The Plexiglas Principle states that a team that unexpectedly breaks out one year should be expected to decline the following year; rather than continue improving, one should expect that team to decline from their previous year’s success.
Regression to the mean is the basis of this principle. Not only should one expect luck-based statistics like BABIP to return to the mean, one should also expect skill-based statistics to regress toward the mean. Expecting Raburn’s or (pre-injury) Kluber’s exceptional skill-based statistics to hold steady or improve is an unreasonable expectation, but that does not mean that the odds-on likeliest possibility is an almost total return to the mean, but merely a regression toward the mean: excellent performances, ones that exceeded projections, are much more likely to be ‘good’ rather than ‘excellent,’ but they’re also unlikely to return to being merely average.
But who among the Indians vastly exceeded their projections? Masterson’s above-average season was not the best season of his career; Santana had top-prospect billing had enjoyed a comparably above-average season previous to 2013; Brantley’s season was below-average relative to his career, and Swisher and Bourn posted their worst seasons since 2008. The following seven are the only players who broke out relative to their careers.
- Ryan Raburn
- Yan Gomes
- Jason Kipnis
- Scott Kazmir
- Ubaldo Jimenez
- Corey Kluber
- Danny Salazar
There are two subdivisions of these breakouts: not-really breakouts, skill-based breakouts, and luck-based breakouts.
Kipnis is undoubtedly in the not-really-breakout category. His performance was worth 4.5 WAR; however, his minor league career was defined by offensive excellence, and his previous season – one victimized by a low BABIP – produced 3 WAR. A top prospect having a breakout season should hardly be regarded as some bizarre statistical anomaly.
There also exists an argument that Jimenez’s season was not a breakout – he had, after all, previously been a serious Cy Young contender with the Rockies in 2009-10. However, the Ubaldo of 2010 had far superior fastball velocity than did 2012-13 Ubaldo; given that he was – in essence – a different pitcher than he was in Colorado, it’s entirely fair to describe Ubaldo as having had a breakout year. In amost exactly parallel fashion, Kazmir too broke out. Each of the two are similar, and neither of the two are on the 2014 team – so their regression to the mean is has nearly zero impact on the Indians
Undisputable skill-based breakouts include Kluber, Salazar, and Raburn. None of their results suggest they were the beneficiaries of unduly good luck – Raburn’s BABIP was approximately in line with his career norm, but his breakout came predominantly in his walk rate, an 10.5% rate unprecedented for Raburn’s career. Yet despite this, his power numbers – though the best of his career – were nevertheless similar to his second-best season.
Kluber, once described as having the most effective pitch mix in the Indians system, had never truly seen his effectiveness translate to exceedingly good statistics. His second-degree components (Swinging Strike rate) had always been above-average (making him, perhaps, a good parallel for a certain controversial 2014 fifth starter), but only in 2013 did that translate to a genuinely above-average strikeout rate at the major-league level. Kluber’s true breakout, however, was his walk rate, which took a step forward from good to ‘top of the league.’ This, at least, statistics suggest we must regress to his career numbers – yet given the fact that Kluber’s surrendered only two walks over 20 innings in spring training – one wonders whether he took an honest-to-goodness step forward.
Indeed, given how profoundly Kluber’s skill statistics (xFIP, for instance) underperformed his ERA, even if one regresses Kluber’s skill statistics toward the mean, one should also regress his ERA to fall in line with his skill-predictive statistics, due to the bad luck Kluber had on his HR/FB rate, and due to the astronomical BABIP surrendered; that is to say, even if one expects Kluber to get worse (which is reasonable, given how excellent his 2013 was), one should also project his ERA to substantially improve despite this. So to project Kluber to be substantially worse in terms of run prevention than he was in 2013 is to not only regress him toward the average, but to return him entirely to the average. That’s not responsible statistical regression: it’s pressing ‘return to 2012 factory settings.’
In the case of Salazar, his skills broke out, as well. In his rookie year, he had the sixth-highest K/9 of any rookie starter since 1945, surpassed only by Kerry Wood, Stephen Strasburg, Mark Prior, Max Scherzer, and Dwight Gooden. It’s impossible to project that strikeout rate going forward, but his 14.6% Swinging Strike rate – an extremely stable year-to-year statistic – was surpassed only by Francisco Liriano since such statistic were tracked. Regressing his strikeout rate toward the means is the job of projection systems: and according to FanGraphs’ projection, that sort of regression would put him at 9.53 K/9: one of the best paces in the league. ‘Regression’ makes him merely an excellent #3 pitcher.
Finally, we return to luck-based breakouts, which includes only one player: Yan Gomes. As this same author has discussed elsewhere, Yan Gomes’ batted ball profile in both the granular and holistic sense did not lend itself to a particularly good BABIP – it certainly did not lend itself to a .342 BABIP. Yan Gomes’s offensive performance was entirely unprecedented relative to his career, and it was fueled predominantly by a category long known for being luck-based and unstable. Gomes, in a reasonable projection, is probably a fairly average hitter by virtue of his serious contact problems, his extremely exploitable pull tendency, and his genuinely impressive power. Projecting him for more than that is not borne out by any statistics.
In sum, Plexiglas should push back Kipnis only slightly, Kluber, Salazar, and Raburn slightly moreso, and Yan Gomes fairly severely. Regression should push back only five of the players currently on the team, and only one of them severely. It’s not clear that the Plexiglas principle applies to the Indians substantially more than any other team.
The application of the ‘Plexiglas Principle’ to deny the possible success of the Indians in 2014 is memetic regurgitation. They’re likely to be worse than the Tigers, Red Sox, and Rays, to be sure, but that’s due to talent levels, not because of regression. Seeing the Plexiglas Principle trotted out to explain the Indians’ low playoff likelihood in 2014 is a vicar clutching to his writings, denouncing dissenters as heretics, ignorant of Truth.
The problem is terrible defense, a gigantic question mark in the back end of the rotation, and the fact that they’re offense, while above-average, isn’t above-average enough to compensate for the lack of defense. That’s it. Regression is a concern pondered by the confused. ‘Regression’ would have applied to Kazmir and Jimenez. They’re no longer with the team. Given the make-up of the roster, regression is not the Indians’ enemy in 2014 – in fact, in the case of Kluber, Cabrera, Chisenhall, and Swisher, regression should help their overall result statistics substantially.
And if regression does strike the club, I’ll gladly be guiding fans through the Plexiglas doors.