keyboard_arrow_uptop

As far as February traditions go, it’s moving right up there with rural Pennsylvanians’ abuse of a groundhog and LL Cool J mailing in his Grammy hosting duties. Every year as teams prepare to load up the trucks for spring training, PECOTA gives the Orioles a nice sendoff with a last-place prediction.

It doesn’t matter if they won the year before. And in the small sample size of the last four years, all of which have featured a last-place projection in Baltimore, it doesn’t seem to preclude their winning the division, as they turned last-place outlooks into first-place finishes in both 2012 and 2014.

If PECOTA held a grudge for some bad customer service experience at Camden Yards, it would be easy to explain. Even absent a frightening level of self-awareness from the computer, if it were a projection that projected at the team level, there might be a decent explanation as some trend might repeat that makes them routinely underestimated. However, that’s not how PECOTA works. It treats the team projection as the sum of individuals, all of whom came from different places, are at different points in their careers, and have been part of the usual significant roster turnover in the past four years and beyond.

Yet, here they are again. Last season, they were projected for 75 wins in the early release. That was of course before they signed Nelson Cruz in late February, and PECOTA actually liked Nelson Cruz more than many of the statistical and eyeballed projections. So as a full roster, they were a little closer to .500. They went 96-66, some 15 or 20 games better than their range of projections, and swept a first-round series before getting swept in the ALCS.

Now they’re pegged for last place again, albeit in a much tighter bunch, listed at 78-84.

Assuming it really is nothing systemic to the Orioles, which would be contrary to how the projections are put together, it’s better to look at these players.

How are they going to go all the way from first place to last place? How much of that projected 18-win drop to an expected 78-84 is departures and/or substandard additions and how much is decline or reversion of the guys already there?

Getting to 18

The first five wins are easy. That’s the luck that the Orioles had last year that you would expect to be close to independent year-to-year. A 96-66 team, they had a Pythagorean record based on runs scored and allowed of 94-68 and a third-order record based on the underlying stats that go into run scoring and prevention of 91-71. We had this discussion in 2013 after the O’s had an incredible record in one-run games in 2012 and somehow that skill didn’t carry over.

No shame in that. First place teams generally do outperform. So there’s 13 to go if we’re going to explain away the whole thing.

The division is an obvious spot to look next. Last year broke a six-year streak of the American League East being at least 30 games above .500 in aggregate. It was still a good division—12 games over .500 in the superior league. The division outside of Baltimore this year should be stronger, carried pretty much totally by Boston, where a 15-game improvement distributed evenly over their unbalanced schedule would include two against each of the division opponents.

And then there is the team itself.

Below is a chart of position-by-position WARP on what actualized as a playoff team vs. the PECOTA projections for that position’s WARP this year.

Position

2014 WARP

Top player*

2015 Proj. WARP

Top player**

C

1.9

Caleb Joseph

2.2

Matt Wieters

1B

3.1

Chris Davis

2.2

Chris Davis

2B

-0.4

Jonathan Schoop

0.8

Jonathan Schoop

3B

2.8

Manny Machado

3.9

Manny Machado

SS

2.1

J.J. Hardy

2.0

J.J. Hardy

LF

5.0

Nelson Cruz

1.9

Alejandro De Aza

CF

4.0

Adam Jones

3.5

Adam Jones

RF

2.7

Nick Markakis

0.6

Travis Snider

DH

3.7

Nelson Cruz

2.2

Steve Pearce

Position players

24.9

19.3

*By playing time
** By projected playing time

Some of the positions from 2014 include some others—for instance, neither Pearce nor David Lough was the primary player at any position, but they had a lot to do with the first base and left field totals, as Davis was actually bad and Cruz was actually a DH.

The sums, though, tell the right story that 5.6 of the wins are supposed to be lost on the position player side, which includes batting and fielding. All of that, and then some, are supposed to come from the outfield and DH positions where the team both overachieved last year and then lost Cruz and Markakis on top of that.

As for their pitching, it wasn’t great even when they were rolling the AL East on the way to the playoffs.

Their starters—Chris Tillman, Wei-Yin Chen, Bud Norris, Miguel Gonzalez, Ubaldo Jimenez, and Kevin Gausman combined for 161 starts—had a 5.8 WARP, while their relievers combined for 1.7.

This year, that 5.8 from the starters is projected to decline to 3.7 WARP, while the relievers are supposed to be almost identical within a tenth of a win.

The 2.2-win decline by the pitchers is expected to be attained in a very different way from the larger one on the offensive side. This isn’t about any roster turnover at all, as this is almost the same crew responsible for the division title. The same six starters are still supposed to get almost all of the starts, they’re just supposed to be a small bit worse.

So then the question is: Does this look reasonable?

The Orioles depth chart has the individual breakdowns, and off-hand, there’s nothing that screams out that PECOTA missed on. The departures hurt. Pearce is down, but he’s going to be down on Plexiglas alone. If anyone looks low, it may be Snider, who was very much a competent starter last year but suffers for PECOTA’s longer memory. If he’s truly improved, that’s a spot where PECOTA may have missed low again.

