Testing Conventional Wisdom: Derek Jeter and Alex Rodriguez in the Playoffs

Derek Jeter - the Captain, the True Yankee^TM  He took his first 12 teams in New York to the playoffs, winning the world series 4 times.  A .309 playoff batting average and a .302 World Series average among other memorable playoff exploits earned him the name Mr. November. 

On the other hand, we have Alex Rodriguez.  A-Roid.  A-Fraud.  Reports say he's one of the worst teammates around, and steroid rumors will linger throughout his career.  To top it all off, his playoff performance has earned him the nickname Mr. May.

Playoff performance certainly tends to suffer from the dreaded uncertainties of small sample sizes.  As such, it would be prudent to study the playoff production of these two polarizing figures in recent New York Yankees history.  The first question that comes up, as an objective observer, is that of how to analyze the production.  In order to both preserve some rigor while maintaining some ease in the method, I decided to use playoff OBP as the measuring stick.  Much of my method is adapted from the appendix of The Book, and is based mostly on the concept of regressing towards the mean to find true OBP skill. 

For the Captain, we have 25 different postseason series from 1996-2007.  Over these 11 years, Jeter posted an OBP of .390 in 17145 PAs, and so we can assume his OBP skill to be .390 due to the large sample size.  However, Jeter's postseason OBP in this span is a mere .374 in a significant sample of 546 plate appearances (a number that is remarkable in itself).  This is because we haven't adjusted for the quality of pitching of playoff teams.  Over the 1996-2007 period under observation here, AL playoff teams allowed an opposing OBP of .307, a very impressive number in the period including the high offensive period of the late 90s/early 2000s.  Now, with this we can calculate Jeter's expected OBP: Expected = (Offense + Defense - Average).  With the league OBP roughly equal to .340 over this period, we expect Jeter's OBP to be .356.  Jeter's .374 OBP beats this expectation by 18 points.  Over 576 plate appearances, this is just under 1 standard deviation from the mean.   Among players with an expected OBP of .356, only 20% of players would be expected to perform better.  So it does appear that yes, Derek Jeter may possess some sort of tangible ability to perform above his true talent level in October.

Can we say the same about A-Rod during his time in pinstripes?  Unfortunately, we only have 108 playoff plate appearances, which leads to much more variance, but we can still examine what we have.  In the harder offensive times of 2004-2007, the average OBP dropped to .336, and we also see the OBP allowed by playoff teams drop to .304.  Rodriguez put up a mammoth OBP of .404 in that 4 year span.  Regression over his 3532 plate appearances leaves us with an estimate of .395 for Rodriguez's OBP skill.  Calculating his expected OBP in the playoffs as above results in an expected OBP of .363.  It appears conventional wisdom may be on the right track here, as A-Rod's OBP over his 108 playoff PAs is only .343.  However, since the sample size is so much smaller, this is only .45 standard deviations from the mean.  This result is much less significant than that with Jeter.  Still, we would expect 63% of players with a .363 OBP estimate to perform better than A-Rod's .343.  Perhaps the established reputation is correct in saying that Rodriguez performs worse in the playoff months.

It's important to stress here in this analysis that we are comparing Jeter's "clutchiness" to players with similar skills to Jeter, and we are comparing A-Rod's "clutchiness" to players with similar skills to A-Rod.  Not only does this mean that Rodriguez was still above average in the playoffs (whereas Jeter was, as advertised, great), it means that managers, fans, and the media shouldn't be clamoring to be rid of A-Rod or asking for a replacement with a lower skill level.  Finally, we are still subject to small sample sizes.  Certainly, the 3500 plate appearances from the regular season tell us more about what to expect in the postseason than the 100 postseason appearances, and similarly, 17000 PAs tell us more than 500.  This analysis shows how these players have performed in the past.  If somebody held a gun to my head and forced me to predict a stat line for their next playoff series (whenever that may be), I would certainly err on the side of the larger sample size.