Making Sense of PECOTA’s Jays Projections (Is Actually Possible)

Nobody is going to deny that the range of possible outcomes for the 2018 Blue Jays season is pretty wide, and that it tilts a little too much toward “not good enough” for anybody’s taste. But to project them to 78-84? Six games back of the Rays? Uh… that seems a bit low. And yet that’s where PECOTA, the well-known projection system from Baseball Prospectus, which this morning had its 2018 roll-out, seems to think they’ll end up.

I hope this isn’t just my bias showing — I don’t think it is — but “well-known” seems to me, on first blush, like it’s maybe the crucial term here. I don’t want to denigrate the hard work of those running the computers behind the scenes too much, but if this all materialized from some nobody, without the PECOTA branding on it, how seriously would anyone take it?

Then again, last year I thought the projections were light on the Jays, and PECOTA had them at 81 wins!

Advertisement - Continue Reading Below

But that doesn’t necessarily mean that they weren’t off in significant ways (the 3.2 WARP projection for Tulo and 3.1 for Bautista were both… um… generous). Though, if you start digging on any projection system you’re going to find plenty of things that don’t make a whole lot of sense. Even the Depth Charts projections at Fangraphs, which I feel more comfortable with — though that, uh, probably has a lot to do with the fact that they have the Jays at 84 wins, just two behind the Angels and the second Wild Card spot — has the Jays getting 2.5 wins out of Troy Tulowitzki in 2018, which seems… optimistic, sadly.

Depth Charts also has these Jays as the fifth highest scoring offence in baseball, which… shit, maybe I don’t feel more comfortable with those.

And that’s sort of the thing: none of these projections are really worth much if you just take what they’re spitting out and go no further.

So, with that in mind, let’s maybe go a bit further on what PECOTA is saying before we go and say pissy things like this:

Advertisement - Continue Reading Below


On the Pitching Side…

  • The big, obvious weirdness here jumps right off the Jays’ PECOTA projection page (because his WARP is in red — i.e. below replacement), and that’s Marco Estrada. BP’s metrics do not like Marco Estrada. In 2016, when Estrada was worth three wins per Fangraphs, 3.5 per Baseball Reference, and 3.9 per FG’s RA9-WAR (i.e. WAR based on runs allowed, rather than FIP), BP had him at 0.7 WARP. He projects this season to -0.1 WARP. Even if Estrada repeats his poor 2017, that’s 0.2 WARP, so -0.1? Eff that! Which is to say, fuck that. Bump that up! (In your mind.)
  • Sticking with pitching, the other one that feels light is 1.0 WARP on Aaron Sanchez. It’s understandable why that would be — Sanchez doesn’t have much of a track record, and his injury-ruined 2017 was ugly. But he was a top 25 pitcher by WARP in 2016, posting a 3.7 mark. I’d bet the over on that one.
  • Other than those two, there’s not a whole lot to quibble about on the pitching side: Stroman at 2.6 WARP, Osuna at 1.8, Happ at 1.7 (which is probably light, as he was at 3.0 in 2017, and 1.7 would be his lowest mark since 2014, though that he’ll be 35 years old understandably pulls our expectations back), about another win from the bullpen, and a half win out of starters five through nine (named here as Biagini, Guerrieri, Borucki, Pannone, and Rowley).
  • In the overall, PECOTA has the Jays allowing 792 runs as a staff. That’s 100 more runs than the Rays, who themselves are behind the Yankees, Red Sox, Astros, and the Cleveland baseball franchise. The Jays’ projected total is also behind the Twins, Mariners, and the Angels. Now, 792 runs works out to 4.89 per game, which isn’t too far off where the Jays were at last year (4.84), but should we buy that? Sure, they’ll probably take a hit in terms of middle infield defence, because they had a couple glove-first guys there far too often last year (though Ryan Goins had an uncharacteristically poor year by the metrics — a UZR of -4.7 in 670 innings at short, and -5 DRS). But the outfield defence ought to be better. Starters number six through nine ought to be better. Estrada can’t not be more valuable than the algorithm here sees it. And they ought to get a whole lot more innings out of the 2016 AL ERA leader. So… giving up more runs than last year?
  • Continuing this thought: in 2016 the Jays allowed 666 runs. I don’t think anybody should expect that kind of success again, but if that was their projected total, it would make them the third best staff in the American League (just one run better than the Yankees). I think this is worth considering becase PECOTA’s win-loss totals look a whole lot like Pythagorean records based on what they’re projecting for runs scored/allowed. Adjust the Jays’ runs allowed down to 2016 levels and their Pythag is 92 wins! That’s a bit rich, obviously. But split the difference between what’s projected and what they did in 2016 and you get 729 runs, which when used with their 767 runs scored projection gives you a Pythag of 85 wins. I could certainly buy that.

