows they are in the WC game by this spot, and knows that they'll be playing at home or at Baltimore/New York, this is where I'd start Tyson Ross)

If you're still reading this, I'm well aware that this will not happen unless Bob Melvin enters the priesthood this weekend and Billy Beane gives Dave Cameron the job of managing the A's. It's just worth a little thought.

It's interesting because none of the A's starting pitchers have thrown a postseason pitch. Oakland's most experienced pitcher is the guy at the end of the bullpen — Grant Balfour. Melvin could start Balfour for 1-2 innings (again, dependent highly on stress), then turn it over to some sequence of Ryan Cook, Sean Doolittle and Jerry Blevins to get through most of the game. If the A's need to get two or three innings, Travis Blackley has been slightly more effective as a reliever than a starter (starting FIP: 4.14, relief FIP: 3.92). Assuming the A's want Parker or possibly Milone to start a potential ALDS Game 1 if Brett Anderson isn't ready to go, Straily would also be available for long relief.

As an astute commenter in the FanGraphs story pointed out, this has happened before in a postseason game. In the 1990 NLCS Game 6, Pittsburgh Pirates manager Jim Leyland gave the start to reliever Ted Power, who had not started a game all year, and was by FanGraph's WAR the fourth-best reliever on the team. Infact, it was Power's lone postseason appearance of his career.

While Power pitched well, going 2 1/3 innings, giving up three hits and a run, he didn't get any help from Pittsburgh's offense, which scored just one run that day in the loss.

Is it crazy enough to work, or just plain crazy?