What’s Behind the Texas Rangers’ Pre All-Star Break Collapse?

Since 2010, the Texas Rangers have never been far from the playoffs – until now. The Rangers made the World Series in 2010 and 2011, reached the one-game Wild Card round in 2012, and then lost a one-game tie-breaker playoff in 2013 to get into the Wild Card round. Barring a miracle comeback this season, however, the Rangers’ traveling secretary will not have to plan anything beyond the team’s September 28 regular-season finale against the Oakland A’s.

At the All-Star Break, Texas owned the worst record in Major League Baseball (38-57, .400). The Houston Astros, who have lost 100 or more games for the past three seasons, recently passed the Rangers in the American League West standings. As recently as June 16, Texas was at .500 (35-35). Between then and the break, however, the ledger was 3 up and 22 down.

Injuries would seem to be a ready explanation for the Rangers’ problems. Such familiar names as Prince Fielder, Matt Harrison, Derek Holland, Mitch Moreland, Alexi Ogando, Jurickson Profar, and Geovany Soto either have been, or currently are, on the DL this season. A look at the Rangers’ game-by-game performance in scoring and allowing runs, however, suggests that the injuries may not be as powerful a reason for the team’s slide as some might have expected.

In the first graph, I’ve created a plot that, I claim, is not as confusing as it looks. Just imagine 95 vertical lines arranged left to right, where each line represents a Texas game in chronological order. Along any vertical line will be a blue point, representing the number of Rangers’ runs, and a red point, representing the opponent’s number of runs. As an example, Texas lost its season-opener to Philadelphia, 14 to 10. Rangers’ wins are highlighted in yellow.

The graph above is pretty messy, however, with a lot of day-to-day fluctuation. I have therefore re-graphed the same data, smoothing out the curves for Rangers’ and opponents’ runs. Using a technique known as Loess regression, the curves become less sensitive to game-to-game changes and instead follow trends over sets of several consecutive games. Smoothing the curves appears to follow the spirit of Nate Silver’s advice, in his book The Signal and the Noise, to avoid over-fitting of data (pp. 163-168). The smoothed-out curves are shown below.

Let’s look first at the Rangers’ run production (smoothed blue curve). Moving from left to right (corresponding to games 1 through 95), Texas has generally been producing 4 runs per game, an average of 4.12 to be specific. The smoothed blue curve sometimes goes a little above 4 and sometimes a little below 4, but usually does not deviate much from 4 (at one point, the Rangers’ smoothed run production approaches 5 runs per game, apparently corresponding to three games from May 24-26, in which they scored 12, 12, and 7 runs). Texas’s run production has thus far appeared to be robust to injuries.

Looking at the injured players’ offensive Wins Above Replacement (oWAR), a metric of production with the bat , one can see why their impact may not be great. Before being placed on the disabled list, Fielder’s oWAR this season was -0.1, essentially average, whereas his previous yearly oWAR values mostly were in the range of 3 to 6. During their respective careers, Moreland has tended to be slightly above zero and Profar, right around zero, on oWAR. Soto produced 1.3 offensive Wins Above Replacement in 2013, his last season with a substantial number of games played, but he has attained oWAR values as high as 3.5 and 3.6 in previous seasons. On the whole, losing the services of these players does not appear to have affected team run production much.

Rather, pitching appears to be the bigger culprit when accounting for the Rangers’ collapse. Texas has given up an average of 5.25 runs per game this season, but lately, the smoothed red curve has risen dramatically, passing 6, and then 7, and now climbing north of 8. (The Rangers gave up 8 or more runs in 6 of their last 8 games before the All-Star Break.)

Had injuries depleted the pitching staff over roughly the last 20 games, then perhaps the Rangers’ slide could be blamed on medical maladies. However, this does not appear to be what happened. As shown in the smoothed curved, some of the best Texas pitching occurred during the latter half of May (roughly games 42-56), in which Rangers’ pitching gave up 2 or fewer runs in 6 out of 15 games. Of the pitchers who contributed most to the team’s mound success in May (Joakim Soria, with a 1.13 ERA for the month in 8 total innings pitched; Neal Cotts, 1.54 ERA in 11 2/3 IP; Yu Darvish, 2.10 ERA in 30 IP; and Shawn Tolleson, 2.45 ERA in 14 2/3 IP), all pitched extensively in June and July. Just not as well. (Soria had a 0.00 pre-break July ERA, but based on only 2 2/3 innings.) Of the Rangers’ two sidelined World Series veterans, Harrison has made four appearances this season, none later than May 13 (with a 4.15 ERA in 17 1/3 IP), and Holland has not appeared this season.

Those who have pitched well for Texas this season (especially in May) thus have not been directly incapacitated by injuries. However, other hurlers’ ailments have undoubtedly diminished the Rangers’ pitching depth, perhaps burning out the active Texas pitchers to some extent.