Statistics in baseball are fun. Behind each statistic – conventional and advanced – is an idea, a little hypothesis about what really matters in baseball, and which aspects of performance tell us that a player is a great hitter or great pitcher or (sometimes) great fielder. Advanced stats, especially, often embrace other theoretical issues, like which elements of performance are driven by random chance or are instead things a player can control.
But imagine a situation where the statistic appears to tell us what we want to know even though the idea behind the statistic is wrong. Ptolemy, the ancient Roman astronomer, did a pretty good job telling his contemporaries which path the planets would take along the night-sky, even though his basic hypothesis – that the earth was the center of the solar system – was completely wrong. So it is possible to get useful information from a bad hypothesis.
Fielding Independent Pitching, or FIP, is a stat that attempts to isolate the things a pitcher can control – strikeouts, walks, home-runs – and thus the idea behind FIP is that it tells you how good a pitcher is independent of the fielding skills of the players behind him.
Our FIP leaders so far this year are Matt Harvey, Anibal Sanchez and Clayton Kershaw. That’s a pretty good list.
It looks like FIP works, but the hypothesis behind FIP has a big problem: the three factors of FIP have a chaotic relationship with each other when looked at across the entire league.
Logically that shouldn’t be, right? If pitchers are starting to dominate the game – strikeouts going up, walks going down – then home-runs ought to be (generally) going down as well, allowing for some noise.
Yet, strikeouts are way up, higher than they have ever been before. Every year of the past five years has been the biggest year EVER for strikeouts. We live in the Strike-Out Era.
Walks are also WAY down, at their lowest frequency since 1969.
But, the frequency of home runs this year (.97 home runs per team per game) and last year (1.02 home runs/team game) put both seasons among the 20 biggest home-run seasons in the history of baseball. 2012 was comparable to 1998 (1.04 home runs/team game) in terms of the frequency of home runs.
So now pitchers, as a group, are having unprecedented success on the first two measures of FIP and on the third measure they are doing as badly as they ever have in the history of baseball. Logically, this should not be be the case if what connects these three measures – strikeouts, walks, home-runs – is that “pitchers control them.”
Another way to think about a hypothesis (any hypothesis) is that it is really a sort of prediction. It carries inside it the implication that when all the data comes in, I think we will see X. Well, more data has come in, and home-runs are not behaving as predicted. There may be a reasonable sounding explanation for this, but beware. Ptolemy came up with a lot of reasonable sounding explanations as well.