It took about four months, but it's finally happened; Ken Giles is the Official Closer of the Houston Astros. Luke Gregerson has been making good on his three-year contract, and Will Harris has earned his accolades, but for many of us, there was never doubt that Giles has the best stuff of the trio, at least when he’s "on." Neither Gregerson nor Harris will ever get mistaken for peak Brad Lidge, but that's exactly Giles's comp. Anticlimactic as it is, this move is key not only because it slots Houston's most talented reliever into the ninth inning, but also because it invites a (re)consideration of how relievers should even be assigned to innings, of what happens when statistics and performance diverge.
Back in March, I hated how analysts kept saying stuff like, "Look, it's not a big deal if Giles starts the year as a set-up man. Gregerson or Harris will pitch fine in the ninth, or they'll pitch their ways out of it. Besides, the eighth and ninth are comparably important." I got their point, but here's the thing: Between the two of them, Gregerson and Harris have blown nine saves this season. While some of those games still turned into wins, the 2016 Astros will never get those nine opportunities back, opportunities in which a different reliever might have shut the door, preventing losses or tiring, extra-inning games. I'm not saying that April/May Giles would've been Superman in the ninth, but in letting Gregerson and Harris "pitch their ways out of" the closer role, the Astros have paid an opportunity cost. Imagine how different this season would feel if Giles turned just three more of those blown saves into wins.
Don't get me wrong; April/May Giles was a crapshoot, and I'm not sure whether any manager would've made him their closer early on. But for the majority of the season, Giles has been the Astros' best closer-candidate. In heresy of my sabermetric faith, I think it says something about Luke Gregerson and Will Harris that they've both performed noticeably worse in high-leverage spots, small sample sizes be damned. Some relevant stats from this year, per FanGraphs:
Pitcher | BA against in medium-leverage situations (md-lvg) | OBP (md-lvg) | SLG% (md-lvg) | BB% (md-lvg) | Line-Drive % (md-lvg) | BA against in high-leverage situations (high-lvg) | OBP (high-lvg) | SLG% (high-lvg) | BB% (high-lvg) | Line-Drive% (high-lvg) |
Luke Gregerson | .125 (across 16.2 innings) | .169 | .214 | 5.1 | 12.2 | .235 (across 13 innings) | .316 | .480 | 10.5 | 21.9 |
Will Harris | .206 (across 15.2 innings) | .254 | .254 | 6.0 | 7.0 | .259 (across 15.2 innings) | .300 | .382 | 4.9 | 21.4 |
Anything can happen in 15 innings, but these are some crazy spreads, especially the ones for line-drive percentage. These numbers represent the 10-plus times Gregerson and Harris had the chance to go after the lead-off hitter, to do the little things right, to be The Guy, and they simply failed an excessive number of times.
By contrast, Giles's struggles in 2016 have come mostly in medium-leverage spots. [Insert sentence about small samples sizes.] In fact, I'd argue that Giles’s relegation to a set-up role was itself a reason he struggled. If you, an emerging MLB closer, were traded to a team who coveted your skill at closing games—and if closing games were your primary way of earning raises during arbitration—how psyched would you be to set-up for two other guys, regardless of how bad your Spring Training was? It was a kind of self-fulfilling prophecy; April/May Giles was slotted into a set-up role, and he proceeded to pitch like a guy who shouldn't be closing games. [Insert sentence about how some players initially struggle under the limelight of high-profile trades, as well as the statistical impossibility of using a trade to explain several bad innings.]
For a saber-disciple, the game of bullpen musical chairs raises questions of how well stats can predict a pitcher's performance in the ninth inning. The best way to be a successful closer, it would seem, is just to be a good reliever in general. (A similar saber-argument is used when evaluating "clutch" hitters.) The mental part is secondary or irrelevant, we stat-folks like to believe. But why, then, do there exist relievers whose set-up stats are models of consistency—since losing his closer job, for instance, Gregerson has an ERA of 1.00 and a K/9 of 12.5—but who pitch worse in spots of even higher leverage?
Well, the Astros just traded away a guy who embodies this conundrum: Josh Fields. I've contributed to this website across the past three years, and I still don't understand Josh Fields. How does a guy who’s posted season-long FIPs of 2.09, 2.19, and 2.81 have a career ERA of 4.53 across 159 innings? Well, looking on FanGraphs at Fields's career numbers, the figures from the last six columns of this chart probably torpedoed his roster spot:
Pitcher | BA against in medium-leverage situations (md-lvg) | OBP (md-lvg) | SLG% (md-lvg) | BB% (md-lvg) | FIP (md-lvg) | xFIP (md-lvg) | BA allowed in high-leverage situations (high-lvg) | OBP (high-lvg) | SLG% (high-lvg) | BB% (high-lvg) | FIP (high-lvg) | xFIP (high-lvg) |
Josh Fields | .247 (across 39 innings) | .296 | .347 | 5.6 | 1.71 | 2.91 | .287 (across 24 innings) | .414 | .477 | 15.9 | 4.39 | 4.98 |
Yeah, I'm using a 24-inning sample to support why a guy with two years of team control was traded for a 19-year-old first baseman. Did Fields have some bad luck? Hell yes. But in high-leverage spots, he faltered at what a pitcher should have an iota of control over: his ability to throw quality strikes. (Look at that high-leverage walk rate!) It's no surprise that, amid the woeful Astros teams of 2013 and 2014, Fields could never lock down the closer role. This is what a talented, unreliable reliever looks like.
In the end, the business of predicting reliever performance is even harder than we saber-disciples have admitted. While it would seem that dominance in low-leverage innings would ensure success in high-leverage ones, counterexamples abound. All a team can really do is what the Astros have finally done: Stop getting cute about who has the "hot hand" or the most veteran experience and choose the dude whose Good Stuff is better than your other dudes' Good Stuff. The over-performing guy with the "hot hand" will regress, and veterans in their 30s have a habit of getting injured, but 25-year-olds with elite fastball-slider combos tend to keep impressing. Here's hoping to the start of an era in Astros baseball.