With just a week and a half left in the regular season and most of the playoff races offering little in the way of end-of-season excitement, a lot of attention has been paid to the postseason awards. The rise in popularity of a lot of the metrics we host here on FanGraphs – especially WAR – has led to a lot of discussion about how those numbers should play into a voter’s decision making process. In many cases, those writing pieces about advanced metrics and awards voting have assumed that the statistical crowd is in lockstep agreement on the rigidity of how these metrics should be interpreted, and in most cases, the assumption is that we trust the numbers over all else.
That’s not really true, of course. The statistical community has a broad base of opinions, and in general, few if any serious analysts believe that awards voting should be done according to an ordinal ranking of WAR. We don’t believe that the numbers that we look at regularly are infallible – we just happen to think that they usually do a better job of accurately answering questions than any other statistic does at answering the same question. As Joe Posanski put it last week:
And this gets at exactly why I love WAR. It makes things more fun. Should we be a slave to it? Of course not. But, is that even necessary to say? I mean: should we be a slave to anything?
WAR’s rapid rise in mainstream acceptance has thrust arguments from the statistically-leaning crowd into a more prominent role this year, but in general, the nuances of how we believe WAR (or FIP, UZR, WPA, etc…) should we interpreted usually don’t make it into the conversation. Too often, the argument is framed as statheads who put complete and utter faith in their numbers against those who aren’t ready to accept them at face value yet. However, we don’t really hold the position that the metrics were designed for strict adherence. We believe that it’s a good idea to consider things like WAR or FIP when casting a vote, but you should consider other things as well. We don’t believe that anyone should vote based on sorting our leaderboards and calling it a day.
So, we’ve decided that it would be helpful if, as a site, we wrote up our “official position” on each award – not an endorsement of a specific candidate, but instead, a generally agreed upon thought process about how our numbers should play into the voting for that award. These pieces aren’t going to be campaigns for a certain outcome, but instead, we hope to explain how we think our numbers should be used to help inform the voting process.
Throughout the rest of the week, we’ll have pieces on each of the six “major” awards, talking about where we see the metrics we use most on FanGraphs fitting in, their limitations, and what we deem as responsible usage of those numbers in making decisions about who deserves a vote. We’ll also have a separate piece tomorrow about single-season UZR data, which we know is the most controversial piece of WAR’s calculations.
Through it all, our goal is essentially to clarify our generally accepted positions on the best way to look at the data found on FanGraphs. We do believe that the numbers we look at most often do contain interesting insights and reveal important truths about the game, but we also understand that they do not sum up the game in perfection. We don’t have to settle for an either accept-them-or-reject-them narrative, and hopefully, this series of posts will help show how our metrics can be used to improve awards voting without enslaving anyone to a particular number.
First up this afternoon will be a piece on the AL MVP voting – perhaps the award where our metrics are most often discussed. We’ll go award by award through the rest of the week, and we’ll wrap up on Friday with our staff’s selections for each award. It should be a fun week.