Baseball Prospectus came out with their PECOTA predictions for the 2017 season, and I took a look to see how the Nats fared. Remember that this comes from a set of algorithms that predict performance based on past performance, so there is rigor behind it, although it is bound to be wrong, in no small part because there are probabilities involved.
The Nats, by the way, are predicted to finish 87-75, 1 game behind the 88-74 Mets. It’s basically a statisical dead heat, and… wait a second, this sounds familiar. What did I write last year? (cue wavy lines):
Before we get into the individuals, the Nats are predicted to finish 87-75, 4 games behind the 91-71 Mets. It’s not quite what you’d call a statistical dead heat, there is certainly more love for the Mets there. But, a little bit of overperformance by the Nats and underperformance by the Mets, the exact opposite of 2015, would be enough to swing the standings the other way pretty easily. And overperformance by the Nats is probably as easy (or as hard) as just staying healthy.
Yeah, it sort of underlies what I’ve been saying for the last few seasons. The Nats and the Mets are about the same amount of good at baseball. In 2015 the Nats weren’t very healthy and couldn’t get far enough ahead of a not-totally-healthy-either Mets team before the bullpen blew up and the Mets got healthy and good. In 2016, the Nats stayed pretty healthy while the Mets didn’t, and the Nats coasted. Healthiness is close to godliness when it comes to this stuff.
That being said, as I look at individuals, I won’t spend much time on the health downside risk, which could sink this team as it does to many teams each year. I know PECOTA tries to factor in health, but if someone who should be a good player can’t play most of the year, they’re gonna be worse than any projections, so it’s not really noteworthy.
The projections are based on Baseball Prospectus’ WARP statistic, which is just like the familiar WAR but it has a P at the end of it. So let’s look at some of these projections, and where I see some things I think don’t look quite right.
Adam Eaton (2.9) – Ok, so Eaton had a 7.2 WARP last year, and everyone has already told you why that’s unattainable again because it was based on RF defense, which he won’t do in CF. I agree. But his 4.0 WARP in CF in 2015, following a 2.9 WARP in 2014 at age 25 with only 538 PAs means he’s going to do better than this, assuming health. I just don’t know how much better. I think 4.0 is realistic, much higher may be a stretch.
Daniel Murphy (1.8) – If 2016 didn’t exist, this would be a perfectly good assessment. Looking at his trend, pretending 2016 didn’t exist, you can see how 1.8 fits precisely right in there. But 2016 happened, and this is probably an algorithm not being able to understand why and just assuming it’s an outlier. Sorry infelxible computer, 600 PA Murphy does way way better than 1.8 (keep the health of his buttocks in your thoughts and prayers as always).
Max Scherzer (3.4) – Scherzer’s last 4 WARP totals in chronological order: 6.8, 6.6, 6.5, 6.0.
Anthony Rendon (2.1) – If he plays, he’ll do better than this, which he did in 2016 (3.8) and 2014 (5.4). This looks completely like a health discount.
Jayson Werth (0.8) – This is understandable based on age and a terrible 2015 which factors in to the math. And, of course, his fielding brings his numbers down. But if he hits like he did last year, which I am predicting through my own internal projection system (my heart, dammit, my heart) is pretty likely, he’ll be a win or so higher than this.
Trea Turner (5.5) – This projection feels high, but coming off a 4.0 WARP half-season it certainly seems attainable. If he’s as good as he was last year, he can exceed this, but adjusting to the new position and potential for sophomore slumping seems higher probability to me. Not that I see a big dropoff, but maybe he won’t be one of the top overall players in the league. Maybe.
Stephen Strasburg (3.2) – This number feels about right in terms of a good Strasburg, 20-25 start season. 30 starts he’ll be better. But I don’t even know if he’s really 100% at this point. If he ever figures out how to stay healthy, not go a month with pains that affect his performance, and start 30 games, he’s gonna touch 5 or 6 WARP. For now, I think they’ll be thrilled with a 3.2 WARP season from him (if it includes pitching in October).
Bryce Harper (3.8) – Ok, so 3.8 is a borderline All Star, and it’s the second-best position player projection on the team. That being said, he had an 11.0 WARP in 2015, and even in 2016 he was at 4.6. On the other hand, that 4.6 was probably dragged down by his post-June play. I’m not sure anyone thinks he is the player he was the second half of last year, but I don’t think anyone has any evidence to show this. I lean towards upside with him, but he’s just got too many question marks to say he’ll “probably be better than a borderline All Star.” It probably just goes to health with him, and I have no idea where he stands there.
Derek Norris (1.4) – Had a 0.9 last season and a 3.5 the year before. If he hits, this number is probably too low. If he doesn’t, it’s too high.
Ryan Zimmerman (0.1) – A 0.1 WARP means that they project a player from AAA – say, Jose Marmolejos – would be just as good. Of course, it’s more complicated than that, but Zim is a replacement-level first baseman in this projection. You could easily envision a scenario where he is significantly better than that, even if he isn’t near All Star level. But that’s more in the hopes and dreams category at this point until he shows otherwise.
Tanner Roark (1.1) – This number seemed really low to me, and I was ready to put him in the upside category. And there is certainly some discount for his poor 2015, and lack of time as a starter. But his WARP in 2014, when he started 31 games and had a 2.85 ERA, was 1.1. Now, last year he clocked in at 3.8, probably due to his improved K rate, but there’s something about him that WARP, rather than PECOTA, doesn’t love. I could see him starting 30 games, having an ERA+ over 130, being on the periphery of the Cy Young conversation… and still having a WARP below 2.0. Tanner Roark, the most “eh?” of all the projections here.
There you have it, all the reasons why the PECOTA thinking machine will be wrong. Of course, it could be totally right. But then it would probably have to be destroyed, or we’d risk it’s complete domination of the planet.