Thursday, June 21, 2012

Defending the Aardvark (And Last Night's Game)

Process or results. An aging veteran towards the end of his career is signed and produces a season beyond expectations. Was it a savvy move by the team's General Manager? Perhaps the GM realized decline phase athletes went from over-priced commodities to under-priced. Or perhaps the GM concocted his value out of perception: "I can't believe THIS GUY is available. He's got over 350 career home runs! We're signing him!" It is never as simple as either case, but potentially more important than the why is the question of does the why even matter. Again: process or results.

A couple of years ago I read an interesting article at the Hardball Times which addressed this very concept in relation to the San Francisco Giants GM Brian Sabean. Sabean is constant fodder for many sabermetrically inclined individuals, mostly for good reasons in my opinion. His career MO for a long time had been to sign veteran players to large contracts beyond their value (at least according to the SABR community). Many believed he didn't extract nearly enough value - in terms of playoff success - from the Barry Bonds led teams. And until Pablo Sandoval came along, the farm system for the Giants didn't produce what you could consider an above average offensive position player (sorry, Pedro Feliz). I think these are fair criticisms, but the irony of it all is the article was written in November 2010. If you are missing the relevance, it was just a week after the Giants needed only 5 games to dispatch the heavily favored Texas Rangers in the World Series. That was Sabean's 14th season with the team, the 2nd World Series appearance during his tenure (first win), 5th playoff appearance, and 10th winning season. I think you would be hard pressed to find anyone who would argue his 14 years with the Giants were unsuccessful. Again, you could say they should have been more successful, but that is a different discussion.

So Brian Sabean teams had good results even if his process was considered suboptimal* to some. Last night the Phillies beat the Rockies in somewhat dramatic fashion, needing one of the best first baseman of the last 25 years to suddenly forget how to play the position. If you read my post yesterday, you will realize the simulator thought the Rockies were a really good play. It calculated that last night's game was a bit over a coin-flip (53%) in favor of the Phillies, despite the line (and at least one well thought of simulation) putting it at around a 60/40 game. It is impossible due to the outcome of one game to calculate the future probability one team will beat another, but for most of the game - especially going into the bottom of the ninth - it looked to be much closer to an even odds game than a 60/40 one. Ultimately the simulator was wrong - in a zero sum way - about which way to bet the game last night. And just because the game was very even in its entirety, doesn't mean the true odds for the game were closer to 50/50 than 60/40. But even though mathematically it cannot be proven which was closer to the truth, I'll say that the simulator presented a value which felt closer to the truth. The process made sense and produced an answer that seemed more sensible, especially after watching the game. The results just didn't come.

*There is an obvious question lingering out there which I will not attempt to answer: was his process suboptimal? I think the article does a pretty good job of addressing the major points, so I'll defer you there

No comments:

Post a Comment