Shelby Miller turned a lot of heads in 2015, and rightly so. Over his first two seasons as a big leaguer, Miller was not particularly impactful: his 4.12 FIP was 14th-worst among the set of 86 starters who pitched entire seasons in both 2013 and 2014 due to a very high walk rate (8.8%) and a mediocre strikeout rate (just south of 20%). Most of this, though, was due to a dreadful 2014, as he was pretty solid as a rookie. However, his arm still oozed potential, and some late-season success in 2014 may have portended that he was on the verge of breaking out in a way that went beyond an ERA-FIP gap that made him look potentially better than he was.
Neither ZiPS nor Steamer really believed in Miller all that much. and they both projected him to put up about 1.5 fWAR (which is based on FIP, not ERA) in 200 innings of work due to an FIP in the low 4.00s. I was even less sanguine, figuring he would only be about a 1-win pitcher on the back of an FIP around 4.40. Luckily for everyone, I was really wrong, and ZiPS and Steamer didn't look too smart either on Miller's behalf, as he clocked in with 3.4 fWAR on the back of a 3.45 FIP. His RA9-WAR, which is based on ERA, was 3.7, owing to an ERA about half a run better (3.02) than his FIP. So, at least for a season, Miller shut everyone up a bit, and that's a good thing for the Braves and their fans.
To put his performance in context a bit, Shelby Miller was potentially more than just "pretty good but not amazing" when compared to his peers. There were 87 pitchers who qualified for the ERA title last year. Miller ranked 25th among them in fWAR, 14th in ERA, and 25th in FIP. If you expand the set of his peers out to 100 innings of work as a starter (I like this cutoff because it includes about 130 pitchers, which is basically four-plus starters for each team), Miller is still 14th in ERA, though he drops to just outside the top 30 in FIP at the 32nd spot. So, leaving the hackneyed "ace" debate aside, Miller was pretty much a #1 starter in 2015, even if he was towards the bottom of that range. That in and of itself could stand as a short postmortem to his season:
tl;dr - Shelby Miller was pretty damn good for the Braves in 2015. Yay.
Given that, I do think there were some interesting things worth exploring in Miller's 2015 performance. I find him to be rife with material to write about, both because of his potentially-unexpected level of success, and because of the ways in which he goes about making his living seem to lend themselves to analysis better than a lot of other players and their performances.
Some Stuff about Splits
Nothing about Shelby Miller's basic splits really jumps at you when you glance over them. However, there is something pretty weird there: his splits against lefties on the road. Overall, it's not surprising that Miller is harmed by the lack of a platoon advantage: his career FIP is 3.43 against righties but 4.29 against lefties. This gap was even more pronounced in 2015, as he dominated righties to the tune of a 2.88 FIP, but still allowed a 4.08 FIP to lefties. What's crazy, though, is the fact that this handedness split is driven by where the games take place:
First, Miller has always had somewhat dramatic home-road splits. I don't know what the real cause of that is, or whether it's just a weird anomaly that will straighten out over time (it's normal for pitchers to pitch somewhat worse on the road than at home for various reasons, but not to the extent that it afflicts Miller). But, his platoon advantage on the road is all sorts of crazy, while he pretty much has no platoon advantage or disadvantage at all at home. So there's something about facing lefties on the road that really trips him up, and it got even crazier against him this past season. Note that this isn't just a crazy HR/FB or anything driving the results: while Miller's HR/FB is higher on the road against lefties, I think this is not the sole cause but a symptom of an inability to pitch as well against lefties on the road. The biggest culprit, to me, is that he posts a strikeout rate considerably lower (14%) against lefties on the road than he does overall (20%) or even against lefties at home (19%). The walk rates aren't all that different, so it's not a direct control issue, though it may be command-related. In any case, it's kind of weird, and something interesting to look for going forward.
Pitch Mix and Stuff
Shelby Miller made 33 starts in 2015. In 19 of those starts, he allowed two or fewer runs. His FIP was below 3.00 in 14 starts. On the other hand, he got touched for five or more runs in six starts, and had nine starts in which his FIP was above 4.50.
On a start-to-start basis, I've generally been interested in the following affecting his performance:
- Swinging strike rate
- Pitch mix, in general
- Pitch mix in terms of proportion of four-seamers to two-seamers
- Pitch mix in terms of non-four-seamer/two-seamer pitches thrown.
The chart below plots these things (and looks kind of like rainbow spaghetti).
