Reasonable Expectations: 2012 Royals

Scott, you can be a real piece of work sometimes.

Kcdc has put together a heuristic analysis. This is a perfectly acceptable and normal way to engage in research. In this case he’s started from the reasonable simplifying assumption of identifying the starting players and attempting to predict what their performance next year will be. Future iterations then add complexity to the estimate until such time as the tradeoff between effort and additional information becomes negative.

Most of your “concerns”, by the way, are not issues that affect the point estimate projection, by the way, but rather are issues that widen the error bands on the projection. For example, you complain that it doesn’t take into account differences in playing time…but that’s not really an issue because baseball is a zero sum game and any uncertainty in the projection from this factor would affect all other teams. It increases uncertainty but there is no reason to believe it systematically understates or overstates the point estimate unless we have reason to believe that the Royals bench is significantly worse/better than other teams in the league.

You’d have to make reasonable playing time estimates for each player (which would decrease the overall impact of each of them), and then add in the negative WAR from various bench players and fill-ins.

I don’t know why we’re assuming the bench players would be below replacement level. I’d kind of hope they’d be above replacement level. As for playing time decreasing the impact of the players I’ve listed, I calculated batting runs on a 650 PA basis. That’s a pretty normal number for a regular, so you wouldn’t have to bump down the regulars too much. If anything, it might help because the guys who won’t make it to 650 PA’s are some of the poorer hitters, and the subs (Navarro, Cain) probably aren’t much worse.

And then there’s the problem of using ZiPS ros projection as a projection for 2012 (this issue discussed in another thread).

There were a lot of weaknesses in this method that I didn’t fully discuss. I don’t mean for this to be a scientific paper. But if you think any of the actual numbers are far off what you expect, please take issue with those numbers. I think they all look pretty reasonable. Does a .327 team wOBA sound far off to you?

And then there’s the problem of you estimating/projecting earned runs, rather than runs (and of course SIERA does a better job at the former than the latter).

This is about the last of my concerns. Without additions, I have the Royals giving up about 9 more earned runs than the average team. About 10% of runs are unearned, so if you want to bump that 9 earned runs up to 10 runs, that’s fine with me.

But I think the final product is more like a swag than a reliable product bourne of meaningful numbers

Okay. Sort of. Every projection deals with degrees of accuracy. This one is pretty rough. But that doesn’t make it meaningless. I think it errs a little to the positive side, so we should probably bump our expectations down a bit from the result. But the result is 84.2 wins. And you’ve said that you see a 75 win team heading into next year. Maybe my projection is 30 runs too optimistic, but 90 runs?

The purpose of the post was to get some reasonable numbers up to open debate on where the team is. If you still see a 75 win team, you could give a rough break down of where those 90 runs might go.

I didn’t include defense or baserunning in the projection which hurts Escobar’s total while probably helping Melky and Gia’s totals. I think it balances out overall because the Royals have a pretty average UZR team as a whole.

The projection thinks you’re just about dead on with your 24 WAR. The only projection that looks a little funny is Brayan Pena’s 2.8 WAR, but he’s a placeholder for the 2012 catchers as a whole, which you have at 2 WAR total.

What is it, exactly? Or are you using The Force?

Pena’s numbers in this projection are basically imaginary. I doubt he’ll even be on the team next year, but I don’t know what catchers will be. The best guess might be Sal Perez, but he could just as easily chill in AAA until next September, and even if I wanted to call him the every day catcher, projecting his numbers is a total crapshoot. So I was left with a season of catcher PA’s and no rate stats to apply to them. I ended up giving assigning 600 catcher PA’s to a .301 wOBA, and called it Brayan Pena. The .301 wOBA might be a touch optimisitic, but I don’t think it’s orders of magnitude off.

Giavotella got 550 PA’s of .319 wOBA. That would make him an above average bat with a second base positional adjustment. Maybe that level of offense is a bit optimistic. UZR not being included is likely in his favor as well. But for the team calculations, that’s offset somewhat by Escobar’s WAR being lower than reality.

Moustakas was assigned 550 PA’s of .311 wOBA. That sounds okay to me. If he doesn’t turn it around, they’ll play Navarro or someone else instead.

So, the defense will randomly be average just because you say so.

No, it’s reasonable to project the defense as average because it’s roughly average this year. It’s actually a little better than average this year, but that’s with some outfield assist luck.

All of our pitchers will have better luck next year while none of them will have bad luck, and all of our hitters will either take a step forward or reach their career norms at least. Also, all our young guys will develop in a positive fashion to randomly add three wins.

SIERA neutralizes pitching luck. It doesn’t give good luck. Some pitchers will be lucky next year, and others will be unlucky. It just happened this year that Davies, Duffy, Hochevar, and Paulino ran into some bad BABIP and LOB% luck. If they’d run into good luck, as Chen has to a degree, SIERA would have bumped them down.

Young hitters tend to get better. That’s how development works. It won’t work evenly for every player of course. Hosmer might take a big step forward and Moose might get demoted to AAA. Or it could work the other way where Moose takes a big step forward while Hosmer struggles. Adding a few points of wOBA for a year’s development for players in their low 20’s is reasonable. If young players didn’t improve at all, that would be unexpected.

That’s a hell of a lot of things that will have to go exactly right in order to reach your projected win total.

If you look at the WAR totals, I don’t you’ll find them that unreasonable. I wouldn’t say any of them look like best-case scenarios by any means.

In your projection, every single hitter stays healthy and has a decent or better season.

It’s true that everyone stays healthy. This is an unrealistic assumption. It’s also true that every hitter has a decent season. But you’ll also notice that no hitter has a significantly better than expected season. This is the nature of the projections—every individual projection is about in the middle of what you think they could do, and then half the hitters to better and half do worse.

Not to mention we somehow replace our worst starter with a pretty good FA just because.

I didn’t say we’ll do that. I said if we keep the same roster, this admittedly optimistic projection puts us at about 85 wins. Then I said replacing Chen with a good FA would bump that up 2 more wins. I’m not saying that will happen. I’m just noting the most obvious way to improve the team and what effect that would have.

You are right, there are many flaws and they are obvious..so I’m not sure why you are getting upset when they are pointed out.

I haven’t gotten upset at all with people pointing out flaws. You and Scott are right that the playing time assumptions aren’t realistic. Nobody has made much of a fuss over the facts that the pitching stats aren’t regressed to the mean and no development/aging curve is applied. And I consider those issues to be the biggest flaws.

What has annoyed me was the sense that because the model makes some simplifying assumptions, you have to throw out the results entirely. Scott seems to think that without accurately projecting playing time for all 40+ players every team might use next year, you can’t have any clue whatsoever about how good teams are. I strongly disagree with that notion. I think you should take this projection for what it is—reasonable rate stats applied to ballpark playing time allocations while assuming neutral defense+baserunning and no injuries.

But it’s going to be small.

Let’s say you have a really good player who will accumulate 5 WAR over the course of the year but will miss a large portion of the season—say 20 percent. Even if his replacement is pretty poor—let’s say a 1 WAR/year player—the production for the position will be 4.2 WAR per year. For a guy who is league average 2 WAR type, the cost of replacing 20 percent of his production is barely noticable—1.8 WAR rather than 2.0 WAR.

That’s why its frustrating arguing with Scott—he’s completely clueless about the magnitude of the problems raised by his objections.

I also don’t see a reason to believe that the Royals bench will be below average next year.

Not including PA’s and IP for the worse players probably does skew the win total toward the positive side, but if you’re comparing teams, assuming average bench production on average bench PA’s should have minimal impact. What team is known for its great bench that gives them a 30 run advantage over other teams?