The Astros and Fangraphs' Clutch Hitting Stat

Sometimes I can write an article by expanding on a short comment.  My comment on David's article about the Astros' team splits pointed out that Astros have been an amazing clutch team, as measured by Fangraphs' Clutch hitting statistic. This strikes me as interesting, though I don't have the answers as to what it means for the future.

As a preliminary, let's look at the description of the Fangraphs' clutch statistic.  This clutch stat is different from most clutch measures because it is part of the Win Probability Added (WPA) family of statistics.  Both WPA and leverage index are used to determine the player's clutchness.  The actual measure is based on the player's performance in various leverage situations compared to his overall performance.

According to the Fangraphs clutch statistic, the Astros were No. 1 in 2010 and 2009 and No. 2 in 2008.  Over the period 2008 - 2010, the Astros are +17.77, far ahead of the next best team, the Angels, at +7.55.  Since I have known for some time that the Astros and Angels are among the most consistent teams at beating their pythagorean record in recent years, my comment wondered whether the clutch hitting (as measured by fangraphs) has any relationship to over or under performing pythag.

So, here are the top five clutch teams for 2008 - 2010 and the accompanying over/under Pythag results during that period:

(Clutch Stat/ Games Over-Under Pythagorean)

Astros   +17.77 / +23

Angels  +7.55 / +18

Marlins  +4.85 / +7

Giants   +4.08 / +4

Phillies  +3.51 / +2

The five best clutch teams also were good at beating their pythagorean W/L record during 2008 - 2010.  What about the five worst clutch teams for 2008 - 2010?

(Clutch Stat/ Games Over-Under Pythagorean)

Blue Jays  -10.14 / -15

Indians      -9.44 /  -11

Nationals  -9.12 /  -13

Cubs    -8.53 / +4

Rockies  -8.34 / +4

The pattern isn't as consistent for the five worst clutch teams in 2008-2010, because the Cubs and Rockies were poor clutch teams but still over-performed their pythagorean record.  However, the top three worst clutch hitting teams also strongly under performed their pythagorean wins and losses.  Taken together, a comparison of the top and bottom five clutch teams and their pythagorean performance suggest that the fangraphs' clutch statistic affects a team's over or underperformance of its pythag wins and losses.  It doesn't decisively prove anything, but it's interesting. 

However, I'm surprised that the results are this indicative.  Many factors are likely to affect the Actual vs. Pythagorean Result. (This comparison doesn't address team pitching at all.)  And because Pythagorean over or underperformance is purely based upon the distribution of runs allowed and runs scored among a season of games,  the causation between clutch hitting and pythag performance is likely to be indirect.

As stated in the photo caption accompanying this article, Michael Bourn was the best clutch hitter in 2010, based on the Fangraphs clutch stat.  The Astros' individual clutch hitting stats are here.  The top clutch hitters for the Astros in 2010: Bourn, Carlos Lee, Geoff Blum, Lance Berkman, Hunter Pence, Jason Michaels, Cory Sullivan, and Chris Johnson.

For the period 2008 - 2010, Carlos Lee was the top clutch hitter in baseball, with +4.21.   Michael Bourn, Miguel Tejada, and Lance Berkman are in the top 20 for 2008-2010.  Hunter Pence is at No. 22.

One of the reasons that I chose the period 2008 - 2010 is that it accompanies the period that Ed Wade has been in charge of Astros' player moves.  In the past, I have wondered if the team construction has affected the ability of teams like the Angels and Astros to overperform their pythagorean record.  Jeff Keppinger was the worst clutch hitter (using fangraphs) on the Astros in 2010--even worse than the pitchers--and his role on the team has been reduced, despite one of his best seasons. Ty Wigginton was the worst clutch hitter on the 2008 Astros, despite good overall stats, and he was non-tendered. Coincidence or not?

What would all of this mean for the Astros' future?  I don't know.

Let's start out with the fact that clutch hitting is one of the most controversial issues in sabermetrics.  The majority position seems to support the idea that clutch hitting is largely a matter of luck, i.e., it is not a repeatable skill.  Tango's more recent contest indicates that clutch skill can be identified, but that the effect is relatively small.  Others (including Bill James) argue that clutch hitting skill is too difficult to measure, and conclude that the evidence is not adequate to support a conclusion.  I tend to fall in the latter camp.  In addition to the statistical issues raised by James, if we cannot agree on what constitutes the best measure of clutch, we will always have difficulty reaching a unanimous verdict.  I will add that I have not seen any statistical studies which are based on the Fangraphs clutch statistic.

If the Astros' clutch hitting has been critical to beating the Pythag win-loss record...and if the clutch results are basically luck...then this could be a bad predictor for the Astros in 2011, since it might indicate a regression in win-loss record.  The Astros' consistency in beating the pythag over the years leads me to discount this conclusion, but that argument can be made.  Another possibility is that something about the Astros' team construction favors the team's ability to beat its Pythag.  If that's true, then maybe this is a positive factor for the future.

It's also worth noting that two of the Astros' better clutch performers over the last three years, Lance Berkman and Geoff Blum, will not be on next year's team.  Will that change the clutch character of the Astros?  Again, who knows?

Let's look at the new guys.  Bill Hall has a fairly consistent positive clutch result over his career.  Hall shows a negative clutch result in only one out of nine seasons.  His total clutch score over those seasons is +3.88.  For what it's worth, that compares favorably to Lance Berkman.  Clint Barmes, on the other hand, had a negative clutch stat for five of his seven seasons, and his cumulative clutch score is -2.79.  As for the younger additions to the team last year, it's too early to say much, even though in small samples Chris Johnson and Wallace fared well on the clutch stat, while Castro did not. Carlos Lee will be back, and he is one of the most consistent clutch performers active in baseball.  Michaels, Pence and Bourn, who also have excelled on the clutch statistic, will still be in the lineup. 

Any thoughts or reactions?