Second Thoughts Game #109: Texas 2, Cleveland 12
Nights like Friday can remind of how frustrating baseball can be – the component parts fail to line up in the neat fashion expected, and as a result, a team good at scoring runs can seem inconsistent. It’s a frequent occurrence in baseball for all teams, and it can boil into frustration for a sub-.500 team but for the fact that it was a stretch of six games in April 2013, one that saw Cleveland score 53 runs in 6 games and win those games by a combined 39 runs, that kickstarted a torrid 18-4 run that re-started the season.
It’s not an ideal distribution of runs, winning by ten. It’s not a consistent distribution, but performances like Friday’s serve as a reminder that there’s enough pop in Cleveland’s bats to fuel a playoff offense. In spite of all critiques of inconsistency, they remain fifth in the AL in runs scored.
Defense – different matter.
The Salazar Positive-Regression Train Keeps Rolling
Entering 2014, the Indians’ rotation was described as ‘volatile’ by Mike Petriello of FanGraphs. Volatility did not necessarily, at the time, imply that the rotation would be a net demerit for the team; all Petriello argued was that Cleveland’s rotation had a tremendously high ceiling as well as an incredibly low floor. The one pitcher who exemplified best this volatility, at the time, was Danny Salazar, who had the ceiling, as Petriello noted and which this author (should not have) predicted, of a Cy Young-caliber pitcher. However, other analyses noted Salazar’s floor; he remained a 5’11” pitcher who did not fit the ideal mold for durability, his slider was not a plus offering, and his extreme foul rate made him a less-than-ideal fit for a distance pitcher (despite being a net positive for a pitcher).
Before the All-Star Break, Salazar’s season was an exploration of that same direful floor. The defense – regarded as worst in the league by such stats as Ultimate Zone Rating and Defensive Runs Saved – had given Salazar no help, as he had run a trivia-question-esque .369 batting average on balls in play (BABIP), whereas the league average is .300 and even a pitcher BABIP as high as .330 is regarded as anomalous. Much of Salazar’s 5.53 first-half ERA falls on the shoulders of the defense; this much is clear.
On the other hand, the erosion of his strikeout, walk, and home run rates – the three facets of Fielding-Independent Pitching (FIP) – contributed to the decline of his effectiveness and necessarily have nothing to do with his defense. At 1.77 home runs per nine innings pitched, Salazar’s home run rate was 192% the American League first-half average of .92 HR/9. His walk rate was also far above-average, but whereas his first-half strikeout rate (25.3%, 20.0% AL first-half average) was more than enough to compensate for his deficient walk rate (9.2%, 7.9% AL first-half average), his home run problem fueled what FIP determined would have been an average-defense, average-luck ERA of 4.70 – better than 5.53, still far worse than his 2013 FIP of 3.16.
The home run problem, however, was curious in the sense that it was so anomalously high; in the case of the overwhelming majority of pitchers, the ratio of home runs to fly balls typically hovers around 10% - a 13% HR/FB% is viewed as unsustainably high. Salazar’s first-half HR/FB% ran up to 14.8%, one that prompted Cleveland to send Salazar to AAA.
Because HR/FB% is a highly unstable statistic, it is a statistic wherein the mean – 10% - is likely a better predictor of future HR/FB% than is past HR/FB%. Salazar’s HR/FB%, hence, was likely to positively regress to the mean of 10%. Salazar’s Expected Fielding Independent Pitching, xFIP – which assumes a league average HR/FB% - puts Salazar’s expected first-half ERA to 3.82; in other words, given his combination of strikeouts, walks, and fly balls, Salazar would be expected to have a 3.82 ERA. Above that, approximately .90 of his ERA is as a result of unsustainably high HR/FB% rates, and .80 as a result of defense or BABIP luck, adding up to his 5.53 ERA. Given positive regression, one would expect that his ERA would come down the rest of the way by approximately one full point.
‘Positive regression’ is not synonymous with an upturn in luck; it means either that Salazar is making mistake pitches at a lower, more average rate or that those pitches are getting punished at a lower, more average rate. Whatever the cause, positive regression was likely.
Salazar’s post-All-Star Break return has been a tale in violently positive regression, improved performance, good luck, and good defense. Since his return, Salazar’s HR/FB% sits at a robust 0.00%, with him having allowed not even a one home run. Whereas previous Second Thoughts have dissected how Tomlin has failed to induce pop-ups at a rate requisite to counterbalanced the increased home run rates component of flyballers, Salazar’s infield fly ball rate since his return has been unsustainably high – at 31.5% of all fly balls after the all-star break, a rate good for 8th among pitchers with 10+ IP in that time period, Salazar’s IFFB% is 8th-highest and has proven to be a real, luck-independent BABIP-suppressor, given that popups turn into outs at approximately the same rate as strikeouts. Popup-induction is a sustainable skill, in the same vein as strikeouts are a sustainable skill – above-average rates tend to remain above-average, though nobody expects any player to average fifteen strikeouts per game; likewise, Salazar is unlikely to run a 30% IFFB% the rest of the season.
Positive regression, in its most cliché form, has come in the form of Friday’s outing, wherein Salazar allowed four free passes (3 BB, 1 HBP) and only struck out 4 batters in 6 innings. To his merit, Salazar induced two infield fly balls, which are like strikeouts in that they are nearly 100% outs, but this merely balanced matters. Salazar’s performance on Friday was exceedingly middling day from a fielding-independent perspective, that same perspective that documents those pitcher actions that lead to run prevention. In spite of this, Salazar threw six innings with only one unearned run scoring in the sixth.
Positive regression, typically, isn’t a dispensary of cosmic justice; if a pitcher throws a first half with a .400 BABIP against and a 20% HR/FB%, ‘regression’ does not in turn imply that his second half will bring a .200 BABIP and 0% HR/FB% to balance out the anomalous first halves. Likewise, if a pitcher’s ERA is a full two points higher than his xFIP, it’s typically not the case that ERA then overperforms xFIP the remainder of the way to even the scales of base-and-ball justice. Fortunately for the Indians, they seem to have not gotten the news: in contrast to Salazar’s xFIP being nearly two full points lower than his ERA for the first half of the season, Salazar’s single-game ERA was 0.00, his single-game FIP was 3.80, and his single-game xFIP was 5.06.
In a season as extreme as Salazar’s, in which he has endured unsustainably high home run rates, occasionally brisk walk rates, and excellent strikeout rates, regression to the mean is inevitable. It’s wholly unlikely that a 30% IFFB is in the least sustainable, but between the cosmological expectations on Salazar and his nadir of May 2014, it's likely that nothing, good or bad, should be surprising.
John can be reached on Twitter at @JHGrimm. He can also be reached by e-mail at firstname.lastname@example.org.