Did Dellin Betances really wear down last September?

The 2016 season ended disastrously for Dellin Betances. After the Yankees rode Gary Sanchez’s August power surge into surprise playoff contention, the ninth inning of September Yankee games became very important. Betances, thrust into the closer role with the trades of Aroldis Chapman and Andrew Miller, struggled in those ninth innings.

The popular narrative is that Betances, probably the most heavily used reliever in recent memory, wore down under a strenuous workload. The surface numbers support that idea. I’m here to persuade you that that narrative might be spurious, but the surface numbers are so bad I can’t be offended if you remain unconvinced.

Here are Betances’ numbers, split between everything that happened before September 1st, and everything that happened after:

The split is so pronounced, Betances simply must have been much worse in September, right? It feels obvious that the big right-hander had to have worn down, after an exhausting three-year stretch in which he led all relievers in innings.

Look closely, however, and it’s difficult to see any way in which Betances was a fundamentally different pitcher down the stretch. Yes, his results were awful, but the way he arrived at those September results was no different than how he arrived at his pre-September results, suggesting that he hardly wore down at all.

What do we mean when we say a pitcher “wears down”? Anyone that isn’t literally Dellin Betances, or at least not on the Yankees, can’t really know for sure how he the pitcher is physically feeling, so we need to take context clues. Typically, when it’s said a pitcher is worn down, I believe we look for a loss in command and control as well as a dip in velocity/stuff.

You might see Betances’ strikeout and walk rates and reasonably conclude his control did desert him in September. His strikeout rate plummeted while his walk rate doubled. Yet if we just look at the rate at which Betances threw strikes, it doesn’t look like he actually lost his control.

According to FanGraphs’ plate discipline data, 42.8% of Betances’ pitches prior to September were in the strike zone. From September onward? That rate was 44.6%. According to Baseball Reference, 63% of Betances’ pre-September pitches were either called or swinging strikes. That figure was 62% in September. Essentially, the rate at which Betances controlled the zone hardly changed in September.

If that’s so, then what caused Betances’ flurry of walks at the end of the year? For one, a more selective sample of hitters could have contributed. Opposing batters in September swung at pitches out of the zone less frequently, and pitches in the zone more frequently, than they did prior to September. Also hurting Betances in September, however, were the vagaries of small sample size.

I went back and watched Betances’ outings from September, and a number of 50-50, borderline calls that could’ve flipped an at-bat from an out to a walk, or worse, went against him. Take, for example, this Melvin Upton at-bat in a September loss in Toronto:

Two pitches taken for balls that could easily have been called for strike three. Or take this Josh Donaldson at-bat from earlier in the same game:

Betances’ 3-2 pitch nicked the outside corner according to MLB Gameday, but Donaldson took it for ball four. Or even take the 2-1 pitch to Hanley Ramirez in that devastating loss to Boston in September, which can be seen at 4:01:05 of this video:

Ramirez pretty clearly fails to check his swing at a sweeping Betances’ curve, but the umpire called a ball to take the count to 3-1. Ramirez ended the game with a homer on the next pitch, a fastball down Broadway. Over the course of a long season, Betances probably gets some of these calls, but in September, they all seemed to go against him.

Now, you might assert, perhaps Betances’ stuff indicated he was worn down, even if he hadn’t lost his control. Again, there just doesn’t seem to be evidence of that. Betances relies on a fastball and knucklecurve, so here are how those pitches looked before and after September 1st (data courtesy of Brooks Baseball):

His fastball in September was harder, had more horizontal movement, and generated more whiffs, albeit in a small sample. His curveball was also harder, had more horizontal movement, and generated nearly as many whiffs. His fastball velocity in September was actually the second highest velocity he has posted in a full month in his career, trailing only June of last season, when he sat at 99.05 mph.

I have tried to find underlying numbers that support the hypothesis that Betances was far worse in September, and I simply can’t. His swinging strike rate in September was higher than prior.

I pulled Statcast data on all his batted balls, and the exit velocity he yielded in September (86.5 mph) was lower than the rest of the season (87.0 mph). He also allowed a lower launch angle (3.5 degrees) in September than before (7.3 degrees) indicating he wasn’t allowing more optimal launch angles. I even pulled the spin rates on all the pitches he threw. The spin rates on both his fastball and curve before and after September 1st were virtually unchanged.

It’s honestly striking how little Betances’ fundamental game changed in September, when compared to his massive drop in results. He gave up runs and walks in bunches, which led to an easy narrative about how worn down he was. After looking closely at Betances’ performance, however, I don’t believe that narrative holds up to scrutiny.

Draw your own conclusions, but I think the biggest lesson to take from this exercise is that baseball, as always, is weird, especially in small samples. Just because a pitcher’s ERA and walk rate spike doesn’t mean he has suddenly become a new pitcher. Sometimes, stuff just happens.