Hot Tip: Tim Lincecum: Bad Luck or Bad Pitches?

Dave Cameron wandered little ways off the sabermetric reservation on Monday, 6/18/2012, over on Fangraphs with an article looking at Tim Lincecum's struggles from a statistical standpoint.  Giants pitchers in general, and Matt Cain in particular have long been a source of interest and exasperation to the sabers due to an apparent ability to sustain better ERA results than their normalized stats, FIP and xFIP would indicate is possible.  Now Timmy is presenting the opposite statistical phenomenon and it's starting to cause a stir.

As we all know, the season is rapidly closing in on the half way point and Timmy's ERA is still a very ugly 6+.  At the same time, his FIP and xFIP are a much more attractive 3.87 and 3.82.  Since these two derived statistics are based on normalizing parameters that are generally believed to be subject much more to random statistical variatition, luck if you will, the obvious conclusion is that Timmy has simply had a prolonged stretch of bad luck that will turn around any day now.  He's become the classic fantasy Buy Low candidate.

The only problem with that analysis, as Dave Cameron points out, is that it flies in the face of what we have all observed from watching Timmy turn in one rough start after another.  He hasn't pitched well!  We've all seen the inability to find the strike zone, the walked batters, the capitulation with meatball pitches down the middle that get hammered.  Dave Cameron has apparently seen it too and wavers from the sabermetric party line by saying maybe what we have been calling "luck" is more a statistical variation in execution rather than random events that occur after the ball is struck by the bat.

Predictably, the sabermetric faithful struck back later in the comments insisting that outliers like this are completely consistent with Defense Independent Pitching Statistics(DIPS) theory, which basically says balls in play are out of the control of the pitcher.  From there it kind of degenerates into a bad luck/bad pitches, yes it is/no it isn't, tastes great/less filling argument.

In fact, there is a very good explanation for the statistical phenomena that is Timmy right now that accepts his bad pitching as a fact, yet stays true to the theory behind FIP/xFIP.  FIP and xFIP rely on normalizing certain statistics like BABIP, LOB% and HR/FB to league averages on the theory that if you get a large enough sample size, those parameters will regress to league average and you will then have a pitcher's true talent performance.

What we have all observed with Timmy is that he tends to do fine as long as the bases are empty and he can work out of the windup.  As soon as he allows a runner to reach base, especially if there are less than 2 outs in the inning, his mechanics while pitching from the stretch fall apart and he gives up hits and runs in bunches.  Since Timmy's struggles are not evenly distributed throughout the game, and are clustered around when he pitches from the stretch, his LOB% goes WAY down.  The hits and walks he gives up start a cascade that makes them come in bunches increasing the probability that any give runner will end up scoring.  It only takes that happening 3 times a game for an ERA to go from 3.00 to 6.00.

Will Timmy eventually regress to his career averages or at least league averages in these "random" statistical parameters?  Probably, but not until he straightens out whatever is going wrong mechanically when he has to pitch out of the stretch.