Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Friday, November 16, 2007

vorosmccracken: Me Being Arrogant Again

Direct from the punk section at TRAX…the psychedelic fur is flying!

I’m sure David Cameron will love this since he knows how thoughtless and unreasonable I am, despite having never met me. But once again I’m forced to express an unpopular opinion, because it happens to remain my unpopular opinion.

But in an interview with JC Bradbury, Keith Law brings up an issue that gets me in trouble. Now don’t get me wrong, I like Keith. I like Keith a lot, not the least of which because he reminds me a bit of Jon Cryer and I like Jon Cryer.

...Again, this is not like a wide ranging indictment of Keith Law or some bizarre boasting about my own superior methods, but no one in this field does anyone any good pretending to believe something they don’t believe. And that’s the fact: I don’t believe it. If that makes me arrogant, fine. But I’m not backing down from my insistence on things like evidence and peer review. “You just know” doesn’t cut it with me. When someone can satisfactorily explain the step by step process of separating the difference between getting your knees buckled on a slider and “showing fear” at the plate, and then demonstrated it through evidence, I’ll be fully converted. Until then I’m going to call ########.

Repoz Posted: November 16, 2007 at 04:46 AM | 236 comment(s) Login to Bookmark
  Tags: community

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

Page 2 of 3 pages  < 1 2 3 > 
   101. GGC don't think it can get longer than a novella Posted: November 16, 2007 at 07:58 PM (#2617677)
Mike, I thought that the main quibble with the methodology that Voros used was that the group of pitchers he used in his study were all good enough, that they had to be good at preventing hits on balls in play. Thus there'd be guys like Kyle Snyder (IIRC, gaelan or someone used him as an example) who had good peripheral stats but still got hit hard and thus didn't pitch enough to be in the McCracken study. I'd be interested in seeing what the problems are with studying correlation between sequential time series points is, because I think that's how many of the famous (in certain circles) baseball studies went (eg, Dick Cramer's study on clutch hitting and Keith Woolner's "Field General or Backstop.")


Voros, I doubt that you know me from Adam, but I've been enjoying your blog even though I may not be the world's hugest DIPS fan (helping the Red Sox win it all goes a long way in my book.) I like that self deprecating humor.
   102. birdlives is one crazy ninja Posted: November 16, 2007 at 08:05 PM (#2617688)
I give up. It's like I'm speaking Klingon.

Could you at least explain why people are being unreasonable?
   103. BDC Posted: November 16, 2007 at 08:07 PM (#2617691)
I look exactly like Rowan Williams, the Archbishop of Canterbury, so much so that it's kind of scary. I was walking around Canterbury Cathedral not long ago and folks were lining up to kiss my ring.
   104. The District Attorney Posted: November 16, 2007 at 08:07 PM (#2617694)
Does "correlation between sequential time series points", translated into English, mean "when the guy does it one year, he tends to do it again the next year"?

If it does, then yes, I'm completely baffled as to why that would not be a good way to determine if someone truly has an ability or not.
   105. birdlives is one crazy ninja Posted: November 16, 2007 at 08:12 PM (#2617699)
Does "correlation between sequential time series points", translated into English, mean "when the guy does it one year, he tends to do it again the next year"?

If he's talking about conducting a time series analysis, I believe, but I'm not 100% sure, that it's called autocorrelation in technical language and that creates statistical problems in the analysis.
   106. JPWF13 Posted: November 16, 2007 at 08:14 PM (#2617703)
Au contraire. Most of the detractors have read the new stuff, too, and realize that the new stuff doesn't address the single most basic issue, that the method by which the problem has been analyzed is wrong.


Mike I think you are confusing where you stand among the pool of DIPS detractors in general. As far as I can tell most (as in a majority) have not read the new stuff, will not read the new stuff and are interested solely in repeating the assertion that, "Voros said pitchers have no control over BABIP- he's wrong" endlessly.

You are definately in the minority among DIPs detractors because you actually have a coherent and reasoned argument- others do not want to believe that DIPS could be right or even partially correct- it's the same mindset that wants to completely discount MLEs (not work on improving MLEs- just discount the idea entirely) and still wants to use RBI and Pitcher wins as metrics in analysing players.

If nothing else, Voros' work, and later work has shown that pitchers have far less control over BABIP than was commonly believed or assumed 10-20 years ago.
   107. Mike Emeigh Posted: November 16, 2007 at 08:26 PM (#2617721)
Any actual evidence will do.


OK.

In 2003, I saw JJ Hardy play for the first time. He had a horrible series, and I made some comments about his glove at the time which the Brewer fans jumped all over (most of them probably still remember that). When Carolina played Huntsville in the Southern League finals, Hardy played much better, especially defensively - he was probably Huntsville's MVP in that series, although the Stars lost - and I filed away the record.

