USATSI_9227030_168381442_lowres

How the Blue Jays Can Win 100 Games

It’s inevitable that, as April approaches, various pundits will weigh in with their predictions for the upcoming season. Such will soon be the case at BP Toronto. However, having already heard many people venture a guess, as well as having taken note of the results of various projections, I find myself with an outlying opinion. My prediction for your Toronto Blue Jays during the 2016 regular season lies 10-15 wins north of what the majority of other observers have discussed. When I ventured that prediction on Twitter, my esteemed colleague Ewan Ross was skeptical.

Yes, with that rotation. Allow me to elaborate upon why I’m so confident.

The Caveats

Before I address the focus of his question, though, let’s establish some common ground. I think it’s fair to assert that no team in the AL has an offense in the same conversation as the Blue Jays. In fact, even in all of baseball, only the Cubs come close. Moreover, it is at least arguable that the Blue Jays defense is on a par with the best in baseball. Six of the eight fielders (pitchers not included) taking the field on most nights are in the conversation for the best at what they do, and in a seventh position (left field), the defense will be greatly upgraded over the 2015 aggregate. Not only is defense not going to be a drag on the claims I will make about the pitching staff, but instead it should install confidence that the staff will achieve even better results than those which might be otherwise be projected.

Another note for clarity – when we say “staff” we’re really speaking of the rotation here. No one questions the abilities of Storen/Osuna/Cecil, and predicting the outcome of the middle relief squad of any team is a mug’s game. Therefore the accuracy of my bold prediction comes down to how the starting staff will fare, more than any other single factor, save for health. Even then, with health being a relatively equal risk factor for all teams, we can do little more than acknowledge that any team plagued by injuries will under-perform expectations.

Finally, a disclaimer. While discussing my concept for this article, Joshua Howsam pointed out that Baseball Prospectus’ own Rain Watt discussed this same subject (marginal upside of a starting rotation collectively) almost a month ago. I wasn’t aware of it and purposely didn’t read it until I’d roughed out my draft for this post. I’ve attempted, in refining this one, to avoid plowing too much of the same ground that Watt did, and instead go further into the specifics of the Blue Jays’ particulars than he did. I’ll touch back on his article at the end of mine.

The Methodology

For the purpose of this analysis, I’ll be using BP’s WARP statistic, both as it has been accumulated to date, and as it is projected for 2016. In doing so I want to attempt to demonstrate why I’m so optimistic that projection systems are struggling to codify the staff’s potential. For contextual purposes, I need to point out that PECOTA projections are noticeably weighted towards the mean. To illustrate, 31 AL pitchers accumulate 2 WARP or better in 2015, yet only 16 are projected by PECOTA to do so in 2016. The total will surely be more than that, the system just isn’t designed to project that many.

Another example: Chris Sale has the highest pitcher WARP projection for the upcoming season at 4.2, yet Dallas Keuchel reached 6.2 last year and five others exceeded 4.2. The point being that my commentary is not meant to suggest figures in a vacuum, but rather to push back, where appropriate, against that regression since the regression is not always logical when one gets beyond the algorithms. It’s also necessary to add this disclaimer: I don’t remotely understand how the nitty gritty of PECOTA works. I’m simply using these figures for illustrative purposes, to support my prediction with something more concrete than admittedly biased opinion. With that said, here we go.

The Analysis

First data set: the total projected WARP for AL rotations in 2016 ranges from Cleveland’s 11.2 to Kansas City’s 4.3. In the AL East, the Red Sox shockingly come in second with 10.2, the Yankees are 4th at 9.5, the Rays with 8.6 are 7th, and the Blue Jays are all the way down at 12th with a mere 5.3. I take issue with that, as you will see, but that’s the cold number. So the gap between the Blue Jays and the best total in the AL East (assuming for a moment no one else moves off their projection) is 5 WARP. Remember that number.

Second data set: Here are the individual numbers for each potential Jays starter (note that the PECOTA managers thought Hutch would be the No. 5)

Marcus Stroman – 1.9

RA Dickey – 0.4

Marco Estrada – 0.6

JA Happ – 1.2

Drew Hutchison – 0.9

Gavin Floyd – 0.2

Aaron Sanchez 0.1

Laying aside the reality that he’s not the ‘5th starter’  hands up from everyone who would have been confident Hutch is going to accumulate more WARP than R.A. Dickey or Marco Estrada. No one? So, right away it’s sensible to be suspicious.  The WARP projections, which appear on the depth charts are the 50th percentile figures for any given player. On their player cards you can find a range of potential outcomes, ranging from the 90th to the 10th, with the idea being that the closer you get to the 50th, the more likely the outcome. With that in mind, let’s look at them individually.

