One of the issues with most pitching statistics is that they show a very low correlation between current and future success. FIP and xFIP certainly help in that area, but as I discussed an in earlier piece on PER, there are some basic assumptions made in this calculations which don’t apply to every pitcher.
By no means does PER solve the issue of linking current and future success, but it can be a helpful tool.
I first became truly sold on PER in the 2010 season, and Doug Fister was a key reason why.
I had developed a formula I liked and showed it to a colleague. The leaderboard at that point in the season (through the end of May), looked like this:
Roy Halladay – .854
Ubaldo Jimenez – .851
Cliff Lee – .834
Adam Wainwright – .828
Doug Fister – .818
Matt Cain – .810
At the time, Doug Fister was a nobody. He had made 10 uneventful starts in 2009, had a career 4.22 ERA in the minor leagues and was never even mentioned among the team’s top prospects. In fact, prior to the 2009 season, he failed to crack the list of the team’s top 20 prospects, or even the honorable mentions.
So Fister’s place on this PER list generated a mocking review of the stat from my colleague. While he was on board with the concept, he thought Fister was evidence of the current formula being flawed. My response was: let’s wait and see.
Five years later, Fister has established himself as one of the most consistent pitchers in the game. And has proven capable of dominant stretches, repeating his two-month dominance again with the Tigers in 2011 and currently with the Nationals.
Another young pitcher who helped confirm my belief in PER during the 2010 season was Cardinals rookie Jaime Garcia.
By the end of May, Garcia was second in the majors with a 1.32 ERA through 10 starts. He was getting a lot of publicity as a potential third ace, joining Wainwright and Carpenter, but PER didn’t love him.
Garcia’s PER through the end of May was a solid but unspectacular .788, which ranked 23rd in the majors. It was certainly a respectable performance, but no where near the type of success his ERA indicated. While he was limiting runs, he was allowing baserunners at an alarming rate. As a result, over the next four months Garcia fell back down to earth with a 3.53 ERA and has failed to come close to matching his impressive numbers from early in 2010.
A third pitcher from 2010, veteran Jon Garland, also proved the value of PER.
Like Garcia, Garland got off to a red hot start with a 2.15 ERA through May, fourth best in the majors. As an already established pitcher it seemed reasonable to think that Garland was benefiting from pitching in the Padres’ spacious park against National League hitters. But PER downgraded him due to his walk rate and his inability to pitch deep into games (in six of 11 starts he was bounced before the seventh inning). As a result, during the same time span in which he ranked fourth in ERA, he ranked 45th in PER.
Over the rest of the season, Garland fell back down to the level his PER predicted—roughly a league-average pitcher with a 4.13 ERA in June through September.
So does PER have value as a predictive statistics? Since it’s purely an evaluation of what has happened, it can’t put a number on a pitcher’s actual performance versus bad luck in the same way as a stat like xFIP. However, PER is less prone to wild fluctuations than stats like ERA and Game Scores.
I think the best usage of PER is a way to either confirm or deny the sustainability of a pitcher’s short-term performance. Much like with Fister in 2010, if a pitcher appears to be performing at a high level, then check his PER. If it corresponds, then his performance is probably sustainable. If it’s way off, like Garland, he’s destined to fall back down to earth.
So how does this apply to the 2014 Indians?
The pitcher most worth examining is Corey Kluber, who is having a well-documented breakout year.
How sustainable Kluber’s success is depends mostly on how you view him. By no measure is Kluber pitching at an elite level. His strikeouts are impressive, but he’s only slightly above average in most meaningful categories.
Among 112 pitchers with at least 10 starts this season, Kluber ranks 32nd with a PER of .777.
Based on his rank, he’s obviously doing well, but that isn’t a number to get too excited about. Pitchers who sustain a PER above .800 for an extended period of time are the ones most likely to continue to perform at a high level. The guys in Kluber’s range, however, are a little harder to predict.
The other Tribe starter I’d like to look at is Trevor Bauer, who many fans have been pleased with so far this season.
Through his first six starts, Bauer has posted a PER of .735, below the league average of .747 this year. Six starts is a really small sample size, but what does that say about our expectations of Bauer that we’ve been impressed with his performance?
In his six starts, Bauer has yet to produce a PER giving the Indians a win probability above 63 percent.
There’s still plenty of time for Bauer to develop, but until he cuts down on his walks and pitches deeper into the game, he isn’t going to be someone the Tribe can rely on.