2016 was a weird season for the Atlanta Braves in a lot of ways. For one, the offense was historically poor in the early months of the spring and summer, but turned it around dramatically as the season drew to a close. The rotation was the opposite: thanks to a good month from Matt Wisler and Julio Teheran’s resurgence, the rotation was unremarkable but steady for the first half. In the second half, though, it endured very rough waters. Despite Julio Teheran soundly rounding it a bounceback season and Mike Foltynewicz showing signs of putting it all together, seven different starters posted sub-replacement level performance levels in the season’s second half. The worst offenders were Tyrell Jenkins and Matt Wisler, but a series of gong show replacements like Joel de la Cruz and Ryan Weber similarly failed to offer any semblance of support.
As a result, albeit somewhat strange for a rebuild based largely on acquiring pitching prospects, the Braves reloaded before the 2017 season by acquiring three veteran starters: the peerless Bartolo Colon, UCL-bereft knuckleballer R.A. Dickey, and Jaime “it’s just a flesh wound” Garcia to round out the rotation with Teheran and Foltynewicz. As a result, the rotation looks reasonable on paper, as far as projections go. But with Colon and Dickey getting on in years, and Garcia featuring substantial injury risk, there’s a lot that could go wrong that hopefully doesn’t.
Unlike the projections for hitters, there have been a few changes made for pitchers compared to last year. If you’re interested, I suggest at least a skim of the points below. But, if you just want to hop down to the numbers and the players, feel free, but keep in mind one thing: unless you’re looking at the row for R.A. Dickey, it’s probably best to ignore the rWAR column: it’s there mostly for fun.
Q&A: Pitcher Edition
What projections is this exercise looking at?
Three: Steamer, ZiPS, and a very rudimentary system I’ve come up with myself, which I jokingly call IWAG. Steamer and ZiPS are both featured on Fangraphs. As best I can tell, Steamer is maintained by Jared Cross, Dash Davidson, and Peter Rosenbloom. ZiPS, of course, is curated by the inestimable Dan Szymborski. The data from Steamer and ZiPS were collected off of Fangraphs’ individual player pages.
What stats is this exercise looking at?
Specifically, there are four endpoints projected for each pitcher:
- rWAR per 200 innings pitched;
- FIP; and
- fWAR per 200 innings pitched.
In case you’re only familiar with ERA, here’s a quick rundown. rWAR is basically just a way to convert the number of runs a pitcher allows into WAR. It’s not much more sophisticated than that, though it does take a pitcher’s parks played in and league into account. FIP is the basis of how Fangraphs values pitchers: it tries to approximate ERA, but does so only using walks, strikeouts, and home runs allowed. Those three stats are used because they’re ones where only the pitcher and batter are involved: every other outcome involves the defense making or failing to make a play, and therefore FIP chooses not to credit or debit the pitcher for any of those outcomes. fWAR, then, is just the same WAR principle as rWAR is to ERA, but for FIP. It is also park- and league-adjusted.
Some errata that’s vaguely important to consider: 1) fWAR also includes infield pop-ups elicited, because infield pops are basically strikeouts; 2) rWAR is somewhat confusing, because Fangraphs and Baseball-Reference use different calculation methods. This doesn’t matter for projections, but while the rWAR published on the two sites will track pretty well, it won’t be identical, which is important to think about if you’re using one or the other for precise calculations or analyses.
More important, though, is the fact that FIP is a useful statistic because it’s a better predictor of a pitcher’s future ERA than his current ERA. Projections, in and of themselves, are fairly blunt instruments: they take some numbers in, and spew some other numbers out. It is just as simple to make a projection that uses past ERA to estimate future ERA as it is do the same for FIP. However, since we already know that FIP is a better predictor of future ERA than ERA, it makes far more sense to use projections based off of FIP and fWAR to project future pitcher performance than ERA or rWAR. Just on a lark, I’ve still included ERA and rWAR projections in this exercise, but they’re really only useful for R.A Dickey, who “breaks” FIP as a knuckleballer. Otherwise, only the FIP and fWAR columns are particularly of use.
A final consideration: Steamer provides estimates for all four stats (ERA, rWAR, FIP, fWAR). ZiPS does not include rWAR. The conversion of ERA to rWAR and FIP to fWAR is subject to a few complex factors, like league/park adjustments, and adjustments for a pitcher influencing his own run environment (which is overly complicated but minor). For across-the-board consistency, I do the following:
- Using Steamer projections for ERA —> rWAR and FIP —> fWAR specifically for Braves players, I build a model that emulates this relationship. Therefore, IWAG projects ERA (again, just for fun) and FIP only, and then takes those outputs and plugs them into the rWAR and fWAR model to estimate rWAR and fWAR for players playing in however Steamer is thinking about the run environment for Braves pitchers in 2017.
- ZiPS provides fWAR projections, so I simply use those. However, I do the same calculus as above for the ERAs estimated by ZiPS. Again, you should pretty much just ignore the ERA and rWAR projections, but I want to make it clear that ZiPS/Dan Szymborski do not provide rWAR projections on the Fangraphs player pages, so yell at me if there’s something that looks strange there.