I saw Joel Guzman play, for the first time, with Jacksonville in 2004. At the time, he was a hot prospect who'd made it to AA at the age of 19 after hitting for average and power at Vero Beach, a tough place to do both. His numbers were still good at Jacksonville, but after watching him play in three games at Five County, I had some skepticism. His reactions at the plate were slower than I expected from someone with his numbers, and he wasn't particularly quick in the field, either. Mindful of the Hardy experience, I held back forming a judgment - and then when I saw Guzman a year later, nothing was different, and he was in the process of moving off SS. I filed that one away, too. One year later, Guzman was traded to Tampa, and the prospect sheen was off. I was talking to a scout earlier this year; he asked me about Guzman, and my instant reaction was "slow bat". He told me that he had been asked to follow Guzman by his team when the Dodgers were shopping him, and filed very similar reports recommending against his acquisition.

Now, what's a "slow bat"? How would I convince you that Guzman had a "slow bat" as far back as 2004, especially when his numbers hadn't yet decayed to the level of 2006 in Durham? What would you deem "acceptable" evidence? There are certain tell-tale signs of a "slow bat": popups and foul balls straight back on good pitches over the plate, no ability to handle inside pitches at all - but ultimately it comes down to how the batter "looks" at the plate. In essence, it really is nothing more than just "his bat looks slow to me".

The problem you have - and it's not uncommon in this debate - is that you want independent confirmation of evidence that derives from a so-called "expert", because you start from the assumption that the "expert" has no idea what he's talking about. But I start from the assumption - based on my own observations and chats with people at the ballparks I attend - that anyone who watches more than a handful of games learns to pick up many of the same visual cues that the scouts use and can come to many of the same conclusions.

To go back to what villageidiom says in #91: if you want evidence that the scouts know what they are doing, go do some scouting.

-- MWE
   108. Mike Emeigh Posted: November 16, 2007 at 08:31 PM (#2617726)
Mike I think you are confusing where you stand among the pool of DIPS detractors in general.


I am taking about the pool of DIPS detractors "on this site". Most of them understand the issues with the method.

-- MWE
   109. CrosbyBird Posted: November 16, 2007 at 08:43 PM (#2617745)
There are certain tell-tale signs of a "slow bat": popups and foul balls straight back on good pitches over the plate, no ability to handle inside pitches at all - but ultimately it comes down to how the batter "looks" at the plate. In essence, it really is nothing more than just "his bat looks slow to me".

"His bat looks slow" is just a conclusion. There are some events you can describe that lead up to that conclusion.

You've already begun to explain what "looks slow" means in some descriptive fashion. It sounds very much to me like if you had film of some PAs, you could point out the trends you're spotting that go into your conclusion.

That's evidence. Someone else could examine film on 100 random players and see if those trends did or did not have any correlation with success. That's a testable methodology. And it's still not really quantitative in the sense that you're measuring the bat speed. Although maybe you'd want to and that would be a useful data point as well.
   110. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 16, 2007 at 08:45 PM (#2617747)
Re scouting: Has any study ever been done on the relationship between scouting reports and MLB performance, *after controlling for their statistics?* I.e., you'd take a historical prospect's PECOTA projection for his first 5 years in the majors, compare that to his actual MLB lines, and subtract to get the error. Then you see how strong the relationship is between his scouting scores (on the 20-80 scale or whatever) and that error. That would seem to me to be the ideal way to decide which scouts are actually worth their salaries and which ones just take up space. Who did the Marlins have scouting Hanley Ramírez? And who told Boston to trade him? If you could sum that up over the Hanleys and all the other toolsy SS with superficially unimpressive MLE's, I'd think you could get a real competitive advantage over other teams.

Re DIPS: My understanding of Voros's revised line on DIPS (and he of course can correct me) is not that pitchers have no control over balls in play. Rather, it's that the year-to-year variation is so great, and the standard deviation of the actual ability so small, that *for the purposes of projecting next year's performance* you are better off pretending that a pitcher has a league-average BABIP ability rather than trying to figure out whether his true talent BABIP is .287 or .293. By the time you have enough data--6-7 years, say--to conclude with any sort of confidence that a pitcher really does have some such ability, it's likely that he's aged enough that he no longer has it. (This may be the case with Barry Zito).
   111. Dag Nabbit: secretary of the World Banana Forum Posted: November 16, 2007 at 08:47 PM (#2617751)
"His bat looks slow" is just a conclusion. There are some events you can describe that lead up to that conclusion.