If you’re going to call the system out for undue hating on the pitching side, it’s probably going to be with Gausman. Tillman’s been a one-win pitcher two straight years at 200-plus innings apiece, so his projection for that neighborhood doesn’t look all that out of line. Gausman, however, you could make a better case for a breakout. He’s young, just entering his 24-year-old season, and for the two-thirds of the season he did pitch, he easily exceeded his projection. His strikeout rate was down from his major-league debut, and if he can return to that, he could be huge for them in the rotation.

On the whole, though, this projection doesn’t look like a travesty (yet).

From 96 wins, they’re projected to drop by 18:

5 to luck
2 to the division
8 to defections in the offense
2 to minor regressions in the pitching

Which makes 17.

The missing one must be with the bellhop.

You need to be logged in to comment. Login or Subscribe
fawcettb
2/12
I don't get BP & PECOTA. If you keep dressing up the village idiot and pretending it's a supermodel, it doesn't raise the probability that it will actually turn into a supermodel. It's still the village idiot.
redsoxin2004
2/12
What exactly is your issue with PECOTA?
zacharylevine
2/12
I'm lost. Am I the idiot or the model in this metaphor? Only asking because my mother has called me both.
therealn0d
2/12
How unlikely is it that the supermodel is also the village idiot? I don't think the two are exclusive.
NYYanks826
2/13
That was a very poor metaphor.
CrashD
2/12
I am a fan of the Orioles, PECOTA, and authors who can meaningfully cite Poe and Wolfram in the same article, but not of some of the assumptions in this article. Namely: 1 The whole is the sum of its parts. The Orioles have in recent years performed better than that. Blowing off their 2 or 5 wins above Pecota 1st or 3rd order projections because facts differ from theoretical projections works fine for a while, but folks, it's getting old. 2 Nelson Cruz is (was?) Superman. You have him contributing 5.0 WARP in left field in 2014 and 3.7 WARP as DH for a total of 8.7 WARP, when his WARP last year was actually 4.2 (and you project his WARP this year will be 2.5, which may explain why the Orioles let him walk). 3 Chris Davis was (will again be?) Superboy. You have him contributing 3.1 WARP at first base in 2014 when his WARP last year was only 0.7. This year you project him at 2.2 4 Delmon Young is a Nonperson. Listening to Duquette and Showalter recent comments, the primary DH slot is his to lose. With or without reference to these comments, you project his 2015 WARP to be 0.0. Somebody is wrong here (but possibly not you). To save us all some time here, even if Delmon is a Nonperson, just taking away the 4.5 and the 2.4 excess WARPs from 2014 Cruz and Davis respectively makes it harder to argue that the position players are 5 games worse, even on paper, in 2015. On the other hand, some of your analysis seems spot on, such as the projected improvement in the rest of the AL East. And, you may be more optimistic about the starting and relief pitching than I (or, by some of his recent comments, Duquette) might be. So, why am I writing this? Its because over the years I've been reading BP, it's seemed to me that Pythagoras and PECOTA have become deified, and in that process ossified. This in turn, it seems to me, has led to their occasional uncritical use, as in the examples cited above. These are great tools, but if they were the last word, and that easy to use, I wouldn't have to read BP any more. In the case of the Orioles, the last two GMs and the current manager may have had WARPs above 0.0, and their fielding may have contributed more to their record than current metrics quantify. I'm not saying this is what you missed; what I'm saying is, Don't Stop Looking. And, keep up the overall great work.
cburnell
2/12
Excellent points all. Nice work.
jtrussell10
2/12
The '2014 WARP' isnt the number at which the player, produced, its the number at which the *position* produced. Like Steve Pearce had 4.8 WARP, some of that in LF some of that as a DH, which is why the LF and DH dont add up to Nelson Cruz's WARP. Both Cruz and Pearce contributed, among others. The 'top player' is just the player who had the most appearances at that position, by all means just ignore that as its rather irrelevant.
therealn0d
2/12
I would hope that everyone realizes that it's just math. You can criticize the results, and therefore the math, but it's not personal. It's just math.
BrewersTT
2/13
The article does a good job of explaining why PECOTA doesn't like the team's 2015 chances. However, it only tangentially addresses why the projections have been so far off for several years in a row, and that is where there is the most opportunity to learn. I agree with CrashD above on this: the statement "Assuming it really is nothing systemic to the Orioles, which would be contrary to how the projections are put together," seems a little too pat. Maybe it is all just luck. It's possible to roll snake eyes four times in a row with fair dice. At what point do a string of consistent projection errors become "statistically" unlikely, leading us to reject the hypothesis of luck? I'm not interested in attacking PECOTA. I'm interesting in discovering any potential hidden truths behind this.
Schere
2/13
The Orioles didn't win the division in 2012.
zacharylevine
2/14
Sorry, postseason, not division. -Z
crowdog27
2/14
Nice article. I guess it just makes me want to see a strict comparison of those previous years' projections against what actually happened in the real world. Wouldn't that speak to the issue a little. Was it the pitching or the hitting that outperformed the projection.
tim270
2/14
The Orioles are excellent defensively. This is considered on the positional side, but not on the pitching side. I think this explains a lot of the discrepancy.