Rays xBABIP Normalizations for 2010

About this time last year I took a look at the idea of xBABIP versus BABIP and how it affected the Rays in 2009.  I wanted to take this same concept and apply it to the 2010 team to see which guys were lucky or unlucky and how it altered their season.  This is work that was pioneered by Chris Dutton and Peter Bendix (Bendix has worked for the Rays as an assistant to their research and development program that features some of the brightest minds in the game) and I will be using a tool that Dutton helped bring to us mere mortals.

If you are familiar with BABIP then all you need to know is that this tool seeks to find a rate that a batter should be expected to reach base upon putting a ball in play.  I think it makes sense that a ground-ball hitting speedster should be expected to garner more hits than a lumbering fly-ball hitter.  If you’ve watched a lot of baseball, you’ve probably realized that when someone hits a ground-ball it has a higher likelihood of turning into a base-hit than a fly-ball.  Of course there is a power trade-off being made, but that’s a topic for another day.  This isn’t just theoretical mumbo-jumbo as we can see here that the BABIP by hit trajectory looks something like this:

There is nothing small about this sample as it comes from over 62,000 plate appearances in 2010.  Concepts like this form the foundation for Dutton’s tool which will be used throughout this piece.

First off, we can plug and play within the tool to do a quick comparison between the actual BABIP and the xBABIP spit out:

#H uses the difference between the two rates times the number of at bats to see how many hits a batter was either lucky (positive numbers) or unlucky (negative numbers) to receive.  For this analysis we will assume that each of these base-hits turned out to be or would have been a single.  For example, we can say that according to this Evan Longoria lucked into 11 extra singles over the course of the year, while the opposite extreme has Carlos Pena being unlucky enough (more likely the shift is a critical nemesis) to be robbed of 48 singles.  That is an incredible amount of base-hits taken away.  On a final note for this chart those team totals are weighted-averages for each guy listed so you can see that as a team, the Rays under-performed their xBABIP by roughly .026 points, which works out to almost 140 singles taken off the board.  Wonder why our batting average seemed so low this year?  Perhaps here’s a reason.

The next step is to plug in the number of singles that each player would have had if they performed at the level of their xBABIP.  We can then plug these numbers into the slash lines (batting average/on base percentage/slugging percentage) and weighted On Base Average (wOBA).  That looks something like this:

The numbers to the left of the divider were the actual results as calculated by me.  The numbers to the right of the divider take into account if the player had performed at his xBABIP and how all those extra singles would have shaped their slash lines, wOBA, and WAR (batting runs or wRAA is the only thing that I have changed from actual to expected).  Obviously, the guys with the biggest difference in actual vs. expected BABIP are the big swingers here.  The perfect example is Carlos Pena who actually had a 0.196/0.325/0.407 with a 0.334 wOBA.  The effect of those extra 48 singles added in takes his line all the way up to 0.295/0.407/0.506 for a wOBA of .409.  That line is reminiscent of his great 2007 line of 0.282/0.411/0.627 for a 0.430 wOBA.  Less power, but pretty close on the batting average and on-base percentage portions.

Pena is a poor example, in my eyes, due to the extreme shift that he faces.  In 2007 most of those scorching grounders to the right side were getting through, but by 2010 teams were on to his game.  An example of a player with a less obvious reason for their lack of “luck” would be Ben Zobrist who had 27 singles taken off his expected line.  How much happier would you have been if BenZo had put up a 0.287 batting average instead of 0.238 or an on-base percentage of .387 instead of a walk-fueled .346.  Even his SLG% would have seen an increase from 0.353 to 0.402 with the conservative guess that all 27 of those hits would only have been a single.  Seeing a wOBA of 0.367 looks a whole lot better than his actual, barely above average wOBA of 0.331.  That difference represents approximately 20 extra runs that he would have contributed with the bat and he would have seen his WAR go from a respectable 3.2 all the way up to an MVP-conversation 5.3.

Most guys didn’t see such drastic swings, but we can see that Longo was a bit lucky and that’s about it across the team.  Crawford and Brignac got a couple of extra bases, but nobody over-performed anywhere near as much as Zobrist, Pena, Upton, and even Navarro, Aybar, and Dan Johnson under-performed.  We could see these guys have some nice bounce-backs in 2011, though Pena, Navi, and Boots will have to try to improve their luck somewhere else.  Interestingly, if you clicked the link to last year’s article, you’ll see that Navarro, Pena, Upton, and Aybar were again the most unlucky players, in order.  This sort of carry-over flies in the face that BABIP should normalize over time, but perhaps that has more to do with the high strikeout rates for Pena and Upton, particularly.  You’d think with less balls put in play that luck would have a stronger skew to the data.

Of course one area where this got it right was predicting that Jason Bartlett would come crashing back to Earth (21 luck hits in 2009 vs. 9 unlucky hits taken off the board in 2010), but you didn’t need xBABIP to tell you that Jason Bartlett played over his head in 2009.  I really like this tool, and we’ll get to see it in action this year as several of these guys could have benefitted from a bit more luck.  Namely, check out Dan Johnson (actual wOBA of 0.344 vs. expected wOBA of 0.424!) or Matt Joyce (0.362 vs. 0.380) or even Johnny Jaso (0.349 vs. 0.364).  If the Rays can get a bit better luck on balls in play out of Upton, Zobrist, and the young guns, perhaps the Rays offense isn’t quite as decimated as those out of the know seem to think.


About Jason Hanselman

Rays fan.
This entry was posted in recap, statistics and tagged , , , , , . Bookmark the permalink.

2 Responses to Rays xBABIP Normalizations for 2010

  1. Someone left this comment on my Facebook page about this

    “In general I’d suggest looking at a wider range of BABIP to xBABIP results to establish a better idea of the shape of the data – what is a standard deviation on turf and off turf matter for BABIP. Probably also good to check the line drive bias in the data set.

    It’s interesting but 140 singles, essentially a hit a game, doesn’t seem like it’s luck.”


  2. Jason Hanselman says:

    It isn’t my study or analysis, I’m just using the tool. I doubt it’s flawless, especially in smaller samples. I looked only at Rays that had 100+ PAs in 2010 and it seems to yield some quality information. I’m sorry this person doesn’t appreciate the hard work that went into this.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s