Astros Batters and RE 24

Some of you may have read Ben Lindbergh’s article about Josh Reddick entitled “King of Unclutch.” Lindbergh points out that Reddick has a history of “unclutch” OPS in high leverage situations (with former Astros hitter Richard HIdalgo second worst), and a string of consecutive “unclutch” seasons which is exceeded only by Sammy Sosa. The author acknowledges that Reddick is having a good overall season, but posits that the poor high leverage batting splits “lack a clear explanation” and call into question his true value. I don’t have a clear explanation for his high leverage splits, but we do know that the existence of clutch hitting as a “skill” is highly contentious and unproven. Undescoring this, Lindbergh notes that the Astros, an analytic oriented organization, did not even consider “clutch” in their decision to sign Reddick.

It’s possible that Reddick’s leverage splits don’t mean much at all. Adequate sample size has always been one of the problems with solving the clutch question. Reddick has almost 600 PA in high leverage situations, which seems like a lot. But splits data may require 1,000 - 2,000 PA to be meaningful in an unregressed form. Otherwise using Reddick’s split for a one season equivalent of high leverage PAs would require significant regression (i.e., estimating a result that is substantially closer to league average). Reddick’s known weakness against LHPs may also play into the high leverage results, since opposing managers may be more likely to bring in a tough lefty reliever to face him in those late game situations.

Another oddity is that Reddick’s poor high leverage results appear to be highly influenced by one inning--the 9th. Over his career, Reddick’s tOPS is 77 in the 9th inning—worse than any other inning by a considerable margin. In close games, the 9th inning is usually the highest leverage inning. Maybe this is just an anomaly. Or maybe Reddick doesn’t match up well against closers. According to Baseball-Reference, Reddick hits best against finesse pitchers and worst against power pitchers.

RE24

A more interesting point is that Reddick’s RE24 results show a different conclusion than clutch stats. I have written about RE24 previously (Link, Link, also, see Joe Posnanski’s article here). RE 24 (“Run Expectancy: 24 Base/Out” situations) tabulates how much a player’s plate appearance increased or decreased the expected runs for a given base/out situation (as an illustration, runner on 1st and 2d with 1 out is one of the 24 base/out states). RE24 excludes the leverage components related to the score and the inning. RE24 also recognizes some productive outcomes that are not included in normal linear weights stats such as wRC+, wOBA, and OPS. Productive outs (such as moving runners over, sac flies, avoiding double plays) are given credit in RE24. Although both clutch stats (e.g., clutch, “late and close,” RISP) and RE24 are context dependent, arguably RE24 is more skill-dependent than clutch statistics and thus more likely to be repeatable. If I was a GM and had the choice of leverage splits or RE24 for player valuation, I would pick RE24.

RE24 can be compared to a player’s runs above average (wRAA) to determine whether he produced more runs than expected for his given linear weights measure (e.g., wOBA or wRC+). wRAA is a linear weights calculation of expected runs for a given combination of walks, singles, doubles, triples, and HRs based on the average league outcomes for those events. If RE24 is higher than RAA, this (roughly) indicates that the batter’s wOBA was more productive than expected. And, this comparison is more favorable to Reddick than his leverage splits.

Over his career so far, Reddick is 55 runs above average based on RE24, which is 23 runs higher than his linear weights RAA. Perhaps Reddick’s approach in high leverage situations is more focused on situational hitting than hitting the ball hard—at least compared to his other plate appearances. That may explain part of the difference between his high leverage splits and RE24 results. It would be consistent with Lindbergh’s conclusion that Reddick’s plate discipline isn’t worse in high leverage situations, but that he doesn’t hit the ball as hard during the high leverage splits.

Astros and RE24

Now let’s make the transition to examining how RE24 measures the Astros as a team. The Astros’ offense collectively has a very high RE24 so far. The Astros current RE24 (145) is almost 50% higher than the RE24 (99) of the second best team, the Nationals. However, as impressive at it is, the Astros RE24 ranking isn’t surprising, given that the Astros rank first on most major offensive statistics. The Astros’ also enjoy a large margin over the second place Nationals in wRAA.

The tabulation below shows the top six Astros’ hitters in RE24. The first number is RE24 and the second number shows the difference between the player’s RE24 and the linear weights wRAA.

Astros Hitter:RE24/RE24 > wRAA

1. Correa 29.82 / 6.52
2. Gonzalez 29.09 / 9.79
3. Springer 28.52 /-0.88
4. Reddick 22.35 / 9.95
5. Altuve 20.55 / -5.05
6. Gurriel 12.81 / 8.81

As indicated in my previous 2016 article, Carlos Correa’s short career demonstrates a prolific ability to pile up RE24. As in his prior seasons, his current production outperforms his linear weights runs above average. If it seems like Marwin Gonzalez has been at the center of the team’s run production, his RE24 confirms that conclusion. Gonzalez, Reddick, and Gurriel are notable for the ability to produce runs at a higher rate than predicted by their standard slash line stats. Although Yuli Gurriel’s wRC+ is not exceptional for a first baseman, his RE24 significantly outperforms the linear weights wRAA—leading to an outsized contribution to run production. It isn’t surprising that Springer and Altuve are on the list, since they may be the best hitters on the team. But neither hitter’s RE24 outperforms his wRAA. Altuve’s RE24 frequently underperformed his wRAA in the past—most likely because of his susceptibility to grounding into double plays.

On the reverse side, Alex Bregman and Nori Aoki have not performed well on RE24 so far this season. Both players have notably negative results for run expectancy, and have RE24 results (-5.78 Bregman; -11.34 Aoki) which significantly underperform their wRAA (difference:-6.68 Bregman; -4.94 Aoki).

The Astros have had an exciting first half of the season, and the collection of RE24 results may hold the key to why the offense has been spectacular.