Friday, September 30, 2011
Going to WAR for the Cy Young Award?
Posted by Derek AmbrosinoAs I was putting together my ballot for The Hardball Times postseason awards, I arrived at an intellectual impasse. The NL Cy Young Award has three very strong candidates, and while my instinct was to lean toward Clayton Kershaw, I was tempted to consult the numbers—not the cosmetic, mainstream stats, but the gold ol' BABIPs, and FIPs. I was tempted to go find out whether one of the worthy candidates benefited more from luck than the others.
But then I was struck with a question—is it appropriate to discount luck for these awards? What should we be rewarding, inherent performance or outcome?
I’ve long since comfortably reconciled the alleged false dichotomy of the word “valuable” in the context of the MVP; value can only have one reasonable meaning in that context: the highest absolute value beyond zero. But, the question of whether to consider what “should have” happened to a pitcher (or position player) introduces a new, more nuanced wrinkle into how one interprets best performance. Considering wins for example, clearly privileges (and inaccurately credits) outcome too far above underlying performance level—but where is the balancing point for what did happen and what should have happened? I know ERA+ should be privileged over wins in this context, but should ERA+ be discounted too, in favor of FIP?
While I’m not completely convinced that I will not change my mind on this, my thought is that what did happen is far more important in the context of determining the best player that what quite possibly should have. Outliers may be outliers, but that doesn’t mean that they didn’t happen. If, for example, that player got quite lucky with his home run/flyball rate, or strand percentage, I’m fine with rewarding that—those events still benefited his team and resulted in win value.
Crediting that is not on the same level as being swayed by W-L records. At some point, we have to recognize what actually happened. The guy who hits 10 improbable roulette bets does after all go home a rich man, even if he is a rich, dumb man.
One other observation I’d like to make here: I do find it somewhat telling that I had no temptation to look up the core indicator stats for position players. Perhaps that has to do with the fact that the choices for best position players in each league seemed clear cut (not to say Justin Verlander shouldn’t receive MVP consideration). Overall, though, this probably speaks to an inherent difference in the default degree of legitimacy we grant semi-advanced offense stats (like OPS+), and the greater scrutiny we feel obligated to apply to pitching stats. And, perhaps there’s some merit to that feeling.
Derek Ambrosino aspires to one day, like Dan Quisenberry, find a delivery in his flaw, you can send him questions, comments, or suggestions at digglahhh AT yahoo DOT com.
I agree that our considerations should focus on what happened rather than what should have happened. For example, if a Cy candidate had a .260 BABIP, we don’t KNOW exactly what his BABIP really should have been. There’s a very outside chance that it should have been .260 or lower. More likely, it’s somewhere between .270 and .300 with another outside chance of being higher than that. We can use xBABIP to narrow it down, but that comes with a sizable error term too. The result is that we have to fall back on our biased interpretations or flawed proxy stats to distinguish the correct number.
Meanwhile, we know exactly what physically happened. It’s much firmer data to play with.
Who were your obvious MVP candidates? I grappled with Kemp/Braun in the NL and Bautista/Ellsbury in the AL.