What does run expectancy tell us about the Astros’ offense?

I have been writing about sabermetrics for a long time. And one of my favorite stats is RE24, which can provide insight beyond the summary statistics like WAR and wRC+. RE24 is part of the win probability family of baseball stats; “RE” stands for run expectancy and “24” refers to the 24 base-out states (such as “bases empty, no outs,” “bases 1, 2, 1 out,” bases 1,2,3, 2 out, etc.).

You probably have seen run expectancy tables, which show the league average probability of scoring runs from each of the 24 base-out states. For each hitter, RE24 examines the base-out state before and after each plate appearance in order to arrive at the runs above or below league average for all of the base-out states that the batter faced. An article I wrote in 2013 provides a good explanation for using RE24.

Baseball traditionalists complain that advanced stats omit important acts like productive outs. RE24 recognizes productive outs because everything that improved the base-out situation (including productive outs, moving runners over, etc.) are taken into account. RE24 will credit a hitter for a sac fly, and debit a hitter for a GIDP. In theory, teams with good base running skills will have higher RE24 results because they have an above-average ability to advance runners.

RE24 has some similarities to the Clutch and the Win Probability Added stats because it is situational, but it varies from those statistics because it does not weigh situations for inning and score. Unlike Clutch and WPA, RE24 is based more on repeatable skills.

In the past, I have presented RE24 stats for individual players, but this article will examine RE24 on a team basis.

RE24 By Team

Fangraphs RE24 stat for batting by team (as of 8/9/21) is shown here. For comparison, Fangraphs runs above average (RAA) batting stat, which is similar to other linear weights summaries like wRC+, is shown here.

An initial observation from this comparison: the Astros are No. 2 in RAA, slightly (less than 2 runs) behind the Blue Jays, but are much better than any team on the RE24 metric. The Blue Jays are almost 50 runs behind the Astros based on run expectancy. In fact, the Blue Jays’ RE24 is 5 runs worse than its RAA. The Top 10 RE24 teams are shown below.

RE24 Offense Runs Above Average

1. Astros 119.62
2. Dodgers 109.5
3. Rays 77.48
4. Giants 77.06
5. Blue Jays 71.6
6. Padres 69.1
7. White Sox 65.85
8. Reds 61.92
9. Atlanta 61.34
10. Philly 39.7

The Red Sox, Nationals, and Twins are among the top offensive teams (RAA) which fall out of the Top 10 if RE24 is the metric. The New York Yankees are interesting because they are mid-ranked (15) according to RAA, but are closer to the bottom (24) in RE24 (-31 runs). That seems to confirm popular belief regarding the weaknesses in the Yankees lineup construction. Only time will tell if they remedied those weaknesses at the trade deadline.

In order to examine variables that may account for teams’ RE24 performance, I examined the correlation between other offensive stats and team RE24. I used the R-square statistic, which shows percentage explanatory power. Some statistics have little explanatory power: namely speed, GDP, K%, and base running. Isolated power (ISO) and on-base percentage (OBP) seem to have a lot of explanatory power (R-square of 57% and 68%, respectively). BB% has some explanatory power (26%), but K% is minimal (5%). BABIP also has a noticeable positive R-square (22%).

My conclusion is that the Astros, when at full strength, have a lineup construction that maximizes the ability to improve run expectancy, which in turn makes the offense better than indicated by the summary stats like RAA and wRC+. The Astros and Dodgers exhibit RE24 significantly above those teams’ RAA. The Astros’ RE24 is 43 runs better than RAA and the Dodgers’ RE24 is 57 runs better than RAA. For these two teams, is the whole offense better than the sum of the parts? Maybe.

RE24 is also applicable to pitchers, and is quite useful in evaluating relief pitchers. I may look at pitching RE24 in future articles.