The 2015 NL Cy Young award was one of the tightest contests in recent history, and despite what you think about Greinke’s old-school ERA case or Kershaw’s new-school WAR case, Jake Arrieta’s victory showed that voters (a) fall prey to recency bias, and (b) place a premium on stretches of dominance.

This post isn’t about whether or not Arrieta deserved the Cy Young or not. It’s going to look at the concept of a stretch of dominance, and see just what kind of “stretch” a pitcher needs to win the Cy Young when he’s in the same league as a guy with a lower ERA, and a guy with a higher WAR who also had the first 300-strikeout season in 13 years.

Defining the “stretch of dominance” in question is the first step here, and it’s an easy step for me because the memory of Jake Arrieta single-handedly carrying my fantasy team to a championship with his burly arms is forever seared into my brain. I’ll define it as his last 12 starts of the season, spanning 88.1 IP from Aug. 4 @ PIT to Oct. 2 @ MIL.

We’ll start with the most pertinent stat to our investigation, which is Win Probability Added (WPA). While Cy Young voters likely weren’t looking at this stat, it perfectly captures the kind of concentrated dominance needed to skew voter perception because it is the ultimate context dependent statistic. From the Fangraphs glossary: WPA captures the change in Win Expectancy from one plate appearance to the next and credits or debits the player based on how much their action increased their team’s odds of winning. In basic terms, it is a net measure of all the things that a player did to increase his team’s odds of winning, so it will weigh a clutch play like a bases-loaded strikeout in the ninth with your team up a run more than a strikeout with the bases empty in the first.

Jake Arrieta accumulated +3.63 WPA over his last 12 starts, which is the same amount of WPA David Price—who finished fifth overall among pitchers in WPA—managed to accumulate over the entire season. In other words, Arrieta condensed 220.1 IP of David Price-level clutchness into 88.1 shimmering innings.

For the whole season, however, Arrieta’s 5.87 WPA was a distant second to Greinke’s 6.79 WPA, which gives you a good idea of the insane burst of utter dominance needed to effect enough recency bias in the voters’ minds to outweigh two other pitching seasons that were by many accounts objectively better on the whole.

Here are some other fun stats from that stretch:

  • He outperformed his minuscule 1.84 FIP by a run and a half, posting a 0.41 ERA
  • He held opponents to a .136/.182/.172 triple slash
  • That .172 SLG allowed is a result of limiting your opponents to just eight extra-base hits, which breaks down into 6 doubles, 1 triple, and 1 home run
  • That lone home run? It belongs to Phillies rookie Aaron Altherr, and happens to be the only home run allowed by Arrieta over his last 96.2 IP. For your viewing pleasure:
A rare mistake pitch grooved belt-high down the middle.
  • While we’re on the topic of home runs, lets talk about his league-leading 0.39 HR/9. Arrieta allowed just 10 home runs on the season, and according to ESPN Home Run Tracker, the homers hit by Ryan Howard and Alex Gordon at Wrigley wouldn’t have gone over the fence at any of the other 29 ballparks. Just for fun, subtracting those two cheapos from his season gives him a 0.31 HR/9, a full 0.10 HR/9 better than second place Gio Gonzalez (0.41 HR/9)
  • Despite starting on an average of 4.3 days rest during the stretch (compared to an average of 4.7 days rest prior), Arrieta found a way to dial up his fastball velocity consistently. Particularly impressive was his second last start against the Pirates, where his fastball never dipped below 95 MPH once.

Screen Shot 2016-04-13 at 11.35.13 PM

I wanted to end this post with historical comps of this stretch, but couldn’t quite get the information. Baseball Reference’s Play Index is awesome, but unfortunately it doesn’t let you filter by parameters as arbitrary as “12-start stretches”, so the next best thing I could do was compare 2015 Arrieta’s September/October (his last 6 regular season starts) to every the September/October other individual pitching season from 1949-2015 (with IP>30).

The craziness begins at this arbitrary month-based cutoff: Even though the search excludes the six August starts over which he posted a ridiculous 0.43 ERA with a 9.19 K/9 and allowed zero home runs and pitched a 12 strikeout no-hitter, his Sept/Oct. split is somehow even better than his August split. We’ll call this his überstretch.

Another unfortunate thing is that Baseball Reference doesn’t give you the league and era adjusted stats that are good for comparing all-time seasons like ERA-, but the traditional stats are pretty juicy anyways:

Screen Shot 2016-04-14 at 3.01.09 PM

Jim Kaat, for your information, was a three-time All-Star who won 283 games over 25 seasons. His 4530.1 IP are the 26th-most all-time by a pitcher, and his Hall of Fame candidacy peaked at 29.6% in his fifteen years on the ballot.

Screen Shot 2016-04-14 at 3.01.26 PM

Next closest was Johan Santana’s .376 OBP allowed in his first Cy Young season. Few batters reached base, and even fewer got past first against 2015 Jake Arrieta in Sept/Oct.

Screen Shot 2016-04-14 at 3.01.49 PM

Next closest was 2001 Mike Mussina’s .610 WHIP. That makes 2015 Jake Arrieta as the only pitcher since 1949 to post a Sept/Oct. WHIP below .600.