Lies, Damn Lies and Statistics
I appreciate the work they do over at fangraphs.com, and I recognize that there are a lot of smart people over there doing some very interesting work. But sometimes I come across some number they’ve crunched that just makes me apoplectic.
While researching today’s earlier post on KBO pitchers being posted this off-season, I happened to learn that fangraphs values Hyun-jin Ryu‘s 2014 regular season performance at $19.0 million. He had a good year as the Dodgers’ third starter, but he missed some time with injuries, so the valuation seemed high.
It made me wonder how Ryu’s purported value compared to that of Giants’ ace Madison Bumgarner. Fangraphs values MadBum’s regular season performance at $19.9 million.
These two numbers just can’t be accurate. Did someone at fangraphs input the wrong numbers into one of the formulas, or does fangraphs really think Ryu’s regular season performance had 95.5% of the value of Bumgarner’s? Bumgarner threw 65 more innings with an ERA four-tenths of a run lower. Bumgarner was the undisputed ace on his team, while Ryu was clearly the Dodgers’ third best starter.
Even assuming that Dodgers Stadium was a much better hitter’s park in 2014 than AT&T Park was, I don’t see how that could have been anything other than a one year fluke. For the seven seasons from 2007 through 2013, the two ballparks were almost identical pitchers’ parks.
I’d like to believe that fangraphs’ numbers are accurate, since it would be nice to know what players’ performances are really worth. It would also be nice to have a clear sense about how much a player’s defense contributes to team wins. The problem is that I come across numbers like these too often to look at fangraphs’ numbers with any great sense of confidence in their accuracy.