When Joba Chamberlain came out of the bullpen on Tuesday night Larry asked me if I thought we’d ever see Joba in the rotation again. We both felt that the answer was probably not, regardless of the fact that we both endorse the idea of Joba getting another shot as a starter. The bottom line is, the Yankees aren’t treating Joba like a reliever who may eventually be used as a starter. The team has seldom allowed Joba to pitch more than one inning at a time. Converting him back to a starter in Spring Training may not only further damage his confidence, but possibly his arm as well.
That question has since motivated me to return to a topic that was heavily debated before the season began. For those who remember, there was some uncertainty as to whether or not the 5th spot in the Yankee rotation would fall to Joba Chamberlain
or Phil Hughes
. At the time, I voted for Joba. He was better as a reliever in 2007 than Hughes was in 2009, and until this season he had also been better as a starter than Hughes had been. But Joba wasn’t bad last season; he just wasn’t great. Isn’t that exactly the same thing we’d say about Hughes’ 2010 season: not bad, but not great?
The data in the table below are taken from Baseball Reference and Fangraphs. The data are what you’d expect, except I’ve added a rough estimate of how many innings each pitcher gave the Yankees per start. This will overestimate the number slightly because it ignores one-inning relief appearances, like the one Hughes gave recently, but it still works as a serviceable estimate of each pitcher’s efficiency.
The numbers show that in 2009 Joba was a somewhat inefficient, but largely average starter. (For those who don’t know, ERA+ is adjusted so that a score of 100 is league average. This means that a 100 in 2009 may be different from a 100 in 2010, the “Year of the Pitcher”.) He didn’t give the Yankees much in the way of length in his starts, which is probably related to his high WHIP, but he was still better, younger and cheaper than either A.J. Burnett
or Javier Vazquez
has been this year.
has been a better pitcher in 2010 than Joba was in 2009, but the difference isn’t the slam dunk that it has become in the media narrative. ERA+ is designed so that each point represents a 1/2 percentage of improvement. Hughes in 2010 has been about 1.5% better at keeping runs off the board than Joba was in 2009. If Hughes has won a role in the team’s rotation for the foreseeable future why, then, didn’t Joba’s 2009 at least keep him in the conversation?
The answer may lie in WAR. According to Baseball Reference Phil Hughes
has a bWAR of 2.4 this season (which is just a hair above the 2.1 he accumulated in 2009). Meanwhile, Joba was replacement level in 2009 according to his bWAR, registering a 0.1, which is bad.
Hughes’ WAR is about average, which aligns nicely to his ERA+ this year. Joba’s 2009 WAR is replacement level, which should translate to an ERA+ of 80-85, yet his actual ERA+ was 98. An analysis meant to point out that Joba was at least good enough in 2009 to warrant mentioning for the rotation again in 2011 has also uncovered one of the issues with WAR and other advanced stats. WAR penalizes pitchers with poor peripherals, regardless of how those peripherals translate to actual runs allowed.
With a K/BB ratio in 2009 of 1.75 Joba had poor peripherals (Hughes’ K/BB ratio in 2010 is 2.59, just about the same as CC Sabathia). While I don’t believe the Yankees were looking at WAR when they made their decision about which pitchers would start and which would not, they may have been examining the raw data that influence WAR. Those data confirm that Joba in 2009 was far from great, and that the Yankees probably made the right decision in selecting Hughes over Chamberlain this season.
However, given that Joba was far from a complete and utter failure as a starter last season, one might think that 160 innings of 98 ERA+ ball pitching primarily in the AL East would merit further consideration as a starter, but that’s an argument for another time.