<$BlogRSDURL$>

Wednesday, July 20, 2005

FREE SAMPLE
I'm sure you've noticed Vlad's recent slide -- he's 3 for 38 over his last ten games, knocking his line down from 344/399/590 to 314/367/543. But you've probably also noticed that he's been hitting the ball well of late; he's smashed line drives right at people, driving outfielders to the track, been robbed of hits on leaping catches, and justmissed a few balls that would have been homers had he not been off by a centimeter.

All of which to say that Vlad's going to be fine, and his recent statistics are a great indication of why small sample size has to be taken with a boulder of salt. Looking at his line would lead one to believe that he's been swinging the bat poorly, when for the most part that has not been the case.

Also in a mini-slide is Adam Kennedy, who is 2-15 over his last five games. Adam's performance illustrates the dangers of sample size in the other direction; though he's a good player, he's not really a .350 hitter, and was bound to see some regression sooner or later. And when he was hitting .350, though he was hitting the ball well, he was getting all of the breaks, and getting a lot of bloopers to fall in. That's not happening so much anymore.

The good news is that we know Vlad's capable of going on a tear at any time, and, the way he's been hitting the ball, he's only a inch or two away from being on one right now. And we also know that Adam Kennedy can have his average drop another sixty points and still be a productive ballplayer. So these "slumps" don't concern me. It's just sample size.

Tonight would be a good night for the breaks to come our way, having to face Barry Zito and all. Zito has been en fuego since June 22:

G   IP   H  HR  SO  BB  ER   ERA
5 37.3 20 5 21 11 6 1.45
This may sound crazy, but I look at that line, and I think, "We can get this guy."

First of all, the strikeout-to-walk ratio is unspectacular. He's also allowed more home runs than he should have. The Angels can get to tough lefties who allow a few too many home runs.

Furthermore, two of those Zito starts came against Seattle. Yeah, Seattle whupped us for four games going into the break, but they aren't that good a team. Zito also had two starts against the ChiSox in there; the White Sox rank an unspectacular sixth in the league in runs while playing in a good hitters' park; their success has come from pitching and defense more than offense.

And let's dismiss another possibly worrisome factor: the fact that the Angels hit a mere 257/314/389 against left-handers (as opposed to a 274/324/418 against righties), so that would seem bad. But Zito, for this year and much of his career, has been tougher against right-handed batters than left-handed batters:
Year        vs RHB   |    vs LHB
AVG OBP SLG | AVG OBP SLG
2005 205 294 338 | 250 306 420
2004 248 306 413 | 323 419 479
There are a few reasons for this. One, as we learned before, we need to be aware of sample size; Zito faces many more RHB than LHB. Two, the LHB are usually the best LHB because the lessors get the day off when Zito pitches.

Three, I believe there is a real factor in Zito's pitching style: his cut fastball. He runs it in against RHB and can really mess up their game -- think of the Sunday night game last year that was the first time Vlad ever faced Zito. It's a neutralizing pitch against righties, and Zito doesn't really have anything special against lefties -- he just has the same curveball he always uses.

So, the fact that the Lads struggle against lefties, and that Zito is a lefty, doesn't bother me. He's a reverse-type pitcher, he's really like a right-hander.

But, I'm sure, now that I've pointed out all these reasons Zito is hittable, he'll go out and throw a perfect game against us tonight. So, if that happens, blame me: I can take it.

Comments:
How do you explain using the smaller sample size of AK as closer (or moving towards) his "real ability" and the larger sample size of Vlad as explaining Vlad's "real ability?" This is the problem I have with SABR stuff. Isn't it possible that it has nothing to do with luck and Vlad has just gotten out of a grove? Sure, he is still hitting the ball hard (it's Vlad, when doesn't he?), but maybe he is hitting pitches that are worse for him to hit. And maybe AK really was a .350 hitter (not a lucky hitter) and something changed in his swing since the All Star Break that has him screwed up?
 