This is a big chart, and it's not actually as interesting as I'd have hoped. You can see that ERA is really volatile (makes sense), FIP is much less volatile (also makes sense), and that there's a fairly calm period between around starts 12 through 26 (this is pretty much June through August) where Shelby really shone in a pretty consistent way. You can also see that in April and early May (first seven starts) he exhibited a pretty strong ERA-FIP gap that pretty much evened out soon thereafter. Shelby's only "struggles" really came in September, where his stats basically backslid horribly across the board.
In the end, though, Miller's start-to-start numbers don't bear out a lot of things that seemed like fairly decent propositions, at least to me. In the end, I think he was worse in September because he was in worse in September, perhaps from fatigue. Overall, though, Shelby Miller just largely found a way to be Shelby Miller, and kept going at it.
Swinging strike rate
In general, swinging strike rates are really important for pitchers. I wrote about this very early on in the season, and luckily Shelby Miller's whiff rate ended up at 9.20%, well above his paltry 5.75% average over his first four starts as a Brave. One interesting thing is that Shelby Miller's whiff rate was in the lower half of qualified starters this season, and just slightly below average overall (average is about 9.50%). However, he still managed to be a good pitcher despite not eliciting many whiffs. In fact, he posted a paltry 7.75% whiff rate in June, but managed to put up a sub-3.00 FIP for the month. Not many guys can pitch to contact and put up good numbers - this past season, only a handful of guys like Kyle Hendricks, Jordan Zimmermann, and Lance Lynn generated fewer whiffs while still putting up sparking FIPs and 3+ fWAR - but Miller definitely appears to be one of them.
Overall, throughout Miller's 2015 starts, there was almost no relationship between the amount of whiffs he generated in a start and his ERA or FIP therein. You can see this in the chart (it's the faded green dotted line) - sometimes the line being in a prolonged valley corresponds to a strong period (see starts 10 to 15ish), sometimes an increase in the line matches up with relatively poor results (starts 16 to 18, 28 and 29, etc.) I was worried that an inability to generate swinging strikes might doom Miller; however, he's picked up the pace to non-historic, slightly below average levels, and succeeded without getting as many swing-throughs as his peers.
Two-seamer rate
Sinkers have a fairly interesting place in today's baseball environment. On the one hand, outs are outs, and grounders sometimes lead to rally-killing double plays. On the other hand, balls in play can only be somewhat controlled by the pitcher eliciting them, and as such, getting outs via strikeout is generally preferable in player valuation to getting a bunch of rolled-over grounders to short and second.
Shelby Miller mixed and matched his two-seamer and four-seamer. He threw more of the former in 19 of his 33 starts (including one where he threw the same amount of both), and more of the latter in the remaining 14. His ERA and FIP were more or less the same regardless of his fastball mix: 2.96 ERA and 3.48 FIP in the 19 starts with more two-seamers, and a 3.11 ERA and a 3.40 FIP in the 14 starts with more four-seamers. Overall, pretty much no correlation exists between Miller's fastball choice and his outcomes.
From the chart, you can see that Miller's two-seamer rate ebbs and flows over fairly long durations. However, these durations don't directly match up with his effectiveness or lack thereof. Another thing to chalk up to him learning how to pitch effectively with the tools at his disposal.
Cutters and other stuff
If the two-seam and four-seam fastballs are Miller's 1A and 1B, then the cutter surely clocks in as his third pitch of choice. Overall, Miller used it about a fifth of the time, and this usage was fairly consistent: he never threw it less than 11% of the time in any single start, and only five starts featured him using it for 30% or more of his pitches. Incidentally, there's nothing special about those five starts: while one of those was his worst start of the season (didn't make it out of the fifth against the Nats while allowing seven runs) and featured his highest overall cutter usage (36.2%), the remaining fourwere pretty average Shelby Miller starts by FIP. ERA-wise, he was roughed up in two of those starts, but one came against the Blue Jays post-Trade Deadline and the other was at Coors Field, so again, not much to draw on in terms of trends or conclusions here.
At some point, I surmised that Miller's use of the cutter was tinkering based on the fact that his latter-month starts were just killing time in a lost season. That still might be the case, as his cutter ratio increased substantially in September. But even that lasted for just a few starts, so it's hard to see anything definitive there. Unlike other stats, there's a small correlation (R=0.40, R^2=0.16) between cutter usage and his ERA, but even this ill effect is probably driven by those few outlier starts and doesn't show up in his FIP. Tinkering or not, there's no indication that Miller's cutter is impacting his effectiveness or that he should stay away from it, or anything like that.
