Wednesday, February 09, 2005

An Unexpected Screed and A Lengthy Redundancy

Chris at Capitol Punishment points us to an interesting exercise at the Replacement Level Yankees Weblog, where a fellow called SG (a Stargate fan, no doubt) simulated 100 seasons based on ZIPS, one of many competing player projection systems. Good news! We average out to 81-81, indicating that Phil Wood is the greatest genius who ever lived, the Miss Cleo of baseball, the Bizarro Will Carroll.

But don't get too carried away, or you'll wind up as blinkered and lazy as a Baseball Prospectus columnist after the PECOTA projections come out. Here's the thing about all these methods of predicting what a player's going to do: they're all almost always wrong. Somebody whose name I don't remember in an article I can't find pointed out that you can't project with any more than 70% accuracy (UPDATE: It was Ron Shandler - big ups to Basil the Inquirer). So it's fine to use these things to while away the long weeks until the season starts, but it's quite another thing to produce column after column filled with criticisms of GMs who make moves that your projection system says aren't going to work out.

Take, for example, this Baseball Prospectus filler, which takes issue with a prediction that Oakland is going to finish last in the AL West. I find that conclusion less than plausible myself, unless MLB switched out the Rangers for the Red Sox without my hearing about it, but author James Click (presumably of African bushman descent, given his ononmatopoeic name) rests his argument largely on some shit one of his coworkers made up. PECOTA says the A's rotation is going to be just fine and even presumes to tell us how many innings each pitcher will provide. Skeptical? Well, buckle up - Click provides a helpful link to a Baseball Prospectus article proving conclusively that Baseball Prospectus is the smartest. I haven't been so thoroughly convinced since Philip Morris told me that cigarettes cure gout. Anyway, it's all bullshit - that impossible-to-predict 30% can mean a hell of a lot, and it's unfortunate that BP expects us to pay for its rainy-day musings and what-ifs.

This wound up being much more of a screed than I intended it to be, but if you're going to go, you may as well go all the way: All player projection systems are sons of whores and can jump up my ass.

While BP gets itself through to April by pretending that Nate Silver can predict the future, I kill time by whining endlessly about how no one appreciates Livan Hernandez, official pitcher of Distinguished Senators (and congrats to to Washington Nationals MLB News, BTW, for snagging Jose Guillen. Jose Vidro can be yours for fifteen bucks). So here's my latest superfluous exercise: it's generally assumed that Bluegrass Brad Wilkerson is our best player. I maintain that B-Wilk is a sweet player and good guy, and I'm beyond thrilled to have him on my home team, but that he does not rank among the elite players in the game, while Livan does. Let's look at it this way: would Wilkerson be the best hitter on a playoff-caliber team, and would Hernandez be the best pitcher on such a club?

Eight teams made the playoffs last year, as has been customary since 1995: the Braves, Cardinals, Astros, and Dodgers in the NL; and the Yankees, Red Sox, Angels, and Twins in the AL. I'll be using VORP for this endeavor (glossary coming very soon - maybe tonight, but probably not), since that is the only way to capture the swankness of Livan. B-Rad Wilkerson put up a very impressive 48.2 VORP in 2004, putting him in the company of Sean Casey, Travis Hafner, and Ichiro. St. Louis, which won the more games than any other NL team, had three position players with superior VORPs (Pujols, 103; Edmonds, 88.9; Rolen, 73.7). Actually, those are much superior VORPs - the Cards frigging ruled. The Astros had three (sort of), as Lance Berkman checked in 83.7, Jeff Kent at 55.2, and Carlos Beltran provided 43.8 with Houston and an additional 30.7 in KC. The Braves and Dodgers had one guy each better than B-Wilk: J.D. Drew in Atlanta (78.7) and Adrian Beltre in LA (87.1).

Over in the American League, the Yanks sported a whopping five players with VORPs over 48.2: Sheffield (63.4), Rodriguez (62.3), Jeter (59.7), Matsui (57.5), and Posada (48.9). Boston had three: Ortiz (71.3), Ramirez (68.6), Damon (51). Vlad Guerrero (88.5) was Anaheim's only contribution. The highest-VORP position player on the Twins was Lew "Lou" Ford (44). Therefore, only one of the eight playoff teams didn't have a hitter superior to Brad Wilkerson.

Livan VORPed it up at 58.3 last year. No one on the Cards was close (Chris Carpenter was best at 41.6). Roger Clemens was slightly better for Houston (61.3). Atlanta and LA couldn't hang with Orlando's Rubenesque half-brother (40.3 and 49.7 were good enough to lead those teams, respectively). Things were a little better in the AL. The Yankees (Gordon, 39.8) and Angels (Escobar, 53.2) didn't have anyone better than Livan. Minnesota, however, had two: Johan Santana led everybody with 88.8, and Brad "Rad" Radke posted a 60.1. Curt Schilling of the Sox also exceeded Livan with 72.9. Livan Hernandez was a better pitcher than anyone on five of the eight playoff teams.

Here's what I'm getting at: Brad Wilkerson is not a legit #1 position player on a good team. This isn't to take anything away from the guy; he's very good and worthy of our respect. Livan, however, is absolutely good enough to be the ace of a great team. For some reason, this fact is lost on just about everybody.

22 comments:

Basil said...