Marcus Stroman, at 1.9, jumped out at me immediately. In 24 career major league starts, and an irrelevant dash of relief pitching, Stroman has accumulated 3.1 career WARP. Just taking his past work as evidence, pro-rated accordingly, gives you a figure slightly higher than 4 WARP, which is almost twice what the 50th percentile figure is, and in fact it’s considered by the system to be his 90th percentile outcome. Admittedly, and I cannot say this emphatically enough, I do NOT know all the factors which are weighted in PECOTA, but I’m forced to speculate that it doesn’t really know what to make of the lost time last year. I’m highly confident that Stroman will at least replicate his previous success, and I see no compelling argument that he’d significantly regress.

R.A. Dickey is, perhaps, the easiest sell here, since it is accepted wisdom that he breaks all systems. While the PECOTA projection gives him a mere 0.4 (a figure he’s not dipped to since 2008, before  he began throwing the knuckleball). Over three seasons in Toronto he’s been measured at 2.9, 2.5, and 2.8 and there’s no real reason to presume he’ll fall from that neighborhood. His 90th percentile is 2.4, so let’s go with that. As with Stroman, I don’t see the argument that he’d collapse to the extent necessary to post 1/7 the value he delivered in 2015.

Marco Estrada is of course not able to point the same consistency that Dickey can. However his tiny 0.3 projection is, in my view, just plain silly. Let’s throw out the 3.5 last year as, potentially, a career peak. Still, over 2 seasons of full time starting in Milwaukee, his WARP totals were 2.1 and 1.8. It’s true he employed a somewhat different pitching style then (more on that in a sec) but I’d argue that those figures still represent a reasonable example of his value, even as he regresses. The system, though, pegs that previous level as his 80th percentile, which is to say 1.9, but I’m going to go with the 90th again (yes, I know how this looks but I’m at least defending the claim, right?) which happens to fall about halfway between those previous results, and what he did in 2015. That figure is 2.6, which I repeat still represents a regression from 2015. But I think the choice is justifiable, and here’s why:

Estrada added a cutter last year that he apparently found lying around in the in the bullpen before his third start of the year in Houston, and his use of that pitch correlates with a noticeable uptick in his results. From that point through the end of the season his ERA was 3.08 but, more notably, by mid-June he had refined his use of the pitch and in his final 20 starts he posted a 2.62 ERA and limited opposing hitters to a stunning .558 OPS. For comparison, David Price limited opposing hitters to a .555 OPS during his 11 regular season starts for the Jays. Estrada followed that with a 2.33 ERA in three playoff starts. This may be the boldest claim I’ll make in this post, but I stand by it.

Next we come to J.A. Happ, who’s projection is 1.2 and I find that reasonable. I won’t tamper with it. In a post I wrote for Blue Jays Plus, I detailed my reasons for believing the rotation was at least equal to last year’s results and in doing so I pointed out that in the year of start between the break in 2014, and the break in 2015, Happ was essentially a league average or slightly above SP, with an ERA just under 4, and this was before he landed in Pittsburgh. However, excepting his run in Pittsburgh over the last 2 months of 2015, Happ has not finished a season higher than 0.9. WARP since 2009. I simply have no statistical support to predict that he’ll outperform what PECOTA assigned to him.

That leaves the fifth starter slot, which the Depth Chart lists as Drew Hutchison  The system calculated a 0.9 for him over the 23 starts that was assumed for him at the time. In the midst of uncertainty over who fills this slot, what we did know from pretty much the opening of camp was that it wouldn’t be Hutchison. You may know that I counted myself firmly in the “Aaron Sanchez of course!” Party when it comes to who should fill this role, so I’m obviously very pleased with the team’s choice, and I already laid out earlier in the week one possibility model for how he could accumulate 25-30 starts.

I’ll further state without having the data to support it, that should he get that many starts he’ll absolutely better this figure, easily. I wouldn’t be shocked if he tripled it  over a full season. However, there’s a LOT that we simply do not know in terms of how he’ll be handled, and data that can only be accumulated by watching him do it. As a starter this is a sort of rookie year for him, particularly given his physical development and the ongoing refinement in his technique.  So, on balance, I’ll leave it alone. Well, except to adjust to the closest percentile, so 1.3 is the figure here.

PECOTA also projects three other pitchers regarded as spot starters and, regardless of who they actually end up being, there’s another 0.3 in value there so we have a total of 11.8, which is much more respectable than 5.3, right?

“But,” you’ll say (I know you’re saying it right now) “You’re not considering what other teams are doing – you can’t just evaluate one team in a vacuum.” That’s quite true, which is why I applied the same rosy optimism to every other team in the AL East (darn hard to do for the Orioles, let me tell you) as well as the four other AL teams that fared best in the PECOTA projections – Cleveland, Chicago, Houston, and Seattle.

This is somewhat arbitrary, admittedly, but it seemed right and proper to consider every division rival, and the others were selected based on the likelihood that they would still come out with a higher total than Toronto when the study was complete. I confess that I’m trying to skate by with 8 teams here instead of all 14, but I don’t think it matter to defending my contention whether or not the rotation of, say, Texas, is being underrated by PECOTA based predictions.

