Pirates' Sean Burnett For Nats' Joel Hanrahan: A Good "Sabermetric" Trade

Why would the Pirates trade reliever Sean Burnett with a 3.15 ERA to the Nationals for Joel Hanrahan with a 7.79 ERA?

Put another way, would you trade a pitcher with a 4.78 sabermetric ERA for one with a 3.44 sabermetric ERA?

Believe it or not, Burnett has the higher (worse) numbers in the latter regard. (All statistics are as of this morning, and are courtesy of FanGraphs.)

Hanrahan ought to be "lights out" as a reliever. He strikes out 10 batters every nine innings, while giving up only four walks.

He gives up a home run about every 11 innings. His strikeout rate is outstanding, and neither of the other two statistics are particularly bad.

Burnett strikes out six batters every nine innings, gives up the same four walks, and gives up a home run every eight innings on average.

Intuitively, Hanrahan is the better pitcher. Yet Burnett has given up 23 hits and 12 earned runs in 34 1/3 innings, while Hanrahan has given up 30 hits and 30 earned runs in a very similar 34 2/3 innings. What gives?

What we call sabermetric ERA is otherwise known as field independent pitching (FIP) statistics.

These are based on the three factors mentioned above, over which pitchers are believed to have complete control: home runs, strikeouts, and walks.

In these three cases, the result is determined solely by the pitcher (and batter), presumably without any input from fielders.

All other contacts of bat to ball result in a ball in play. Historical statistics suggest that over time, the batting average on balls in play—or BABIP—is just over .300.

That is, a given fraction of balls put in play will fall outside the control of fielders and be a hit, while the remainder will be controlled (caught or thrown to first) by fielders for an out.

This might vary from pitcher to pitcher, or hitter to hitter, or more likely from team to team, depending on fielding abilities, but not much.

Hanrahan's BABIP of .460 is about 50 percent too high, probably because of small sample size. Sean Burnett's is .205, and should be about 50 percent higher. Statistics say that Burnett has been lucky, and Hanrahan unlucky over a handful of late innings in half a season.

That's because these statistics are valid only over long periods of time: three years for a starter, twice as long (or more) for a reliever.

If you believe that the BABIP's will equalize over time, then plugging the above numbers into a standard formula confirms that Burnett is a worse pitcher than Hanrahan.

Could it be that the BABIPs are right because Hanrahan is just a soft tosser, and Burnett is an unusually hard thrower? It's possible, of course, but remember that we are talking about professional pitchers, so the exceptions to this rule usually involve injured players.

In 2007, Zach Duke, who was injured for part of the year, had a BABIP of .360, which is why his ERA of 5.41 was decidedly worse than his FIP of 4.95.

Chien-Ming Wang of the New York Yankees has a BABIP of around .400 in 2009 (and his ERA of 9.64 is almost double his FIP), but he has been on and off the disabled list for a year.

In the 2004 off-season, the Los Angeles Dodgers' General Manager, Paul DePodesta was derided for signing the Red Sox' Derek Lowe. That's because his ERA was 5.42, but his FIP was 4.26.

In the following two years, his ERAs were 3.61 and 3.63, meaning that the 2004 FIP was a better predictor of future ERA, vindicating DePodesta (who was, however, fired for other reasons, not least of which was "bad luck.")

Unlike some of the others, Burnett for Hanrahan looks like a good trade. The Pirates' new managers are big on "scientific" management. In this regard, they are very clever, perhaps to a fault.

Our quarrel with them is that they seem to overlook the "soft" stuff. Like player citizenship. Not to mention fan and player morale. All of which win or lose games. That is called "being too clever by half."