Sunday, March 9, 2008

Computers Don't Know Anything....After All, They Can't Love Or Hate Or Feel Pain!

Rick Morrissey has his own theories about how to predict the fortunes of baseball teams. There's a thing called a "hunch", people. We don't need no stinkin' computerboxes.

If computers ran the world, Steven Seagal probably would have won a few Oscars by now, assuming they judged him on the $2 billion his movies have earned. If computers had a way of measuring acting ability, he'd be running a martial-arts school in a strip mall.

Okay, equating judging an actor by how much money his movies make is sort of like evaluating hitters using batting average and evaluating pitchers using wins. It's stupid and useless. The computers you're about to declare worthless don't do anything that simple. Chalk up another to "terrible similes and metaphors."

But they don't run the world, yet, which means we can still type in our credit card numbers online without worrying that all our money is being sucked into a fund earmarked for global dominance by a dastardly computer.

Attention all forms of cerebral parasites: do not feed on the brain of Rick Morrissey. It is clearly diseased and "possesses the properties of the insane."

Computers have no use for heart, or least they can't quantify it. They can't analyze what's inside an athlete, for example. They can't tell you who has the heart of a lion or the backbone of an earthworm.

They can't!?!??!??!?! Whew. Good thing for computers that having the heart of a lion doesn't make a human being any more likely to hit a curveball. In one corner, we have a computer using projected stats to formulate run totals from run elements to predict a final season record of a team. In the other corner, we have Rick Morrissey arbitrarily placing people on a scale of (Earthworm) Jim to Simba to calculate wins. Vegas's odds on Ricky are 234.5:1. If you like Eckstein a lot, financial opportunity is knocking......

Computers can't tell you that White Sox first baseman Paul Konerko is upset with how he played last season.

They don't need to. Everyone who knows anything about Paul Konerko knows that. The job of computers (not that dissimilar to that of journalists) is to tell people things that they don't know.

All they can tell you is that he hit .259 in 2007, that he just turned 32 and, therefore, he must be on the downside of his career because that's what the model says is supposed to happen to him.

Really, that's it? How totally uninformative.

(Interesting side note: BP 2007 listed Tino Martinez as one of Konerko's top PECOTA comparables and pointed out that Martinez had a severe production decline at age 31, which was exactly what happened to Konerko. Damn unreliable squawkboxes!)

If you saw the piece about Baseball Prospectus' 2008 predictions in Sunday's Tribune, then you know the publication's computer has the Sox going 77-85 and finishing third in the AL Central, and the Cubs going 91-71 and winning the NL Central.

I know as much about computers as I do about astronomy


Which puts you in just an oh-so-good position to dismiss them as irrelevant, doesn't it?

but I believe the computer term for Baseball Prospectus' Sox prediction is "fatal error."

The prediction for the 2008 White Sox literally made the computer implode. Rick's not joking.

I have the Sox winning 85 games and giving Cleveland a run for its money for second place in the division.

Okay, sure. I'm positive you'll tell me what this is based on shortly. But as long as we're being insanely optimistic, why not go all the way and say they'll contend for the division lead? After all, there's no way Cleveland is that far below Detroit (if you ask me, Cleveland's slightly better).

I know, I know: The Indians are loaded with talent, and if it weren't for Detroit spending gobs of money, they'd be the favorites in the AL Central.

Actually....BP still has them as favorites anyway. But, moving on, moving on.

But, again, what about heart?

That thing that beats out grounders to first base? No....that's speed. That thing that makes the ridiculously tough play in the outfield? No....that's speed too, and good judgement and athleticism. Hit home runs? No....that's power. Striking players out? No.....good arm strength and pitch placement. And we all know that it can't be walking, because walking is like the anti-heart. Where does "heart" fit in there, Rick? I'm pretty sure that the Sox aren't trying any more gooder than the other teams. Plus, they punted Erstad (HAHA! HAHA! A PUN!), so they're already out 51 "cardions" in the heart statistic. See? You CAN quantify it!

Hal (or Smitty or Shecky or whatever the computer's name is)

We'll call him "Slobby", thanks to Ed Hardiman and the slobbermetrics it uses.

and I pretty much agree about the Cubs, which, given my track record on predictions, should make Hal/Smitty/Shecky do a lot of soul-searching

So you're pretty much admitting that you're wrong a lot on predictions? I could never have fathomed this! You base the things you say on so much excellent evidence!

