Have you seen one of these guys?

Q: How can I tell the difference betweeen 'Spreadsheet Baseball' and 'Fun with Google Spreadsheets'?

A: The first is posted (usually as a "Story") by one of the site moderators. The second is posted either as a FanPost or FanShot by some creep who lives on the wrong side of the 49th parallel. The first is well-written and clear. The second attempts to be colloquial, but the maddening typos and run-on sentences make it difficult, if not impossible to read. The first, when dealing with player projections for the upcoming baseball season, utilizes  Baseball Prospectus' rightly revered PECOTA projection program. The second uses whatever free projections happen to be lying around the internet -- and who knows where they've been. The first comments on said projections briefly and with informed insight. The second complicates things with (most likely poorly applied) mathematical formula and combinations of metrics and offers inane commentary cloaked in an Emperor's wardrobe of objectivity.

Most importantly in the current context, the first is rare and thus valuble, like gold. The second comes unbidden, unwanted, but yet unstoppable, like a sinus infection. Which one are you more likely to receive this winter?

In other words, only NHZ can save you now...

Here (finally) is the pitching companion to my earlier piece discussing the Bill James Projections and the 2009 Royals offense. I held off because I knew that this probably wouldn't be as good, and also I was a bit uneasy with how to convert ERA/FIP, etc. into wins/runs above replacement. Thanks to Sky Kalkman of Beyond the Box Score fame, I now have some simple formulae for doing so, and so away we go...

[Yeah, it was a long, rambling introduction. At least I use the "intro paragraph" function so that, just in case the post makes the front page, it doesn't push half-a-week's worth of posts off.]

As with the piece on James' hitter projections, I want to remind everyone that I am not endorsing or criticizing the James projections either as a whole or in part. I simply thought I'd post a bit about them and try to translate them into terms that make it easier to see whether projections have the Royals getting better or worse, overall. The hitting projections, as I surmised, have the Royals offense improving by 5-6 wins over 2008 in 2009 (although I did note that the James offensive projections are notoriously optimistic). So I'm going to translate the James projections into runs above replacement so that we can get a better sense of what some people are projecting the Royals pitching to do next year beyond simply "hey, it looks like that guys ERA might suck next year" and stuff like that.

Boring Methodology Stuff

(If you find this stuff intolerable, skip down to "Starter Projections")

There are plenty of places to read about replacement level and the debates about whether it is useful and what it should be on the internet, so I won't get into it here.  I will simple say that the most basic and intuitive way to think about a player's value above (or below!) replacement level, whether in retrospect or prospect, and whatever stat you are using, is as follows:

rate above replacement times playing time = value above replacement level.

Pretty obvious, no? Then for projections, the idea is

projected rate above replacement times projected playing time = value above replacement

Of course, while the formula is simple, the assumed knowledge is difficult: What is the projection? How much playing time will the player get? What is replacement level?

Well, the first question is easy for the purposes of this post: we're just going to take the James projections and see what they say. They also project playing time, so that's taken care of (although that's even iffier than the ERA/FIP projections). I will also prorate the suggested rate over replacement over an equal number of innings for all starters and relievers for better comparison, and to allow people to do their own simple calculations based on personal playing time estimates.

So... what is the replacement level for pitchers? Since their playing time works out different than hitters, we can't just prorate say, 2.25 wins (about 23 runs) below average over a year with respect to plate appearances. Rather, we'll use a percentage of rate stat (ERA, FIP, tRA, etc.) as replacement level. I will follow the suggestion that for starters, replacement level is 128% of league average. For relievers, replacement level is 107% of league average. That's a partial explanation for using the following formulae for determining a pitchers runs above replacement with ERA as the preferred rate stat:

Starters: ((lgavgERA*1.28)-playerERA)*(IP/9.0)

Relievers: ((lgavgERA*1.07)-playerERA)*(IP/9.0)

These are pretty simple formulas. There are more complex ones based on pythagorean winning percentage and all that, but frankly, this was quicker and more easily understood.

One more thing before I (finally) get to the projections. Tables are hard to fit into margins for fanposts without really screwing things up. For the full tables with all sorts of boring stuff, check out the detailed Google Spreadsheet I made for this. Here, I've only filled in just what I could.

Here are the some explanations for the headings that appear in this post:

ERA, IP, HR, BB, SO, and FIP are all taken from the Bill James projections posted at Fangraphs.

ERArar is ERA-based runs above replacement, as detailed above. For starters, the formula is ((lgavgERA*1.28)-playerERA)*(IP/9.0), for relievers, it is ((lgavgERA*1.07)-playerERA)*(IP/9.0)

FIPrar is FIP-based runs above replacement. Since (in theory) ERA is usually supposed to regress to FIP since FIP is a better representation of "luck-free" peformance, I've scaled it against league ERA replacement level. If that's messed up, shoot me. For starters, the formula is ((lgavgERA*1.28)-playerFIP)*(IP/9.0), for relievers, it is ((lgavgERA*1.07)-playerFIP)*(IP/9.0).

ERArar+/- and FIPrar+/- measure how much the James projects have the player improving in runs from 2008 to the projected 2009 .

FIPrar200 and FIPrar70 prorate each pitcher's FIP-performance over "full-time" performance (200 innings for a starter, 70 for a reliever) for comparison purposes.