To reference Watt’s article, for comparison purposes, he dialed up every rotation to the 90th percentile and in so doing demonstrated that the Blue Jays benefited more from such an escalation than all but 3 AL teams, none of whom benefit from my methodology as much as the Jays did. His findings correlates to my argument but, as I’ll explain later, is not exactly the same argument.

I’ve spent some part of every day the last 2 week going over these rotations and trying to look at them as a fan of that team might, without going past what I could rationally support with the equivalent data to what I used for the Blue Jays. My results increased the total of each team, and no player was downgraded.  Here are a few examples.

PECOTA has David Price at 3.8, yet he was 5.6 last year and preceded that with a three year average of 4.3 – so I split that (to the closest percentile) and gave him 5.1 (his 80th percentile).

Chris Archer was assigned 2.9, down from his 4.9 in 2015 – I went with his 80th percentile of 4.2.

Dallas Kuechel was given a messily 2.8 after his 6.2 in 2015 and the 4.9 recent average he brought into that season. I went with his 90th which is 4.8 as a more obvious conclusion.

And finally, for one non-ace example, Boston’s Eduardo Rodriguez came off of a season in which he’d put up 2.3 WARP in not quite 2/3 of a season, which pro-rates to 3.5 yet his projection is 1.1. Anything less than 2 (his 80th percentile) is underestimating him, in my opinion.

These are certainly random examples, but they are only offered as an illustration of how I applied the methodology to other teams. It might be said that my choices are arbitrary, and to the extent that when one interjects human opinion into the cold mathematics offered by PECOTA, arbitrariness is an ever present demon. However it is a, if you will, consistent form of arbitrariness. I went with the 80th percentile for Price and Archer for the same reason that  I did not try to max out the Blue Jays “fifth starter” figure based on a guess that it would be Sanchez and that he would excel. It’s not my intent to just guess wildly, but to adjust the figure based on what I can support safely, conservatively, from their past performance. That just so happens to be the 80th percentile for Price, but the 90th for Kuechel. The method has some inherent element of arbitrary assumptions, but it’s not being employed in an arbitrary fashion.

The relevant conclusion for the purpose of discussing the Blue Jays is this (you don’t REALLY want me to go into detail of what I concluded for 40 different SP that play for other teams, right?): Instead of ranking 12th among AL teams, the Jays rank 6th. Instead of being 4th in the AL East, 5 games back of the Red Sox, the division looks like this:

Boston – 13.1

New York –  12.1

Toronto – 11.8

Tampa Bay – 11.6

Baltimore – 7.3

That’s a mere 1.3 WARP swing between the Jays and the top of the division, less than 2 from the best team in the AL.

 

In Conclusion

Here’s where I need to circle back to Rian Watt’s excellent “Life at the Margins” (which, if you haven’t read, you should definitely do so) one more time. Watt made a good point, and drew valid numerical conclusion about what happens if every team reached it’s 90th percentile level for every starting pitcher on the staff. But where my point differs, particularly concerning the Blue Jays relative to the rest of the league, is that I’m NOT assuming every pitcher, even on the Blue Jays, will do that. My argument hinges on the claim that this particular set of pitchers has a better argument for greatly exceeding expectations relative to their direct competition. Yes, the margin between the Blue Jays and the Red Sox only closes a little (4.3 to 3.6) if you run both staffs to the 90th as Watt did but that improvement does mitigate for my premise. But I hope I’ve illustrated that it’s more likely, based on the track record, that you can get that result out of Stroman, Dickey, and Estrada  than out of Buchholtz, Porcello, and Kelly.

I’ll be the first to admit that for any team, things can go horribly wrong. I’ll also readily concede that there may be a complexity here I’m not seeing, but the simply stated conclusion which I support my prediction with is this – the Blue Jays rotation, when compared to their AL counterparts, is at least league average and when it’s placed in the context of the team’s offense and defense is much more than adequate to  the task of exceeding 95 wins.

I am, of course, giddy with anticipation.

Lead Photo: Kim Klement-USA TODAY Sports

Related Articles

3 comments on “How the Blue Jays Can Win 100 Games”

Arjonn

Do the WARP projections factor in that the quality of defence varies from team to team? If not, it would seem to make intuitive sense that pitchers on teams that defend well would be at least a bit more likely to surpass their projections.

Jacob

WARP, like all wins above replacement stats, is designed to be team neutral. That means it tries to say “if we dropped this player on a random team how good would he be?” The Jays pitchers will be helped out in standings results and team dependent stats like ERA by our superior D, but their WARP should remain constant based on things they control, like K rate.

Arjonn

Thanks for confirming what I guessed would be the case. This looks like yet another example of the whole not being the sum of the parts.

Leave a comment

Use your Baseball Prospectus username