...I...I thought we just named him Slobby.....

which is impossible because it doesn't have a soul, just an evil chip that makes it want to mate with Marie Osmond and produce robots that sing show tunes.

Oh no he di'int!

But: and this might seem like an irrelevant question, Rick. Does the soullessness (I'm inventing words, this guy is so stupid) of a computer stop it from predicting the outcomes of baseball seasons. Does it? Just answer, I won't go crazy and argue it has a soul.

Do feelings count? Or hunches?

Hunches....you mean those things that by definition are based on absolutely nothing? I have a hunch that Alfredo Amezaga will outproduce Matt Holliday this year. Just a hunch. It counts, dammit!

Where is there room in computers for the inexplicable?

Nowhere, just like there's no room in baseball for it either. Improbable, yes. Inexplicable, no.

Does the fact that it's the Cubs' 100th season since their last World Series title mean anything in the computations?

I dunno, Rick? Do you think that the last 100 Cubs seasons (which, by the way, primarily featured players not currently on the Cubs' roster) have any bearing on how many games this team will win?

Does it mean anything that the Cubs could be driven by the challenge of a century of dryness or, conversely, that they could cave in under the pressure of it and finish 10 games below .500?

Okay, let's assume that's true. Those two possibilities cancel each other out approximately, leaving the same initial prediction. The computer doesn't think these things CAN'T happen, Rick, just that they are unlikely. It just takes the record in the middle of the range of the most likely outcomes. Why is this so fucking hard to understand?

I believe the Sox are embarrassed by what happened last season and, not to belabor the point, there is nothing in a computer's innards that can measure the effects of that.

So what are you saying, an embarrassed with-something-to-prove Jose Contreras will be better than his physical ability?

That the Sox dropped from 90 victories in 2006 to 72 games last season was one of the shocks of the baseball season. But not to Baseball Prospectus, and the people who run it deserve their props. They chalk up a lot of what happened on the South Side last season to the inevitability of time catching up with older athletes. I chalk it up to a number of players having down years at the same time.

Right, you attribute it to that because you're not paying attention at all. The team is losing, so you exacerbate what went wrong, and fail to pay attention to what went right! For every player on the Sox who had a bad year (relative to the expected), there's one that had a pretty good year by his standards. Jim Thome fought aging very well, and had a very good season. Bobby Jenks broke out. Mark Buerhle, Javier Vazquez, and Jon Garland all had very good seasons by the standards of their ability. Players stopped being so absurdly healthy like they were when the team was performing very well.

Isn't there room for a number of Sox to have good years at the same time? Say, in 2008? If Jim Thome stays healthy, he could have an excellent season.

Just like he had last season! .328 EqA+!

The Cubs don't have a good enough rotation to do the impossible and win the World Series

The '06 Cardinals were famous for having an AMAZING starting rotation. Guys like Jeff Suppan and Jeff Weaver. Top-of-the-line talent, really.

but perhaps Carlos Zambrano's feistiness becomes contagious and the staff starts pitching like the '69 Mets did. Can a computer comprehend feistiness? I don't think so.

Oh dear Lord! The computer's calculations were leaving out feistiness! Silly computer! Move the K/BB ratio and GB/FB ratio projections, and remember that Carlos Zambrano has a fire in his belly! Ryan Dempster and Jon Lieber will rediscover their youth! Jason Marquis won't suck anymore, inspired by Zam punching random objects and pointing to the sky after every inning! I plan to destroy the computer on which I am typing after this post. How could it have missed this?

This is the time of year for predictions, so it's not surprising there would be a few bad tidings, especially for the Sox.

The problem with computers is that you can argue with them until you're blue in the face, and they don't even blink in response. There's no satisfaction in it. You can, however, achieve a higher level of contentment by hitting them with a baseball bat.


Agreed. So long, Compy. I hardly knew ye. I'm sorry for shouting aimlessly at you until my face turned blue.....I just....I just don't understand how you can't comprehend the grittiness of Pablo Ozuna and Ryan Theriot! Forgive me, we're from different worlds. Maybe we'll just never understand each other. ::sniff::

::CRASH!::

26 comments:

Archie Micklewhite said...

Question...did you just make this article up? I mean, it's fine if you did, but just...honestly, how the hell does one article hit EVERY SINGLE CLICHE imaginable? Computers are evil, what about grittiness, can't measure heart...it's just all there. It's like this article was wrapped up nicely and tied with a bow, waiting for people like you to come along.