On the Hitting Side…

  • All of the win totals above assume that PECOTA’s 767 runs scored projection is about right, but it’s worth noting that it’s a big jump from the 693 runs they scored in 2017. Do we really buy that the Jays have another 70 runs in them? I obviously try to be an optimist as much as reality allows when it comes to this team, so take my opinion for what it’s worth, but… yeah, I think so. More Donaldson, Granderson over Carrera, Grichuk over Bautista (frowny face), no Goins and Barney (and if not Tulo and Travis, at least what look like competent MLB hitters in Solarte and Diaz), and better minor league cover up and down the lineup (Teoscar, Alford, Jansen, etc.). I buy this side of PECOTA’s projection more than the other.
  • That said… poor goddamn wonderful Tulo is projected to 2.3 WARP, based almost entirely on an incredible track record that now feels like ancient history. The last time he exceeded 2.3 WARP was 2014. He was worth just 0.3 last season, which even if prorated over 600 plate appearances (a number he hasn’t reached since 2011 — woof!) jumps to just 0.7.
  • On the other side of that coin you have Josh Donaldson, whose 3.7 WARP projection sure seems light. Donaldson played just 113 games last year, struggled through a bunch of those (236 PA into his season, on July 22nd, his wRC+ sat at just 110), and still managed a 3.8 WARP. That total was his lowest since 2012. From 2013 to 2016 his totals were 6.2, 6.9, 7.5, and 5.4. I’d not just bet the over on Donaldson, I’d bet the over and that he makes up the difference between Tulo’s projection and reality.
  • Beyond that everything looks reasonable enough: a little high on Morales (0.8 WARP), a little low on Smoak (0.7); maybe little high on Martin and Pillar, too (2.1 WARP and 0.9 respectively, compared to 1.3 and 0.1 in 2017); possibly some room for more out of Granderson, Pearce, Grichuk, and Carrera (projected 3.6 WARP combined); and maybe even more ceiling from Solarte, Travis, and Diaz, who get just two wins combined.


Advertisement - Continue Reading Below

No, but seriously, take a little closer look, think about the flaws of the system, and I think the picture you get is pretty close to what you’d expect. I began this piece with the line, “Nobody is going to deny that the range of possible outcomes for the 2018 Blue Jays season is pretty wide, and that it tilts a little too much toward ‘not good enough’ for anybody’s taste.” Still sounds about right to me.

  • these projections are just a little too all-over-the-map to be of much value. some are just plain odd and way out of line, like they’re just guessing that guys will have big drop-offs or long-term injuries (JD, sanchez), while in other cases, they’re (inexplicably) factoring in not only sustained health but a return to form not seen in years (tulo). i mean, any one of *us* lowly commenters who had these types of prognostications would be laughed off the site.

      • Paul Beestons Grass Surface

        As Malibu Stacey would say “…Math is hard”. On a serious note….we’ll see. It might be math but its not exact in terms of error ranges which you point out. Otherwise why bother playing. Skip straight to the projected playoffs.

      • i’m sorry, but for JD, how does that ‘math’ actually work? i mean, if i simply run a string of data points together, like X X X X X, the logical next point wouldn’t be, ‘lady doritos,’ it would be X, or a close relative to X (maybe W or Y). projecting him to be roughly half what his historical value will be seems dubious, if ‘they’ are, using this same ‘math’ to project such a huuuge bump up for tulo. the data in use simply seems to conflict.

    • I was thinking that way at first, too. But it’s pretty hard to have a one-size-fits-all formula, I think. And really, like I say in the post, the vast majority of the Jays’ individual projections pass the smell test, so it’s actually doing a pretty good job if we adjust our expectations a bit.