Mark Kotsay, xBABIP, and Confirmation Bias

At baseball, Mark Kotsay hasn't been any good for awhile.  Injuries and age have robbed him of value -- in his prime Kotsay was a very good starter, a solid bet for a 4 WAR Season.  His ability to play a solid outfield along with his skills with the bat were valuable commodities.

That Mark Kotsay hasn't been seen in years.    Most were non-plussed when he was acquired in a minor deal last summer.  A lot of fans are really pissed he's still the DH -- but is there any reason for hope?

Nassim Taleb wrote a pretty well received book about applications of stats in economics and philosophy called "The Black Swan".  Basically, the book is about why we should be skeptical about what we think we know, even in a data saturated world full of linear models.

Nassim illustrates several common logical fallacies, one of which is the confirmation bias:

Confirmation Bias – It is human nature to try to find evidence that confirms our stories, to take past instances that corroborate our theories and treat them as evidence. The problem is, it’s pretty easy to find evidence that will confirm just about any theory.

We can easily see such a bias for hating on Mark Kotsay:  Mark Kotsay has an OPS+ of 69.  He's a DH for crying out loud.  He has no defensive value.  He's like a million years old.  He just SUCKS.

And on the surface he does.  But that's just the Confirmation Bias talking.

Last year the Hard Ball times ran some numbers and ranked the various "batting average balls in play" predictors.   One method in particular, which they dubbed, "xBABIP", proved particularly effective.*  According to their data, Kotsay has had quite a bit of bad luck this season, so much so in fact that most fans' outlook on him would be reversed had he received only average providence.

Interestedly, aside from advanced stats and data modeling, there is a meme around town that rather than just sucking, Mark Kotsay has just been really unlucky.

Get this: according to xBABIP Kotsay should have a .297 BABIP.  Go ahead, download the spreadsheet, go to FanGraphs, and plug in Mark's stats, you'll get the same thing.**

Kotsay's actual BABIP this year is only .218, which is 79 points lower than the numbers expect.  Assuming an average luck year, Kotsay's OPS should be roughly 79 points higher, good enough to go from a slash line of 215/298/338 to 294/377/417.

Suddenly Kotsay is one of the most important hitters on the team, OBP heavy in a lineup filled with impatient sluggers.

The thing here isn't so much that Kotsay is deserving of a roster spot or that whoever is, but that knowledge is hard and complex endeavors such as running a baseball club are often subject to variations the people in charge don't see coming, let alone the pundits.

And here is something completely different and yet related.

via t0.gstatic.com

* One of the foundations of modern sabermetrics is the idea that batters experience a good amount of luck in any given year on the number of balls in play that fall between defenders.   Variance in BABIP between seasons is often used as an example of luck, good and bad.  Nick Swisher was run out of town for a year BABIP suggested was mostly bad luck.

**Look for a forthcoming post from me on differences in xBABIP and actual BABIP for the Sox' regulars.