Archie Micklewhite said...

And by "people like you", I mean "bad sportswriting deconstructionists." Or have we agreed on an easier-to-use term yet?

Derpsauce said...

Archie...awesome name dude. And no, sadly, I didn't invent this, though I was bored enough to.

Derpsauce said...

"Basement dwellers" will do just fine, thanks.

larry b said...

If computers ran the world, Steven Seagal probably would have won a few Oscars by now, assuming they judged him on the $2 billion his movies have earned.

I think what he meant to says was "If Larry B ran the world, Steven Seagal probably would have won a few Oscars by now, assuming they judged him on how enjoyable many of his movies are. Specifically, Exit Wounds and Hard to Kill."

Unknown said...

If computers were ruling the world, wouldn't they have a way of measuring acting ability too? Steven Seagal would have won several Oscars AND be running a martial arts dojo in a strip mall. The Eckstein would probably be there teaching grittiness.

Seriously though, Pnoles points out something that all these writers seem to miss...Morrisey and the rest admit they have no idea what the computers are compiling, how they do it, or the significance there of. They might as well start commentating about the flight characteristics of the F-22.

Tonus said...

What you're seeing is the predictable reaction to the growing success of sabremetrics. Writers are desperately clinging to the old and thoroughly arbitrary way of judging baseball performance and talent, and they're losing ground. You will pry the feistiness out of their cold, dead hands.

By the way, I'm picking the New York Knicks to win their division next year, because they are so embarrassed by their recent performances that they will be overflowing with heart and grit and pissedoffitiveness. Sure, the computer will pick them to have another lousy season, but can a computer measure pissedoffitiveness?

Yeah, I didn't think so.

Archie Micklewhite said...

Sorry dude, I'm pretty sure I once saw Zach Randolph reading Baseball Between the Numbers. He said all the talk of VORP made him excited, and also that he really respected everything Paul DePodesta's been through). So I don't think the Knicks are going to be capable of the necessary grit and/or hustlers with sabermetric losers like that clogging them up.

Chris W said...

frankly I'm pretty nonplussed re: PECOTA.

Notice the following things:

1.) The only actual fact you ever really hear about PECOTA's merits is that it, despite much outcry against it, predicted the White Sox's 2007 Record...not other teams. It's always the 2007 White Sox

2.) The Pecota Predictions for the 2003-2006 White Sox couldn't have been more wrong.....

Look: Pecota is slightly better than Rick Morrissey "going by his instincts" when those "instincts" are based on "gut feelings" and not on "analysis" but it's not much better.

Frankly, statistics as a predictor rather than a guideline is a poor model and experiential, visual, and even anecdotal evidence (like "Konerko's not the kind of guy who will let himself have two consecutive lousy seasons, no matter how old he is," or "Cooper's going to focuse on having Contreras throw more first pitch fastballs") is not a poor substitute from statistical models.

Remember: the human element is still a very big part of baseball. After all, what kind of douchebag would try to predict Jermaine Dye or Matt Thornton's production in 2008 given their mercurial history in the pros?

Anonymous said...

Some Tips on how to monetize Blogs
Making Extra Money With Your Blog.

Derpsauce said...

"Konerko's not the kind of guy who will let himself have 2 consecutive lousy seasons, no matter how old he is" = meaningless and stupid.

"Cooper's going to focus on having Contreras throw more first pitch fastballs" = meaningful.

Do you see the difference between those two observations, Chris? I'm not dismissing the latter as stupid. The former is awful.

Derpsauce said...

On a side note....here's some interesting PECOTA 2008 projections:

Rays win 89 games.

Mariners take last place in the West despite acquiring Bedard.

Chris W said...

The first is relatively meaningless, but not completely meaningless.

Some players respond to a bad season by doing nothing. Others take measures to correct them.

Konerko's a slumpy hitter but he generally gets the numbers. Last year he didn't. That suggests to me a problem of approach rather than a physical deficiency/age-related decline, especially since, from watching the games, he didn't seem to be having problems with his swing/catching up with fastballs.

Obviously this isn't something to STAKE any kind of prediction on, but anyone who watches White Sox baseball would be pretty shocked if Konerko has another bad year. That's all.

I still hope we trade him for prospects

Jeff said...