A fair question.

I would admit that it's possible that Vlad is out of a groove and it's possible that Kennedy really had the ability of a .350 hitter for a few months.

But my eyes tell me those things aren't true, and their numbers tell me it isn't.

I look at Vlad and see a guy who hammered a ball to the right-field fence that Jacque Jones had to leap to catch; a guy who flew out to the deepest part of Angel Stadium; a guy that smashed a line drive that Gold Glover Eric Chavez just got enough of a glove on to keep in front of him to register an out. I see a guy who's hitting the ball fairly well.

I also see a guy who has a proven ability to hit around .330 or whatever it is. When a guy who's a .330 hittter hitting the ball hard and going 3-38, I say, "He's hitting in bad luck, he should bounce back at any time."

In Adam's case, according to my eyes, in June and the beginning of July he was hitting the ball well and getting the breaks. He was getting bloopers to drop in for hits. Essentially no one in today's game is an actual .350 hitter (except maybe Ichiro!). You hit .350 by (1) being good and (2) having some good luck.

We also have a bunch of Kennedy's performance that says he's a .270 hitter or .280 or whatever it is. When that guy hits .350 for a few months, you can either say (1) he's really figured something out or (2) he's caught a few breaks.

Sure, (1) happens. Player reach new levels of performance: just look at Melvin Mora. But (2) happens a lot more often, and even when something like Brady Anderson hitting 50 HR happens, the player will normally revert to their usual range.

I am not an expert, but in my observation, Adam's swinging the bat just about the same as he was when he was hot. But let's say I'm wrong, and he really was swinging differently in June: if he can't sustain that -- and there's nothing in his history post-CSUN that says he can -- then, for all intents and purposes, it doesn't matter if he was "really" a .350 hitter or a good hitter having good luck.

I think the hangup here is the word "luck," so let me come at it another way ... in Men at Work, by George Will, Tony Gwynn reflects on a game where he went 1-3 with a home run. The home run disturbed him and the two outs pleased him. Why?

Because Gwynn knew that the swings on which he made outs were better swings than the one on which he got a home run. He knew that, though it might have worked on that day, in the long run, he was going to make a lot more outs swinging the way that brought him the home run than the way that brought him the outs. He was satisfied with the outs because he knew he had done his job, but had just happened to hit the ball right at someone.

When I think of "luck" in this context, that's what I think about: sometimes you succeed doing the wrong thing and fail doing the right thing. Over the long run -- 600 AB, 1500 AB -- those things will generally balance out. Adam Kennedy being a career .270 or whatever hitter is more telling than his hitting .350 for two months.

Do players' true talent levels change? Sure they do. I love Adam Kennedy, but he's going to have to hit .350 for more than a few months to convince me that luck was not involved.

I don't know if that really addressed the objection, it got kinda verbose there ...
 
I suppose my main objection was that you are using different rules for AK and Vlad. Certainly Vlad is a better hitter and typically has a higher average (and .350 is within the upper limit of his BA range in any given season) and AK is not typically a guy that can hit more than .300. Also, I do not have the anecdotal evidence you do (the whole no Angel games on TV thing, which ironically I get to watch one tonight), so I cannot tell if someone is hitting the ball well or not, but isn't the difference between a long out and a home run or a double sometimes the difference in someone whose swing isn't perfect or someone who is trying to hard to hit the ball?

Anyways, back to the different rules you use for the two players...you seem to be saying that Vlad's recent slump is just that: a slump. Whereas AK's recent slump is a regression to his mean. I am not saying that those aren't correct interpretations, but that is a poor way to define causality (AK's slump is caused by regression, while Vlad's is random).

This is the type of thing I would notice Rob Neyer do when analyzing a group of players a lot before he went to ESPN Insider and I stopped reading him (and I absolutely despised it when he did). Once, to make his point about what pitcher was the best of all time, he took about 10 great pitchers and put up comparisons of various stats for them and then would say: well, this stat isn't very important, so these guys are eliminated. And then he would bring that stat back to eliminate other guys. You can't use the same stats to say opposite things.
 