I get the feeling MWE wasn't basing the slow bat conclusion from the pop ups but from the fact he looked slow at the plate. He can speak for himself & likely will in a a little bit, but I think he's saying/implying he's seen other young prospects who hit into pop ups and other results without leading to the slow bat conclusion. It's how he looked at the plate more than anything else.
   112. AROM Posted: November 16, 2007 at 08:59 PM (#2617763)
Re scouting: Has any study ever been done on the relationship between scouting reports and MLB performance, *after controlling for their statistics?*


Probably, but they are doing the work for a team and it will never be made public. People outside the game don't have enough access to scouting reports to do such a study.
   113. Greg Pope thinks the Cubs are reeking havoc Posted: November 16, 2007 at 09:04 PM (#2617772)
*for the purposes of projecting next year's performance* you are better off pretending that a pitcher has a league-average BABIP ability rather than trying to figure out whether his true talent BABIP is .287 or .293.

I believe that if you have the know-how and the time, you just use heavy regression. So if a guy has a .285 BABIP and league average is .300, you regress closer to .300. In the first year you'll regress that guy to .298. It's a huge regression until you have a lot of years.

Having said that, if you don't regress, it's obvious that you're better off using league average (.300) than Y1 (.285) because it's much closer to the ideally regressed value. So if you have to choose Y1 or league average, you choose league average.

Note: All numbers made up since I don't know how to do a regression.
   114. birdlives is one crazy ninja Posted: November 16, 2007 at 09:04 PM (#2617774)
Now, what's a "slow bat"? How would I convince you that Guzman had a "slow bat" as far back as 2004, especially when his numbers hadn't yet decayed to the level of 2006 in Durham? What would you deem "acceptable" evidence? There are certain tell-tale signs of a "slow bat": popups and foul balls straight back on good pitches over the plate, no ability to handle inside pitches at all - but ultimately it comes down to how the batter "looks" at the plate. In essence, it really is nothing more than just "his bat looks slow to me".

I get the impression he wants you to translate "popups and foul balls straight back on good pitches over the plate, no ability to handle inside pitches at all" into quantitative parameters so he can go to ballpark and measure the exact same thing as you. I don't know if this is possible or even advisable.
   115. JPWF13 Posted: November 16, 2007 at 09:07 PM (#2617779)
I get the feeling MWE wasn't basing the slow bat conclusion from the pop ups but from the fact he looked slow at the plate.


You can do this watching any game, the different hitters-

As a Met fan- watching Carlos Gomez hit and Milledge hit is a fascinating study in contrasts.
Mets officials and some members of the NY Media are ga ga over Carlos, he's tremendously fast, has a strong arm, LOOKS like he has some power in him- Milledge's physical attribute (tools) all appear to be faded versions of those possessed by Gomez- except ONE.

You can't help but notice that Milledge has a quicker bat than Gomez, his swing gets up to speed quicker and goes through the hitting zone in less time. The difference is really visually dramatic- so dramatic that even knowing most fans and the MSM ignore minor league stats, I'm still amazed so many think Gomez is a better prospect.
   116. Toby Posted: November 16, 2007 at 09:11 PM (#2617781)
Voros,

don't give up. Many people agree with you. I know I do.

But there's evidence and then there's evidence. Just about anything can be evidence, it's just a question of what weight to give it.

I remember a discussion you and I had back in 2004 about Toby Hall vs. Michael Barrett, which one would have a better career. Your statistical projections said Hall would be better. I considered the specifics of his [edit: Barrett's] career to that point (the position switches from C to 3b and back) and his extrordinary struggles at the plate and suggested that he fit the profile of someone who had a stress/mental/emotional injury, and I predicted that he would recover in the long run and be a better player. I think I may have been right.

Around the same time there was a thread about Blalock and his incredible struggles after he came up; you took the position that the Rangers should not have demoted him, and I agreed with you on that one, it just seemed like a small sample size fluke.

The point is that evidence can exist beyond traditional statistics. At the extreme end, a projection for a guy like Darryl Kile after he died would predict that he would continue to pitch effectively; the statistics don't realize he died. Then there are the players who have physical injuries or stress/mental/emotional injuries; if they can get healthy, the statistics don't tell the full story about them. We know that injuries make athletes less effective, but we don't have very good metrics for determining whether they are injured or just underperforming. For that, we have to rely on educated guesses based on observation and testimony.

And that's where scouting comes in, I think -- not to tell you whether player A is better than player B, but whether player A is healthy and functional. The more healthy and functional he is during a period of time, the more reliable his statistics are during that period of time.
   117. snapper (history's 42nd greatest monster) Posted: November 16, 2007 at 09:16 PM (#2617784)
his swing gets up to speed quicker and goes through the hitting zone in less time.

But, couldn't this be easily quantified with video analysis?
   118. birdlives is one crazy ninja Posted: November 16, 2007 at 09:19 PM (#2617790)
But, couldn't this be easily quantified with video analysis?