My main problem with this terrible terrible piece of dogshit baseball analysis is lines like this:
-----------
Computers have no use for heart, or least they can't quantify it. They can't analyze what's inside an athlete, for example.
------------
Is Rick Morrissey saying that we should listen to him, and not a computer (really, a calculation), because he CAN quantify "heart"?

That's an early favorite for "Worst piece of baseball analysis - 2008".

Bengoodfella said...

Maybe I am being too moderate but I believe we could use computers AND logic to determine whether we player and team is going to have a great year. I think it is possible to use both.

I also believe it is dumb to poo poo on computers and what they can show/predict. One thing Mike PECOTA can't predict is "the ratio of articles written about Mike Hampton coming back strong this year" to "How angry I get because I feel like it is jinxing him."

I think the funniest thing I have ever read about old timers not needing projections was the Bill Plascke oriented post(not spelling his name correctly) on Fire Joe Morgan about Andre Ethier and an old scout who did not need numbers to see what a great ball player he was. As if he was recommending an incredible diamond in the rough and not one of the highest ranked prospects in the A's organization.

I just don't see why sportswriters always think those who make projections using PECOTA only believe that, and only that, can determine how a team/player is going to do. They are freakin' projections that can be used for dicussions/fun. I just think sportswriters are not comfortable with potential facts that don't involve intangible variables like grittiness and the will to win. Since you can't measure grittiness and the will to wininess, then the shitty articles they write about these characteristics can't be proven incorrect. I am rambling...

larry b said...

For once I agree 100% with Chris W. (Number of times this has happened since we met in 2002: four. Including this.)

If you have the kind of healthy attitude about projections Ben expresses above, great. Keep it. But there were entirely too many intensely pro-SABR bloggers who made an entirely too huge fucking deal about the fact that PECOTA nailed the 2007 White Sox. Do you know how many wins it picked for the Rockies? I didn't hear much about how that prediction. Again, it's fine if you have a healthy and moderate appreciation for it. But I'm not even a White Sox fan, and I got really sick and tired of reading "ZOMG PECOTA knew exactly how many wins one team in the league would win last year, therefore it's always right and anyone who wants to use any other method is an idiot who doesn't know anything about baseball."

Now I'm rambling. You can see the point I'm trying to make: being the anti-Rick Morrissey is just as bad as being Rick Morrissey. Moderation, my friends, moderation. Except when it comes to living in basements. That should be practiced to the fullest extent possible, with little or no time spent anywhere else.

Bengoodfella said...

When I comment on a post it seems to stop any further commenting. I have noticed this and the reasons could be (a) what I say makes no sense and frustrates others into quitting or (b) i always come in at the tail end of the post.

I do have to agree with Larry that people made too much of a deal out of PECOTA nailing one team. I personally think it is fun to use to predict how my favorite players are going to perform and it helps me get excited/depressed for the new year of baseball. Nailing the White Sox exact total was a fluke, and I doubt it will nail an exact win total for a team this year. There is a lot that goes into why a team wins and loses obviously.

Except the Pirates. They lose because their front office in the past has not cared and you don't need a PECOTA projection to tell you that.

larry b said...

Not a lot of Pirates fans around FireJay ready to defend their team these days, I'm guessing.

Chris W said...

The whole "what did the experts say about the White Sox in 2005" angle is played out but let's be realistic: They predicted like wins in the low to mid 80's for teh Sox in 2005.

Now, that team was a surprise, but so were the 2007 Sox. How is the 2005 butchering of the prediction any less relevant than the 2007 nailing of?

Further, the PECOTA predicted the 2007 Sox's downfall would be starting pitching. Though it also predicted some decline in offensive production, it didn't predict it to the extent that the 2007 White Sox sucked at the plate, nor did it predict that the bullpen would be their main shortcoming.

But, yeah, it totally guessed the win total of a team for one random year despite being completely off in other years and misrepresenting the reason why that win total was waht it was.

More like "poop"-cota

Tonus said...

From what I've seen, the guys at BP would be the first in line to point out that forecasting any player or team's performance based on a rating system like PECOTA is nowhere near to being exact. Many of them stress the need to find a middle ground between statistical analysis and less easily modeled factors.

But the reality is that the day of people like Morrissey, telling us that statistical analysis is all but useless because you "can't measure heart" is beginning to decline. We'll never lose it entirely because sports have an emotional edge to them, and we like that edge. Sports will never be "just numbers" to real fans of sports.