Dude. Please fix your purported "comment" link, which does not actually have the "#comment" on the URL. This is breakage I had to fix with the same template you're using.

Sure, (1) happens. Player reach new levels of performance: just look at Melvin Mora.

I would say, just look at Brian Roberts.
 
I have no idea of how to fix that comment link thing. I've tried it before, to no avail. I'll try again.

Back to Josh's objection ... I think it's as simple as Vlad is a .330 hitter, he's hitting below that, I perceive that he's hitting the ball well, I therefore conclude that it's a slump and he should bounce back.

With Kennedy, he's a .270 hitter, he's hitting well-above that, I perceive that he's had some hits on bloopers, I therefore conclude that his recent minislump is him regressing to the mean.

I don't think I'm using the same stats to say opposite things; I'm trying to say that here are two players who are slumping to some degree, are they going to bounce back or not? The answer to that question is based on the history of each player, which in these two cases is very different.
 
I am sorry, it still doesn't quite jibe with me. The idea of using stats to say that one slump is "just bad luck" while another is "regression to the mean" just doesn't sit well with me. Yes, their past results do matter, but every year is different. At what point is it decided that a certain level of statistical prowess is within a player's reach? When he is age 27? After 5 years in the majors?
 
See, if we really knew the answer to that, we could run baseball teams that won the World Series every year. It's all about using information to make the best guess possible.
 
Oh come on, this argument is getting silly. Vladimir Guerrero is one of the best hitters in baseball, and has been for years now. Adam Kennedy is an above average hitter for a middle infielder, and has been for years now.

The larger sample is not the weeks Kennedy has been hitting .350, the larger sample is his entire career, none of which suggests he's the reincarnation of Rogers Hornsby.

One slump is called regression to the mean because it's bringing his season's numbers closer to his career mean. The other is not because it's not.
 
At what point in a season do you decide that a person is actually hitting that well? Do Derrek Lee's numbers not really count because he is so far above his career norms? What about Adrian Beltre last year? Was it just a fluke or was he for some reason particularly locked for the season? When is a small sample size big enough to not be considered small anymore? When is a career at the point when you can no longer expect improvement? The problem with stats is that they don't account for the human element: there are both physical and psychological reasons a player may be hitting well or slumping at any given time. This is something that annoys me about statistics: you can make them say anything you want. The key is to be careful in how things are presented, and I just feel that the claim was used that a particular small sample size was dismissed as bad luck, while another almost equal small sample size was claimed as a regression to the mean and that is not the best way to state these things. I am not arguing he is wrong, just that I think that the idea of "small sample size" was used in two opposite ways.
 
The thing is: Adam Kennedy going 2-15 or whatever it was is not the small sample size I'm talking about; his small sample size is the past two months where he hit .350.

For Vlad, the small sample I was referring to was the past few weeks.

No one's denying "the human element." Sometimes players are "locked in," sometimes they play hurt, etc.

Every player's performance, for any period of time, is a sample of his true talent level. The bigger the sample, the more clear a view of his talent level we have. Of course, that talent level can be a moving target, as players improve and decline for any number of reasons. That's what makes it hard. If we could always know what perfs were flukes and what were real changes in ability, we would always know who to sign to big contracts and who to avoid.

We don't always know these things. We make our best guess. I saw Adam Kennedy hit .350 for two months, and I didn't say to myself, "Oh, great, Adam Kennedy's become a .350 hitter." I said to myself, "Adam's swinging a mean bat and getting some breaks; maybe he'll be able to hit around .312 this year like he did in 2002. That would be great."

That's my opinion; that's my guess, based on his past performance and my own observation of his play.
 
Post a Comment

This page is powered by Blogger. Isn't yours?