You mean stick a radar gun on the bat?
   119. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 16, 2007 at 09:19 PM (#2617791)
And the point is that if you are going to regress to .298, you might as well just use .300 and save yourself the trouble. If you have a lot of data, it's likely that it's no longer valid because a guy's BABIP ability 6 years ago may be very different than his BABIP ability now (Roy Halladay leaps to mind).
   120. birdlives is one crazy ninja Posted: November 16, 2007 at 09:31 PM (#2617798)
The point is that evidence can exist beyond traditional statistics.

I get the impression that he hasn't heard any arguments that convince him of that point! It's Klingon to him. Look, auantitative analysis is currently favored over qualitative data in a lot of different arenas. The later is gaining respect but let's just point to public policy as an example. Quantitative analysis will carry far more weight if qualitative research is even let in the door.
   121. JPWF13 Posted: November 16, 2007 at 09:39 PM (#2617804)
But, couldn't this be easily quantified with video analysis?


yes

Juts as you can "quantify" the fact that Jeter has the slowest reaction time of any Yankee infielder through video analysis.
   122. Honkie Kong Posted: November 16, 2007 at 09:39 PM (#2617806)
You mean stick a radar gun on the bat?

Am surprised people haven't done this. There was talk ona cricket blog on doing this with cricket bats. They already measure the speed of the ball coming off the bat, something which might be very useful for those pbp metrics.
   123. snapper (history's 42nd greatest monster) Posted: November 16, 2007 at 09:44 PM (#2617810)
You mean stick a radar gun on the bat?

No. How many frames it takes the bat to get from first movement to the hitting zones. Kind of like what CBW did with video of pitchers.

yes

Juts as you can "quantify" the fact that Jeter has the slowest reaction time of any Yankee infielder through video analysis.


I don't doubt that at all. But, then why do we need scouts to tell us about "slow bats" if I can just send a guy to video tape the prospects?

Actually, why wouldn't you do this for a lot of scouting? Send a low wage/skill guy to videotape the hitters/pitchers, and let the true scouts stay at home analyzing. It would save them a lot of travel time.
   124. Yeaarrgghhhh Posted: November 16, 2007 at 09:45 PM (#2617811)
On your terms, that is - IOW, YOU get to decide what counts as "evidence", and what doesn't. In my experience, that kind of discussion goes nowhere.

So, do you get to decide what counts as "evidence"? Are we supposed to just accept what you say because you're Mike Emeigh and we're not?
   125. birdlives is one crazy ninja Posted: November 16, 2007 at 09:55 PM (#2617821)
No. How many frames it takes the bat to get from first movement to the hitting zones. Kind of like what CBW did with video of pitchers.

Interesting idea, just sort of thinking out loud, is the technology good enough to allow for variation in the data? If the difference in the number of frames between the fastest and slowest bat in the MLBs is only a couple of frames because camera technology isn't fast enough, then a traditional mph radar gun reading might work better.
   126. Dizzypaco Posted: November 16, 2007 at 09:56 PM (#2617822)
Back in April, when Dustin Pedroia had a bad start, there were was one poster in particular who watched Pedroia against the Yankees, and announced based on the series that Pedroia was never going to be an effective hitter. This was a Yankee fan who is well respected around here (for good reasons), and if I remember correctly, slow bat was a factor along with other qualitative observations. My response that while I agreed that Pedroia was unlikely to be as good as he had looked in the minors (which I was wrong about), you can't tell anything by watching a handful of games when a player comes up. People often look like they have slow bats or holes in their swing when they are struggling.
   127. JPWF13 Posted: November 16, 2007 at 09:58 PM (#2617826)
Actually, why wouldn't you do this for a lot of scouting? Send a low wage/skill guy to videotape the hitters/pitchers, and let the true scouts stay at home analyzing. It would save them a lot of travel time.


I believe that's how much of football scouting is done
   128. Daryn Posted: November 16, 2007 at 10:01 PM (#2617829)
I look exactly like Billy Joe Armstrong of Green Day fame, only with more kempt hair.

Billy Joe Armstrong
   129. snapper (history's 42nd greatest monster) Posted: November 16, 2007 at 10:01 PM (#2617831)
If the difference in the number of frames between the fastest and slowest bat in the MLBs is only a couple of frames because camera technology isn't fast enough, then a traditional mph radar gun reading might work better.

The problem with this, is that it's probably not "bat speed" per se that we are talking about here (that's important for other things, like power). It's how fast the player reacts/gets the bat to where it needs to be. You can have a "long swing", or "slow bat" that eventually reaches high velocity, but it's too late.
   130. Swedish Chef Posted: November 16, 2007 at 10:05 PM (#2617837)
I think this is a cultural conflict between "craftsmen" and "engineers". The engineers are horrified by the lack of sound scientific basis in what the craftmen are doing, and the craftsmen can only point to their experience but not validate it by mathematical means foreign to them.