So, without any further ado, here are the projections. And please, once again, keep in mind, that I am not endorsing them one way or the other. They are not mine. Don't shoot the messenger (oh, it's not that bad).

Starter Projections

[Detailed Google Spreadsheet]

Comments:

Reliever Projections

[Detailed Google Spreadsheet]

[NB: I have not included leverage index in figuring out what the relievers are worth here. That is very important because, obviously, Joakim Soria should not be (nor do I think the projection systems intend him to be) projected to be about as valuable as Mike Jacobs in a good year. However, I'm not satisfied with my knowledge of how to incorporate leverage index for relievers in general, especially in projections. Leverage says less about the relievers specific pitching abillity than the value of the innings and situations in which he has pitched. Clearly, the reliever has little or not control over this. I could, of course, just calculate it for the Proven Closer(TM), since we have a pretty good idea of how he will be used, but then that just exacerbates the mythical difference between the "ability" of the closer and the other relievers. If I do this for other projection systems, I might include leverage then. Just letting people know what is up.]

Comments:

This seems like a good time to remind everyone that this is just my analysis and rough "translation" into run values of the Bill James projections, not my own personal projections or endorsement of said projections. Yeah... they didn't even bother to project  Gobble, Pumpkinhead, or Rosa (how can they neglect ears like that?). And who knows what roles (if any) those guys will play on the Royals next year. In general, keep in mind that as unpredictable as pitchers in general are to project, relievers, due to their varying usage patterns (and the small sample size relative to "true talent" inherent in said usage) are doubly difficult to project.

Soria is so good it isn't even all that interesting to discuss him at length. His ERA was probably a bit luck-derived this year, but we all know how "good" ERA is as a stand-alone measure of pitching ability. The James' projections recognize this by showing an improvement in his FIP. No worries here. Good thing the rotation is totally fine...

As for the rest of the bullpen... well, I would have included Nunez and RamRodRam
for the sake of once of showing the hilarious MLB mugshot of Ramon Ramirez's
perfectly round head, or, as Royals Review once put it, "he looks like a little kid who is supposed to be in bed peeking around the corner." It really hurts to see that the official MLB mugshot already has him as a digitized Rex Sock. Anyone got the Royals version? Seriously, there isn't much to say, other than the projections show what most of us (well, at least me) thought -- while Ramirez probably pitched a bit or a lot over his hear this year, he's still considerably better than anyone the Royals have left in the 'pen other than the guy who should be in the rotation. Mahay's great first half seems to have made people forget how badly he bombed in the second half. He projects as above replacement level here, but only slightly better than Nunez at best. Maybe adequate as a setup guy, but he's not in Ramirez's class anymore, according to the projections. The wildcards are: (a) Tejeda, given that his projection waffles between him starting and relieving (and who some think might have the stuff to be a very good reliever); (b) Rosa, who isn't projected at all (although I'll leave it to you prospect hounds out there to debate whether he should/can start or not); and (c) Dayton Moore's ability to find another Ramirez on the cheap, especially without another Greinke waiting around like in 2007 to bolster his bullpen-building reputation. Whatever one thinks of the projections, I think it's safe to say that, at the moment, the 2009 bullpen as a whole looks far inferior to the 2008 edition.

Concluding Thoughts

Given the difficulties in projecting pitching (especially relievers, and especially since the projections were done before Nunez and Ramirez were traded away), I'm a bit leery over setting out what they predict as far as overall improvement or decline for the Royals pitchering staff. Remember that 10.5 runs = one win. Excluding Tejeda from starting, by ERA the James projections have the starters improving by about 7.4 runs, or about 0.7 wins. By FIP, however, they are projected to decline by about 39 runs, or 3.7 wins.

The relievers get more complicated because I didn't use leverage, nor do I have projections for the bodies brought in to replace Nunez and Tejada. But I will give my my extremely rough (leverage-free) estimate of what the James projections say. By unleveraged ERArar, the current Royals bullpen is projected to be about 27 runs worse than last season's, or 2.6 wins. By unleveraged FIPrar, the bullpen is projected to be about 13.3 runs,  or about 1.3 wins worse.

Recall that my back-of-the-envelope calculations had the always-hitter-positive James projections seeing the Royals offense improving by about 5.5 wins. Well, we haven't taken defense into account at all, which could improve by a bit with Crisp moving into center and DDJ into left, or a lot if the Royals figure out that Teahen is a superior player (salaries aside) to Guillen with defense included. Anyway, with all the qualifications re: leverage, defense, and as-of-yet unacquired or unprojected relievers noted, using ERArar one might say that the James projections see the Royals as improving by 5.5 + 0.7 - 2.6 = 3.6 wins. Using FIPrar, we get 5.5 - 3.7 - 1.3 = 0.5 wins. Remember to take this with a shaker full of salt -- and that these aren't my predictions.

I hope you've found these posts at least somewhat interesting. It was a nice chance for me to pretend to do real analysis while simply adding up numbers and cracking my usual assortment of dumb jokes.  One last thing -- while PECOTA is NHZ's territory, would  people be interested in me doing this when other projections systems (like CHONE and ZiPS) come out? I don't mind wasting my own time writing stuff like this, but only if other people don't mind wasting theirs reading it.