This is a short Q&A section.
Okay, first off, that’s not a question. Second, pretty much everything you need to know is above. On to the projections themselves!
The table below depicts projections for seven pitchers: the presumptive Opening Day rotation, Aaron Blair, and Matt Wisler. While a handful of teams do get away with using that few pitchers over the course of a full season, it’s probably a safer bet that they use something like 10 instead. One of those might be Josh Collmenter, who will likely start in the bullpen. Another might be Joel de la Cruz, who remains with the organization (for now). The Braves may also make some other depth additions to AAA that could soak up a start or two. In any case, Matt Wisler and Aaron Blair will likely be first in line for spot starts (or if a starter goes down...). With that said, as usual, I’m happy to produce projections for other potential starters if there’s interest, just let me know in the comments.
Player-by-player recaps follow.
How can Bartolo Colon not be first on any list? On the one hand, Bartolo Colon is going to turn 44 next season, which should dull some optimism regarding his future performance. On the other hand, he’s put up 2.9, 2.6, and 2.9 fWAR over the past three seasons. It’s not surprising that Steamer and ZiPS expect a slight step back: there’s gotta be some aging curve adjustment somewhere. IWAG is not that concerned: if he can get older and older while keeping his FIP under 4.00, who is IWAG to tell him he won’t be able to do so again?
Colon is not much of a strikeout aficionado anymore. He mostly just throws his two-seamer again and again, pitching to contact and avoiding walks. That probably limits his ceiling, but he should have no problem rolling out another season above league average unless Father Time catches up with him. Across the three projection systems, Colon always checks in as the best or second-best pitcher in the Atlanta rotation. If he pitches according to those projections, the Braves may be able to get a nice return for him at the Trade Deadline. If 2017 is the year he collapses, however, the Braves may have an ugly season on their hands.
The Braves acquired Garcia for a trio of prospects, including John Gant, whom I was quite sad to see go. In general, I covered a lot of the background regarding Garcia here, and the projections are largely in line with that.
When he’s been able to take the mound, Garcia has generally been very effective. He’s a 2+ WAR pitcher per ZiPS, and around a 3 WAR pitcher per Steamer/IWAG. Garcia had some issues with the longball last season, but unless his HR/FB ratio remains around 20 percent, he’ll probably be at least average. Of course, there’s always the chance that injuries dampen his effectiveness, or, worse yet, knock him out of the rotation for a few months or more. That’s a risk the Braves are taking, however, and the rewards could be pretty delicious if he pitches like he has in the past for the first few months of the season.
Interestingly, Steamer and ZiPS generally agree on his strikeout and walk rates going forward, in that they’ll look pretty similar to his career rates, with some more strikeouts. The main deviation is in the frequency that Garcia gets taken deep: Steamer uses a pretty generic projection of one homer per nine innings, while ZiPS is expecting 1.2 homers per nine instead. Suffice to say, there’s not much consternation about Garcia’s ability to maintain his control and command. If he can keep the ball on the ground and in the park, he should have good results on the mound, and if he can’t, he’ll start to look more like an average pitcher (or worse).
After a tough 2015, Julio Teheran bounced back to stymie National League hitters last season. One of his most-discussed qualities (besides the potential return he could bring in a trade) is his ability to beat his FIP: over the 821 innings that comprise his major league career to date, Teheran has exhibited an 0.46 ERA-FIP gap. Unlike other FIP-beaters, however, Teheran is not a prodigious groundball eliciter, nor does he particularly limit the quality of contact against him. Instead, he tends to coax fly balls that don’t leave the yard out of hitters’ bats, granting him a below-average career BABIP-against (.275) that helps to explain some of the gap. (Other partial explanations include his great ability to pick runners off, as well as the fact that he fields his position well.)
As a result, FIP and fWAR may once again continue to underrate Teheran a bit. You can see this based on what happens when you project his ERA and rWAR versus his FIP and fWAR in the tables above: by rWAR, Teheran may be expected to be well above average, approaching on elite. But that’s contingent on him continuing to beat his FIP to the same extent of around half a run of ERA, which may be a tough sell. Even without that, though, Teheran projects to post an above-average FIP below 4.00. The reality may be somewhere in the middle: Teheran may once again pitch like he did in 2016, with an fWAR around 3.00 and perhaps an rWAR somewhat higher. Critically it looks like he’s put his tough 2015 behind him, and the projections don’t expect another rocky performance for the young righty.
Folt .45 is a good test case for projections might disagree. In 2014, he got rocked at AAA and was used as a reliever for 18 innings with the big league Astros without remarkable results. In 2015, he fared better at AAA, but exhibited too much inconsistency and a lack of effectiveness in 15 Atlanta starts (and a few relief appearances). Last year, he continued to improve upon his ability to retire AAA batters, and more importantly, made 22 starts at the major league level where, when all is said and done, he clocked in as a league-average starter.