The worst trait of skeptics is preemptively dismissing things that they don't believe in. I don't really see any reason one couldn't learn to tell a bad swing from a good one. Of course it would be better to have a sound method for swing analysis than some mostly unconscious rules of thumb. But that kind of knowledge exists and works in all other fields of human endeavour, it would be mightily weird if baseball was different.
   131. JPWF13 Posted: November 16, 2007 at 10:07 PM (#2617840)
People often look like they have slow bats or holes in their swing when they are struggling.


and someone who never liked the prospect to begin with will say that it's obvious that the player will never amount to anything- if it's just a fan no big deal, if it's a manager or decision maker the poor guy is doomed.

Bil James had an interesting take on that WRT pitchers- when control/finesse guys struggle adjusting to the MLB they look like absolute crap- crap with no potential. When power pitchers struggle they still look bad- but the potential is still readily apparent. Hitters can be like that too, some guys still look like hitters when they are struggling- whereas some guys look like crap when they struggle-

I've noticed that guys who are regarded as streaky usually are not any streakier than "consistent" guys (not statistically anyway)- they are just guys who look really bad when making outs.

To a casual observer Darin Erstad looks like an MLB player, even going 0 for 4. Chris Shelton looks like total crap when he goes 0 for 3 with a walk.
   132. birdlives is one crazy ninja Posted: November 16, 2007 at 10:13 PM (#2617847)
The problem with this, is that it's probably not "bat speed" per se that we are talking about here (that's important for other things, like power). It's how fast the player reacts/gets the bat to where it needs to be. You can have a "long swing", or "slow bat" that eventually reaches high velocity, but it's too late.

Good point, thanks for clarifying.
   133. J. Cross Posted: November 16, 2007 at 10:31 PM (#2617875)
The problem with this, is that it's probably not "bat speed" per se that we are talking about here (that's important for other things, like power). It's how fast the player reacts/gets the bat to where it needs to be. You can have a "long swing", or "slow bat" that eventually reaches high velocity, but it's too late.

Good point, thanks for clarifying.


Yes, and this is where I think scout speak contributes to the problem by being sloppy. The difference in actual time that a high bat speed v. low bat speed bat spends in the hitting zone is something like 0.002 s. I'm highly highly skeptical that anyone no matter how well trained can perceive that difference. You can judge bat speed by the launch velocity it gives balls or possibly by sound off the bat but I simply don't think it's possible to see it. What scouts are seeing is whether a guy can be beaten by high or inside fastballs or whether he has a short swing or a long swing. Keith Law talks about "load' which I think is just how far back a hitters hands are when they start to move forward.

You can't help but notice that Milledge has a quicker bat than Gomez, his swing gets up to speed quicker and goes through the hitting zone in less time.

Sorry, I just don't think that this is something that it's even possible to notice never mind possible to miss.
   134. SoSHially Unacceptable Posted: November 16, 2007 at 10:37 PM (#2617883)
Back in April, when Dustin Pedroia had a bad start, there were was one poster in particular who watched Pedroia against the Yankees, and announced based on the series that Pedroia was never going to be an effective hitter.


I remember that. I'm thinking it was McNally (if not, it's probably more fun to pin it on him than SJ or RB).
   135. snapper (history's 42nd greatest monster) Posted: November 16, 2007 at 10:41 PM (#2617887)
What scouts are seeing is whether a guy can be beaten by high or inside fastballs or whether he has a short swing or a long swing. Keith Law talks about "load' which I think is just how far back a hitters hands are when they start to move forward.

But for this kind of stuff wouldn't it be much better to analyze videotape and get a quantifiable answer rather than a subjective impression?

Why don't teams send out video teams rather the scouts, and keep all the scouts in a controlled centralized group where they could all opine on every prospect and bounce ideas off each other, etc.?

I imagine someone could even start a service to videotape games with prospects of interest and sell the film to all the teams. Hell, baseball should start such a service itself.
   136. Backlasher Posted: November 16, 2007 at 10:52 PM (#2617900)
I give up. It's like I'm speaking Klingon.


That is where that $H stuff comes from? I didn't know it was some Star Trek claptrap.

Mike I think you are confusing where you stand among the pool of DIPS detractors in general. As far as I can tell most (as in a majority) have not read the new stuff, will not read the new stuff and are interested solely in repeating the assertion that, "Voros said pitchers have no control over BABIP- he's wrong" endlessly.


Utter nonsense, totally irrelevant and the last bastion of a dying argument. "Well, everybody who said he was wrong was correct. I'm just going to make up some justification for some non-existent group of them and say 'that's why I'm mad'" DIPS version anything is ill conceived in both concept and execution.

The reason there is still an issue is because of situations exactly like what you see in this thread. Someone is very wrong; they are wrong because they have a minimal understanding of the subject matter; and when they are shown to be wrong, they try to throw bad snark at the argument and the most dismissive of fear uncertainty and doubt at the situation. For instance, "any evidence at all" is ridiculous and its intended only for rhetorical effect. Its not even good rhetoric.