***Somebody whose name I don't remember in an article I can't find pointed out that you can't project with any more than 70% accuracy.****

Attaboy, mate---Ron Shandler, marketing genius: http://www.fantasybaseballfriday.com/ajr84.shtml

I thought about commenting on it last week, if only to use the phrase "Marcel the Monkey."

Basil said...

BTW, I know I'm a smartass, but it is sort of ironic that you post this entertaining (and excellent) screed on BP playing with their ding-a-lings, then you cite to a measure (VORP) popularized by . . a BP guy. ;-)

Ryan said...

Thanks for the Shandler pointer, and HOW DARE YOU POINT OUT MY INCONSISTENCIES!? BP does some things better than others. Chris Kahrl and Steven Goldman: good. Will Carroll and Dayn Perry: bad. And hell, I'm still paying those bastards.

Basil said...

I like Goldman, too. Kahrl's fine, except for what I call Kahrlitis---the obsession over the 25th roster spot. I never much noticed Dayn Perry until you exposed his perversions.

And one inconsistency ain't nuthin'. Hell, my very first post mistook a two month old article as yesterday's news (literally, rather than figuratively). :-)

Chris Needham said...

Geez. Extra cranky today, are we? Going through Dayn-withdrawals? :)

I think you're right to a point on the predictions. They're a fun way of assessing how things look, on paper. I would point out though, that Diamond-Mind is traditionally one of the more accurate systems out there. And I think when you're doing something like simming it 1,000 times, some of the 30% even out.

Ryan said...

That's a very good description of Kahrl's odd obsession. I'll never know as much as she does about marginal ballplayers, and I'm not sure I'd want to.

Oh, I'm not cranky (well, not toward Basil, anyway). I'm just feeling my oats. I enjoy looking at PECOTA and ZIPS and all that - it's a nice diversion. But I really think BP relies too much on it for material, and some of their writers allow their tone to get excessively condemnatory when they're still talking about likelihoods rather than facts.

And if Dayn doesn't have something perverted on Monday, I'm taking credit for fixing him.

Chris Needham said...

I hope you don't mean 'fix' in a Bob Barkeresque way.

Yeah, the holier-than-thou attitude and blind obsession towards numbers is definitely obnoxious. (See any MGL post on Primer for a great example.)

But, there's not a paying audience for that happy medium, I guess. (See: Limbaugh, Rush; or Franken, Al)

Ryan said...

Also, Joe Sheehan hates every contract ever signed by anyone. "$75,000 for Babe Ruth? It's too much! GMs are all insane!"

Chris Needham said...

Nah. The Yankees signed him. He'd probably approve. ;)

Basil said...

Did you guys buy Prospectus during their "Young! Young! Young!" phase? (The corollary of the Proven Veteran(tm) dig.) They predicted the Twins to break out every year from about '98-01, just on the basis of going young. Well, they were right eventually.

Chris Needham said...

Yeah... I think that was right after they went through their "Screw Defense" phase. Remember when the A's had Jaha, Giambi, Stairs and Grieve on the field at the same time?

The classic for them though, is the 1998 Yankees essay. They predicted that that was the year that old age would finally catch up to them. They were due to fall off a cliff.

Unfortunately for BP, when they fell, they landed on a bazillion wins.

Josh said...

Chris Kahrl is a she?

Chris Needham said...
This comment has been removed by a blog administrator.
Chris Needham said...

She wasn't until recently.

Note the absence of sarcasm.

...Not that there's anything wrong with that...

Ryan said...

I missed the Twins' breakout predictions, but I guess they're still doing that with San Diego. Keep saying it and it'll come true.

Goldman, BTW, is still in "Screw Defense" mode. I swear he suggested playing Edgar Martinez in center. Well, not really, but he's lagging behind his compatriots in that area. Dude parties like it's 2000.

Kahrl seems pretty up-front about all that. I know there was an article about her - I wanna say in the Washington Blade - and she discussed it on BTF. Good for her.

Basil said...

I think that BTF did link to an article from the _Blade_ about it; that's how I found out, at least. I'm surprised Kahrl talked about it (or anything) on the Primer site. I thought BPro's policy was pretty much to disregard Primer's existence.

Ryan said...

Good point. She didn't stick around long, as I recall. She just confirmed it and thanked everyone for being cool about it.

Yuda said...

Did you happen to read today's PTP? They spend most of the Astros section explaining how and why PECOTA blew its projection for Clemens last year, and then uses this year's projection to explain why the Astros overpaid for him.

Ryan said...

You the man, Yuda. Good catch. A friend of mine suggested that it would be interesting to see how many guys exceeded their 90th percentile PECOTAs. Clemens did, as they admitted today, and I'm pretty sure Wilkerson did as well.

Basil said...

That's pretty funny, Yuda.

Yuda said...

The worst part of it was that the "why" of it blew the Clemens projection last year is because there haven't been many (or any, really) players as old and effective as he is. So they were basically admitting that PECOTA can't get his projection right, unless it just gets really lucky.

Basil said...

Ah, the old error-at-the-edges. That's the funny part about, say, James' Pythagorean stuff. It's a great tool, of course. But it only works when teams' numbers tend to bunch up closer to the middle. So if you want to know whether an 87 win team is a 92 win team, fine---but if you want to know whether the '02 Braves are a 101 win team or a 108 win team, or the '03 Tigers area 49 win team or a 58 win team (or whatever), you're out of luck.