Steamer and ZiPS pretty much agree that this average starter mantle is the one Folty should be wearing heading into 2017. In general, I think that’s a pretty reasonable assumption. But, IWAG is again, very much a blunt instrument, and isn’t forgetting Folty’s sometimes-problematic path to this level of performance. The same inconsistency that plagued him in 2014 and 2015 could rise up again, and IWAG tries to account for that possibility. If I had to guess, I’d go with Folty being a special case that shows that backwards-looking projections don’t always work great, but like I said, he’s a good test case. We’ll see what happens.
Knuckleballers are weird. R.A. Dickey is weird. If a projection system had to project knuckleballers for a living, it probably wouldn’t be employed for very long. By fWAR, Dickey has been worth between 1 and 2 wins in each of the last four seasons. Of course, fWAR assumes that pitchers allow a relatively average mix of balls in play, and that’s exactly what knuckleballers don’t do.
If you use rWAR, which gives Dickey credit for all outs on balls in play, and works off of his ERA rather than his FIP, Dickey’s success has been more erratic, but higher: 2.4, 2.6, 3.6, 1.1. Of course, the worst part of both of these trends is that last season was the least successful, and Dickey actually lost his spot in the Toronto rotation before the season ended. Whether or not that has any bearing on what he’ll do in 2017 remains to be seen.
As indicated, Dickey’s FIP is probably not trending in the right direction. Still, by that measure, Steamer has him as an average pitcher, while ZiPS and IWAG are more skeptical, likely due to Dickey’s 2016 issues. After all, though, it’s just FIP, and for Dickey alone, the rWAR projections may be more meaningful. By that metric, Steamer and IWAG except Dickey to be average-to-above, while ZiPS isn’t giving him that much of a break.
Dickey probably has less of a chance to be valuable enough to be moved during the Trade Deadline, but that doesn’t mean it’s not possible. Once again, knuckleballers: who even knows?
Matt Wisler has had over 260 major league innings to date, as well as over 200 innings in AAA. The projections are pretty much unanimously over him. With projected ERAs and FIPs in the mid-to-high 4.00s, it’s not super-surprising that the Braves opted to stuff their rotation rather than give him another chance.
Still, giving him a chance to maybe improve upon his past results couldn’t have hurt, especially given where the Braves are in terms of the win curve for 2017. In general, though, Wisler has so far shown no particular aptitude for garnering strikeouts or avoiding walks, and he’s also wrestled with a serious homer problem from grooving pitches and allowing a fair bit of fly balls, along with not managing contact too well.
Wisler will probably get some starts here and there, but he’s going to have to show the Braves more than he has so far to avoid sliding even further down on the depth chart. He’s already missed his chance on this front as far as the projections go, but that’s probably not his primary concern at this point. (Or his second, or tertiary concern, either.)
You can largely take the text for Wisler and apply to Aaron Blair. On the plus side, though, Blair has only 70 major league innings under his belt, along with under 150 AAA innings. That increases the uncertainty surrounding his future performance. On the minus side, Blair’s major league results have been worse than Wisler’s so far. Either way, the projections for them are pretty similar.
If I were running the show, I’d be more willing to move on from Wisler, just given his greater exposure; Blair has more latitude to put it together despite earlier struggles. The Braves aren’t leaving themselves much room for one starter to figure it out at the major league level, much less two, so don’t be surprised to see one of them dealt. On the other hand, with a couple of old guys and Jaime Garcia in the rotation, it never hurts to have essentially major-league ready help in the minors.
(As a side note, if I were running the show, I’d give Blair and Wisler a much longer tryout in the rotation, given that the Braves have little to lose and a lot to gain from doing so. But that’s somewhat a different topic than the projection review at hand.
Next time, we’ll handle the remainder of the roster: the varied potential relief pitcher corps for the 2017 season.
But before then, something to chew on. In 2016, starters for the Braves only managed to pitch 880 innings. In general, teams are projected to get somewhat more than that out of their rotation: 920 to 950 innings or so. Using a 950-inning mark, and the same innings distributions as found on Fangraphs’ Depth Charts, the three projection systems estimate the following rotation WAR values:
- Steamer: 11.0 WAR;
- ZiPS: 9.7 WAR; and
- IWAG: 9.6 WAR.
Last season, the Atlanta rotation amassed only 4.7 WAR. The projections therefore expect a fairly substantial improvement. However, as noted above, that figure is overly depressed by a lot of sub-replacement level rotation pitching: the contribution from above-replacement pitchers was 6.9 WAR; the projections are still a fair bit of improvement.
If you recall from the outfielders post, the position player projections ranged from 10.7 (Steamer) to 18.0 (IWAG) WAR. In combination with the starting pitching projections, they range frm about 21 (ZiPS) to about 28 (IWAG) WAR.
Before moving on to relievers, figuring that an average bullpen will be around 3 WAR, you could forecast that the Braves would win between 71 and 79 games based on the current roster. The final piece, the bullpen projections, will add a bit more clarity to those numbers and round out the exercise.