I think this is a cultural conflict between "craftsmen" and "engineers". The engineers are horrified by the lack of sound scientific basis in what the craftmen are doing, and the craftsmen can only point to their experience but not validate it by mathematical means foreign to them.


I agree with your second paragraph, but I quibble with this one. These are not "engineers" throwing FUD at the situation. This is more akin to a bad middle manager trying to get a seat at the executive table with craftsman and engineers.
   137. Zoppity Zoop Posted: November 16, 2007 at 10:59 PM (#2617911)
The reason there is still an issue is because of situations exactly like what you see in this thread. Someone is very wrong; they are wrong because they have a minimal understanding of the subject matter; and when they are shown to be wrong, they try to throw bad snark at the argument and the most dismissive of fear uncertainty and doubt at the situation. For instance, "any evidence at all" is ridiculous and its intended only for rhetorical effect. Its not even good rhetoric.


Ooo, ooo, Is this 20 questions? Can I start?

1. Is it the person who wrote post #136?
   138. BourbonSamurai Is a Lazy Nogoodnik Posted: November 16, 2007 at 10:59 PM (#2617912)
Are we still doing the celebrity thing? Some people tell me I look like the scientist in "short circuit," others tell me I look like the robot.

I call BS on this. Nobody will admit to having seen Short Circuit. You sir, are a dirty liar!


If you're referring to Austin Pendleton, which I think you are, he's a friend of ours. We went out to dinner with him once with another (non-theatre) friend, who at the end of the meal insisted on saying, "God, I loved you in Short Circuit!"

Despite the fact that Austin is a Tony-nominee who has been in many oscar-nominated films, he was very kind.

As far as celebrities, I get Kenneth Branagh/Seth Rogen, depending on how skinny/fat I am. I am skewing skinny at the moment, thank goodness.
   139. JPWF13 Posted: November 16, 2007 at 11:06 PM (#2617926)
Utter nonsense, totally irrelevant and the last bastion of a dying argument. "Well, everybody who said he was wrong was correct. I'm just going to make up some justification for some non-existent group of them and say 'that's why I'm mad'" DIPS version anything is ill conceived in both concept and execution.


Nice to see you BL, I was thinking of you actually when I in-artfully said that the Luddites and not the Emeighs comprise the majority of the anti-dippers.
   140. Backlasher Posted: November 16, 2007 at 11:14 PM (#2617933)
Nice to see you BL, I was thinking of you actually when I in-artfully said that the Luddites and not the Emeighs comprise the majority of the anti-dippers.


Good to see you too. I'm glad I'm in your hearts and minds. Unfortunately, I've had to read the DIPS junk b/c for awhile the argument was "The new DIPS has extra zing." Its just the same song with a new title. It still lacks utility, and its moved so far away from any original contention its meaningless.

Skilled observers are usually able to isolate when a pitcher is serving slop in the same way they can see a slow bat. Pitchers do have that skill and the population of pitchers where you are doing your $T's and klingon rituals have been pre-selected for that skill.
   141. JPWF13 Posted: November 16, 2007 at 11:24 PM (#2617945)
Are we still doing the celebrity thing? Some people tell me I look like the scientist in "short circuit," others tell me I look like the robot.

Pitchers do have that skill and the population of pitchers where you are doing your $T's and klingon rituals have been pre-selected for that skill.


Oh damn you're moving from the luddite anti-dip argument towards the Emeigh position (phrased a bit differently of course)

Your more fun when you're just snarking

Are we still doing the celebrity thing? Some people tell me I look like the scientist in "short circuit," others tell me I look like the robot.


Judging by his webcasts I look and sound like Will Carroll, which is why you will never EVER see me willingly be filmed and broadcast
   142. The District Attorney Posted: November 16, 2007 at 11:36 PM (#2617954)
Pitchers do have that skill and the population of pitchers [you're studying] have been pre-selected for that skill.
Okay, I can buy that, although it of course wouldn't be any reason not to use DIPS in reference to established MLB pitchers.

But what about the criticism that, if true, would both be more damning, and as #101 points out, would also be very weird considering that we've traditionally accepted it in other contexts... the criticism of (and it rolls right off the tongue, kids) "correlation between sequential time series points." Apparently "most of the pool of DIPS detractors on this site understand the issues with the method." Is this a situation where only registered members of the He-Man DIPS Haters Club get to find out what these issues are? Or do ya think you could tell the rest of us?
   143. J. Cross Posted: November 16, 2007 at 11:41 PM (#2617957)
Is this a situation where only registered members of the He-Man DIPS Haters Club get to find out what these issues are? Or do ya think you could tell the rest of us?

