What happened to Chapman this season? Relief pitcher performance has more variance than other positions due to very small sample sizes. In other words, sometimes relief pitchers just get unlucky. However, that was not the case with Chapman in 2017. In the graphs below, I’ve aggregated Chapman’s performance by month, combining the injury-shortened May and June:

Chapman had two very bad months – the May/June period and August. During the May/June period, his strikeout rate remained excellent. However, it dipped significantly after coming back from injury in July, and really took a dive in August. After Joe Girardi benched him in August, Chapman recovered both in strikeout rate and ERA; he didn’t allow a single earned run in September.

But what role did batted ball luck play in all of this? And what about Chapman’s fastball velocity? For these questions, we turn to Statcast’s xwOBA and velocity leaderboards:

Well that’s cool: Aroldis Chapman was remarkably consistent in terms of xwOBA throughout the year, other than August. His batted ball luck was very unlucky in May and June, and mildly unlucky in July. His fastball velocity was stable before shooting up in September. Chapman was only legitimately bad in August, but he recovered in September to have his best month of the season.

The data tell us a clear story: we can all exhale. Aroldis Chapman is probably going to be an elite closer next season. The Yankees didn’t buy damaged goods; the good just got temporarily bruised in August. Aroldis Chapman is still a top-5 closer in baseball, 3.22 ERA or not.

Tanaka’s top line statistics on the season:

• 30 starts
• 178 1/3 innings
• 9.8 K/9, 2.1 BB/9
• 4.74 ERA, 4.31 FIP, 3.66 DRA
• 3.8 WARP

Already, a blurry picture emerges. A pitcher with a 3.66 DRA is pretty good, and definitely worth three years, $66 million. A pitcher with a 4.74 ERA, even in Yankee Stadium, is below average.

Of course, Tanaka got better as the season went on. His ERA was 3.77 after the All Star break, and Tanaka was dominant in the playoffs. Was Tanaka just unlucky in the first half, or is there a bigger problem?

Statcast provides us the perfect tool to answer this question. On every ball in play, Statcast records an exit velocity and launch angle. These two variables can be combined to produce very reliable predictions of the run value outcome of the batted ball. Combined with strikeout and walk data, Statcast produces a statistics called xwOBA, or expected wOBA. A full explainer is here.

How did Tanaka perform according to xwOBA this season? Below is a 3-game rolling average for the regular season:

The data tell a very clear story. At the beginning of the season, Tanaka pitched like an above-average (although only slightly-so) AL pitcher. Something went horribly wrong in May, and he turned the average batter he faced into peak Aaron Judge. He corrected something in June, then performed consistently like his 3.77 second-half ERA for the rest of the season. For what it’s worth, PECOTA projects almost exactly the same ERA for Tanaka.

I don’t look at these data and see an ace. I see a pitcher who, at his best, is pretty good. He also has enormous downside. We’ve known about Tanaka’s downside for awhile now. When he’s off, he allows home runs. A lot of home runs. It makes sense that he’ll have rough patches.

If I were Masahiro Tanaka’s agent, I’d think twice about opting out. There might be a few teams willing to overpay for his amazing playoff performance, but it’s not clear that he’s worth a $100+ million contract. If he does opt out, I hope Brian Cashman passes.

Photo Credit: Robert Deutsch-USA TODAY Sports