The 2017 baseball season hasn’t taken place on the fields and in the stadia of North America yet; however, it has taken place in computers. In fact, it’s taken place multiple times in computers. While this is not nearly as fun as watching baseball being played on the field, it’s February, and it’s cold (or at least it’s supposed to be), and baseball on computers is all we have right now.
Baseball Prospectus recently released its PECOTA projections. One of the most fun and informative things about these projections, as opposed to other publicly available ones, is that it offers a range of possible outcomes for a player. Looking at the 10th and 90th percentile seasons puts additional context around what’s usually seen as the projection, the median, 50th percentile projection. A look at ranges for Wins Above Replacement Player (WARP)—which measures overall value—from bottom to top allows us to imagine the season on a spectrum. The lesson is that the Rockies have the talent to contend, but the youthful roster has a lot of volatility.
★ ★ ★
The 10th percentile projections, with the exception of Nolan Arenado, peg Rockies starters between replacement level and average. Among the expected starting position players, four have 10th percentile projections that sit below replacement level, and three of them are well below. Tom Murphy and David Dahl have low-end projections about one win below replacement. But part of this is probably due to unknowability, as both players are young and haven’t had very many major league plate appearances. I can imagine the hole in Murphy’s swing being exploited, leading to a poor season, but I have an extremely hard time imagining Dahl getting substantial playing time and offering negative value.
Ian Desmond’s -0.7 10th percentile WARP is imaginable because he’s only two years removed from a season just like it. That season, he hit .233/.290/.384, but because he played shortstop, he was a valuable player overall. His 10th percentile projection has him hitting a very similar .243/.291/.408, but a first base adjustment (estimating neutral defense) puts that below replacement level. The projection is not only imaginable, but it’s also demonstrative of the risk the Rockies are taking with Desmond.
The best and worst of possible worlds, position players
| Player | 10 | 50 | 90 |
|---|---|---|---|
| Player | 10 | 50 | 90 |
| Tony Wolters | 0 | 0.7 | 1.9 |
| Tom Murphy | -0.8 | 0.8 | 2.9 |
| Ian Desmond | -0.7 | 0.8 | 2.7 |
| DJ LeMahieu | 0.1 | 1.9 | 3.9 |
| Nolan Arenado | 2.3 | 4.3 | 6.5 |
| Trevor Story | 0.4 | 2.4 | 4.8 |
| David Dahl | -1.2 | 0.6 | 2.8 |
| Charlie Blackmon | 0.2 | 2 | 4.1 |
| Carlos González | -0.2 | 1.6 | 3.6 |
Desmond’s 90th percentile projection further highlights the risk. It projects that he’d have to hit .305/.360/.515 while posting neutral defense at first base to accrue about 2.7 WARP. This best version of Desmond would be the best offensive season of his career, though the numbers are achievable with a little help from Coors Field. He could do it, but if I had to guess whether he’ll end up closer to his 10th or 90th percentile projection, I’d go with 10th.
It wouldn’t be surprising to see the other position players reach their 90th percentile projections, mostly because many of them (Murphy and Dahl excepted) just had seasons that either reached or exceeded them. In particular, Charlie Blackmon and DJ LeMahieu probably had career years in 2016, but if they “decline” to their projected 90th or even 80th percentile projections for 2017, it will be a boon to the Rockies’ contention chances.
Perhaps the most notable projection here is Arenado. His “best version” 90th percentile projection is about one win less than what he produced in each of the last two seasons. Projection systems are, by nature, conservative. But it still seems strange that Arenado can put up consecutive 7-plus WARP seasons and still have a 90th percentile projection below seven. He’ll just have to do it again and change PECOTA’s mind, I suppose.
★ ★ ★
The volatility mostly appears in the pitching. These projection ranges capture what we presume about relying on a young pitching staff to push the team to contention: it can go terribly right or horribly wrong. On one extreme, if each of the four pitchers almost assured a spot in the rotation if healthy and the two most likely to fill in as the number five lurk in the dregs of their 10th percentile projection, it’s easily the worst rotation in team history. At the other end, if each starter lands in his 90th percentile projection, it will be among the best rotation’s the Rockies have ever had.
Let’s recall the conservatism of the projection system and the lack of information young pitchers offer. It’s easy to envision Gray and Hoffman, in particular, reaching and even surpassing these 90th percentile projections.
The best and worst of possible worlds, pitchers
| Player | 10 | 50 | 90 |
|---|---|---|---|
| Player | 10 | 50 | 90 |
| Jon Gray | 0.1 | 2.1 | 4 |
| Tyler Anderson | -0.3 | 1.8 | 3.9 |
| Chad Bettis | -1 | 1.1 | 3 |
| Tyler Chatwood | -1.3 | 0.5 | 2.1 |
| Jeff Hoffman | -0.5 | 1 | 2.3 |
| Germán Márquez | -0.2 | 0.3 | 0.8 |
While it’s not likely that all or most of the rotation will end up on one side of these extremes, it’s useful to think about what it would look like if they did. A young pitching staff is not easy to predict, and these projections demonstrate that. If the pitching, on the whole, trends toward the 10th percentile, then it doesn’t matter as much what the offense does. It will be tough to compete. There’s a lot of justified optimism going into the 2017 season. But the hope should be measured.
★ ★ ★
These 15 players received projections based on what PECOTA knows about the players. But we also know things, and we even know things PECOTA doesn’t, such as the fact that Blackmon’s improvements at the plate over the past two seasons are due to deliberate decision making. Based on what we know, we can level our own judgments about the players, as well as the projections.
Using the 50th percentile projections as a baseline, I’d wager that Dahl is the most likely to “beat” his projection, if only because it is pretty pessimistic for a player of his talent and pedigree. Arenado is also pretty likely to exceed his median projection of 4.3 WARP. The position player most likely to fall short of the median is probably Desmond. On the whole, however, it just feels like most of these players are better than their median projections.
On the pitching side, I’d pick Chatwood as the most likely to fall short of his 50th percentile projection; Gray and Anderson seem like very good bets to do better. They’re the most talented pitchers on staff, and they have both experienced major-league success, and they should be able to parlay talent and experience into above average seasons in 2017.
Ultimately, the best and worst of these possible worlds have to combine in the right way for the team as a whole to contend in 2017. And for that to happen, the young pitching needs to come through. There are ways to put these projections together to get a successful team, and that’s fun to think about for now. But it’s nothing like watching the actual baseball unfold.