Nooooo!!!! The parties are here but surely no one want to get into this again! I know I could just not read the thread but it's like a train wreck once this gets going.

full disclosure: I'm firmly in the Voros camp on this one.
   144. CrosbyBird Posted: November 17, 2007 at 12:08 AM (#2617978)
You can't help but notice that Milledge has a quicker bat than Gomez, his swing gets up to speed quicker and goes through the hitting zone in less time. The difference is really visually dramatic- so dramatic that even knowing most fans and the MSM ignore minor league stats, I'm still amazed so many think Gomez is a better prospect.

I felt by watching Gomez that he was just outclassed by major league pitching. I felt like Milledge was a lot more likely not only to make contact with a hittable pitch, but crush it.

Gomez doesn't look as good as Milledge did at the age of 21 in the majors.

Gomez does look like a better fielder though. That might just be speed. Fast guys usually tend to look like better fielders.
   145. JPWF13 Posted: November 17, 2007 at 12:10 AM (#2617979)
Pitchers do have that skill and the population of pitchers [you're studying] have been pre-selected for that skill.


The argument goes (and others can make it better than me)

1: BABIP is a skill that pitchers have, those who are poor at it get weeded out climbing from A to AA to AAA. (just as pitchers who don't miss bats and pitchers who walk too many batters or give up too many homers are weeded out).

2: As a result the pitchers who make the MLB are at the high end insofar as the ability to prevent hits on balls in play. The variance between MLB pitchers in this skill is narrower than the combined variance between the defensive efficiency of the various teams and normal fluctuation (ie: a .300 hitter can easily bat .225 in 100 at bats or .350 in 100 at bats). The result is that simple studies make it appear that BABIP variations from year to year are random.

But still, why don't we readily see players who are true outliers?

1: There's a floor to how low a pitcher's "natural" BABIP can go- you have 9 fielders and a lot of ground to cover, no matter how good a pitcher is at inducing poor contact some balls will fall in. No matter how good a pitcher is at inducing poor contact some batters are still going to make good contact, guess right etc. No matter how good a pitcher is, he'll hang a curve ball every now and then.
2: Those who are really bad at BABIP, who will never see their BABIP regress to league average, don't last long enough to have their stats represent a reasonable sample for inclusion in a study. (or they will never get a chance in the MLB at all)


All this may be true and may be correct, but a problem I have with this view is that many anti-Dips advocates (Mike Emeigh and a few others excepted) START from the absolute conviction that BABIP is a skill and latch onto the selection argument for main reason that it's hard as hell to test.

Many of the same individuals reject MLEs using a variation of the same selection argument, they want a disprovable argument, because they do not believe that translating non-MLB stats to the MLB with ANY level of accuracy is conceptually (in their world view) possible. But you say, Ron Shandler's MLEs correlate [almost] as well with future perfomance as actual MLB stats do- the anti-MLEer doesn't care, and selection bias gives him an argument, "who cares, that's because scouts know who can hit and who can't, those who won't match their MLEs won't get a chance to play enough to be included in a study"

Sure Voros presents his arguments badly, and he comes off as arrogant, but he was on to something even if he tends to overstate things. Many anti-Dippers simply close their minds, eyes and ears, stick out their tongues and go Nyah Nyah Nyah pttph!
   146. The District Attorney Posted: November 17, 2007 at 12:21 AM (#2617980)
No, I understand the pre-selection argument (although that's a good summary that's valuable in and of itself). I'm asking about the idea that (wait for it... wait for it...) "correlation between sequential time series points" is itself invalid.
   147. Simpson Posted: November 17, 2007 at 12:44 AM (#2617993)
Voros, if you're still here and care...

I'm surprised that if you define approach = plan at the plate, you define it so strictly as "His plan is to hit the damn ball, preferrably hard, and if he's any good, his plan is to avoid swinging at pitches out of the zone." I think approach is something that differs not only from player to player, but from at-bat to at-bat, and certainly from being on a tear vs. being in a slump. Certainly David Eckstein has a different approach (or should) thant Puljols. Approach needs to change as game scenarios change, and also differs based on where you bat in the lineup. Consistently making the wrong change, not recognizing your skillset or role in the lineup, or simply refusing to change at all are symptoms of bad approach IMO. I would think quantifying these behaviors would be fairly easy.

Now I'm no scout, nor a practicing sabrdude, but I did once work for the Sky Sox in 2002 when Jack Cust had just come over from AZ, and he was constantly held up as someone with a bad approach, perhaps ironically because his approach more often than not was a carbon copy of your stiff definition. He was either swinging for the fences or drawing his walks. These are not bad things obviously, but he would infuriate both scouts and the press box folks by never altering that plan, that approach. There would be close games, when we were down a run in the late innings with runners on, when a single would put us ahead. Singles should be easier to hit that homers, but did Cust alter his approach? No, he'd strike out swinging for the fences, or walk to load the bases for an inferior hitter. In many people's eyes, his stubbornness was a big part of his bad approach. For other hitters, I imagine completely different reason, such as lack of confidence. Sometimes, I think bad approach is a way to describe the point where a player's skills have been tapped out--when a guy reaches the limit his talent will take him, he struggles, and his approach is questioned (assuming something like bat speed doesn't tell all you need to know).
   148. Simpson Posted: November 17, 2007 at 12:50 AM (#2617998)
Also, for the record:

I like the analogy of identifying hit songwriters.