It should not surprise you that there's a small correlation between non-two-seam/non-four-seam pitches thrown by Miller and a rise in both ERA and FIP. If his bread-and-butter aren't working he's forced to resort to secondary pitches, which are less effective because they're secondary. There's something there, but the effect is small. Expect Miller to continue working off of his fastballs going forward -- he has no reason not to.
Other random findings
Just out of curiosity, I also ran a bunch of correlations between most of Miller's stats and ERA/FIP on a start-to-start basis. There are some marginally interesting things here, including:
- Because Shelby does not rely on strikeouts as much as other pitchers, there's not that much of an impact on variations in his strikeout rate on either his ERA or FIP. However, his walk rate does influence both numbers substantially. This makes sense, as pitchers who rely on balls being put in play can't afford to issue free passes and get BABIPed to death (this happened to Miller occasionally, especially later in the year, and should not be read as "he got discouraged by the lack of run support" but instead as "BABIP is a thing.")
- Unsurprisingly, his BABIP-against and LOB% have huge impacts on his ERA but pretty much none on his FIP. This is mostly just a reinforcement of why ERA is kind of bogus as a way to measure outcomes resulting from the accrual of plate appearances in which a pitcher participates. But I think we already knew that.
- Here's a fun one: there's pretty much zero correlation between Miller's groundball percentage in a start and his ERA. You'd think that this would definitely hold true for his FIP, but it's really pretty much zero for both. I think this is really BABIP-related silliness, but it's still worth mentioning. Miller wasn't effective because he got lots of groundballs, he was effective by getting ahead of hitters and making good pitches, which often resulted in batters getting out because they hit weak grounders.
- The more Shelby pounded the zone, the better his outing. This is probably true for all pitchers, but among the correlations, this one is borne out, unlike the other plate discipline stats. In other words, something like o-swing% did not have a noteworthy impact on Miller's outcomes, but how many pitches he threw in the zone did. Strike-throwing: it's awesome. However, there wasn't much of a correlation between first-pitch strikes and outcomes, so it's only the overall rate of strikes that seems to have a big impact.
Looking Ahead
I don't really know what to expect from Shelby Miller in 2016. On the one hand, I don't see much of a reason for his 2016 to be different from his 2015, assuming that he can stay healthy and repeat all of his mechanics. While there are some things that might be some cause for concern, like an xFIP that's over half a run higher than his 2015 FIP, it's not hard to believe that he has some homer-suppression abilities, and stuff like his crazy road lefties split evening out a bit should give him a bit of a boost in 2016 anyway. Without doing a lot of thinking about it, it seems reasonable to me that he'll floor next season as an average full-time starter, or a little better (say, 2.2 fWAR). Whether or not you think he'll surpass that is based on a lot of things. For what it's worth (pretty much nothing), Steamer projects Miller to be worth 1.8 fWAR next season. Everyone underestimated him before 2015, we'll see if the same thing happens again. I'm willing to be wrong twice, but if he puts up a 2016 that resembles or improves upon his 2015, I'll definitely stop.
One thing that's come up recently is whether the Braves would consider an extension for Miller. I think that this is a tricky proposition, because it's possible to value Miller either at his 2015 performance level or his overall career performance level, which differ very widely. The Braves would likely want a discount through arbitration and a less-discounted free agent year (or two) as "compensation" for shelling out guaranteed money; whether Miller accepts it likely depends on his evaluation of himself and which basis of future performance the Braves are using to price him.
Using Julio Teheran's extension as a basis yields figures of an approximately 25% discount for a player through their arbitration years, and a free agent season at about a 10% discount, (These numbers are rough, and based not on the overall $/WAR a player would earn on the open market but more what the Braves would pay for such a player given that they don't play in the deep end of the pool much.) Given that, I could see a low starting point for a Miller extension to be $24 million over four years (this includes on free agent season). Miller's camp might counter with something a lot closer to $38M over four years instead, which is a substantial gap that speaks to how much better Miller's 2015 was than his career up before this past season. Overall, I can see a reasonable middle ground be $34M over 4 years or $43M over 5 years, but I would be surprised to see Miller extended this offseason. Depending on what happens next season, a stabilization of performance may lock him in as a guy that will get big bucks on the open market, or a regression might lead him to seek guaranteed money at a price point the Braves might be comfortable with.
In any case, Shelby Miller had a great 2015 season. Let's all celebrate by writing a letter to Steamer wondering why they hate him so much.