First step, of course, is to rid the industry of people who get paid to write awful crap. Looks like you guys have your work cut out for you!

Derpsauce said...

PECOTA isn't perfect, but it's superior to any individual system out there. No, PECOTA didn't say peg the 2005 White Sox as World Series champs, nor did it predict the success of the Rockies last year, but come the fuck on....who did? It's nowhere near infallible, but it's better than just about everything we have available (and certainly better than Morrissey's wild ass-guessing) It's not "deadly accurate" as it boasts, because there's so much shit that can happen in a season and plenty of teams finish off their projection. The standard deviation is significant, but PECOTA hits damn close to the median of possible outcomes. It's an approximation. Maybe too much is made out of the 72-90 Sox projection, but the POINT of that projection isn't the record itself, but basically that the White Sox had very low chances of being contenders. Even if they finished like 77-85 or something, the projection was still helpful, and a solid argument against everyone who thought the White Sox were still going to contend in '07.

larry b said...

PNoles and Tonus- My issue isn't with PECOTA itself or BP themselves. Its with people who take what they say and then run twenty more steps down the spectrum with it.

So we agree. Mostly.

Archie Micklewhite said...

It's worth pointing out PECOTA was actually one win off from the actual White Sox record (they predicted 73-89). They did nail the other Chicago team, though, picking the Cubs to go 85-77. I actually did a really quick analysis of last year's PECOTA and it turns out they were, on average, within 4.867 of a given team's actual W-L record, which I think is pretty reasonable, especially because they were skewed by a couple extreme outliers (the Mariners and the Twins).

I think a final point to add is that the reason why PECOTA deserves a lot more respect than Rick Morrissey is that their system isn't built on some bullshit, fallacious appeal to authority. Instead, the system is constantly revised to take into account the new data set provided by another season. It's built on learning from being wrong, not always forcing oneself to be right. Huge difference.

Also, I have to admit I missed all these ultra-saber blogs that soured some of you on PECOTA. Maybe it's because the only way I read actual sportswriting is when it's being ripped apart by one of the FJM's, which means the only sports journalism i read is shitty traditional stuff as filtered through angry stathead bloggers. I guess if I started reading that blog Mike Celizic and Plaschke run, Fire Will Carroll, I'd start seeing more of those.

In conclusion, by all means let's be reasonable and moderate, but let's also not suggest PECOTA and an analysis of Paul Konerko's emotional makeup are equally valid. There's such a thing as too much compromise, you know.

Archie Micklewhite said...

Also, final point in defense of PECOTA - I'm not 100% sure about this, but I'm guessing it tends to be way more accurate regarding individual players than overall teams. I mean, I don't know whether team PECOTA predictions are derived from the sum of the player projections, but either way it seems probable things get way less clear the more variables you add. And even with all THAT said, PECOTA isn't really making an absolute projection for a given player but a series of probabilities. So some play at their 90th percentiles, others at their 30th. Over the long haul, it all averages back out again.

Is this convincing to non-PECOTA people? I suspect not. But I'm definitely a fan.

Tonus said...

I don't think it's that anyone here is "anti-PECOTA" as much as there is a recognition to understand just what the tool does and doesn't do. It's a good forecasting tool. And a forecasting tool simply gives us a good idea of how things will turn out, assuming that certain factors remain true (ie, injuries and outliers mostly).

I think that there are people who push the idea of statistical analysis to absurb extremes. But also people tend to take sites like FJM (both of them) a bit too much at face value instead of understanding the humor factor involved (yeah I know, what a straight line!). Sites like BP and related pages are much more moderate in their approach.

Then again, most people who are at the extremes don't really read sites like BP, and they've either never read Moneyball, or they've memorized it line-by-line. They're not rational people.

Tonus said...

BTW, that first paragraph I wrote makes just about no sense thanks to the atrocious grammar usage. It should read:

"I don't think it's that anyone here is "anti-PECOTA" as much as there is a recognition of just what the tool does and doesn't do. It's a good forecasting tool. And a forecasting tool simply gives us a good idea of how things will turn out, assuming that certain factors remain true (ie, barring lots of injuries and/or outliers)."

Rest assured that whatever distracted me to that degree was some sort of very manly activity, like a threesome with two ultra hot chicks or delivering a savage beating to some annoying clerk at the local grocery store. And not something lame like my mom asking if I'd like some chicken soup, because I'm bound to catch a cold if I spend all day in the basement.