I very much enjoy watching Arizona Fall League games.

I've never been told I look like anybody.
   149. BDC Posted: November 17, 2007 at 12:50 AM (#2617999)
"correlation between sequential time series points"

I believe that concept is roughly expressed in Klingon by pot'hbe'chug yay qatlh p"eghlu'.
   150. robinred Posted: November 17, 2007 at 12:50 AM (#2618000)
148. Simpson Posted:

I've never been told I look like anybody


Not even Abe or Homer? Or Marge? OJ?
   151. J. Cross Posted: November 17, 2007 at 12:52 AM (#2618003)
Isn't Jack Cust a great example of where scouts failed? A league of statheads would have had this guy in the majors much sooner despite his flaws. When finally given a chance, his "bad approach" didn't stop him from thriving at the plate at the major league level. No doubt, he still can't field but he didn't field in the minors either.
   152. CrosbyBird Posted: November 17, 2007 at 12:53 AM (#2618004)
I believe that concept is roughly expressed in Klingon by pot'hbe'chug yay qatlh p"eghlu'.

Wow. This would have been an interesting quote for that old thread where they intentionally walked that little leaguer to get to the kid with brain cancer.

No, I didn't remember what it meant and had to google it, but I did recognize it as real Klingon and not gibberish.
   153. The Most Interesting Man In The World Posted: November 17, 2007 at 12:57 AM (#2618006)
If I put enough gel in my hair, people think I'm San Francisco mayor Gavin Newsom.
   154. Simpson Posted: November 17, 2007 at 12:59 AM (#2618008)
See, I look at Cust as a guy that statheads missed too--he was supposed to be a potential star. I think his approach limits how good he can be. Sure, scouts missed in that he is talented enough to play regularly in the big leagues, but he's not going to be as good as many hoped just going on the stats he put up early in his minor league career. And at this point, at his age, how many more years like 07 can you reasonably forsee him having? Leaving defense out, I don't think you can definitively state that AZ and COL regret letting him leave their organizations.
   155. Simpson Posted: November 17, 2007 at 01:03 AM (#2618012)
I wish I had the $$ of those others Simpsons... but I will concede that when I was high school, people thought I looked a like Rick Smits from the Pacers. I was (and am) a tall skinny white dude.
   156. The Non-Catching Molina (sjs1959) Posted: November 17, 2007 at 01:16 AM (#2618027)
When I had hair, John Cusack.

Come back, Voros, it's getting good now...
   157. J. Cross Posted: November 17, 2007 at 01:21 AM (#2618036)
See, I look at Cust as a guy that statheads missed too--he was supposed to be a potential star. I think his approach limits how good he can be. Sure, scouts missed in that he is talented enough to play regularly in the big leagues, but he's not going to be as good as many hoped just going on the stats he put up early in his minor league career. And at this point, at his age, how many more years like 07 can you reasonably forsee him having? Leaving defense out, I don't think you can definitively state that AZ and COL regret letting him leave their organizations.

Well, I think he can be as good as hitter as many hoped. I'm not sure that it makes him super valuable given his lack of defensive ability but I do think he can really hit. And, sure, guys with his skill set might not age well. But, I think the point of debate between statheads/scouts re:Cust was whether he can hit at the major league level. Granted, 500 PA doesn't answer that question definitively but early returns are that he can.

Now, I'm not anti-scout. I pick my prospects (for my all important fantasy league, that is) based on what the scouts say more than the numbers but I think Cust is looking like one in the stathead column.
   158. philly Posted: November 17, 2007 at 01:49 AM (#2618061)
Cust is a good example of how limited the whole notion of players that scouts hate or like really is. Cust was a 1st rd draft pick out of HS. Whether or not you think that college stats are good predictors of eventual major league success, I'm pretty sure nobody would claim that as a stat based pick. But some scout in the AZ organization made Cust a millionaire at age 18. Come to think of it, that's the kind of "hate" I could use.

It's also true that even with his various limitations and things that scouts "hate" about him, he was very highly ranked in the BA Top 100 for a couple of years. I think he peaked in the mid to high 30s, I'm sure ahead of many "tools goofs" that scouts "love".

I agree that many of the random anonymous quotes from scouts about established professional players do not suggest a huge level of insight above and beyond good statistical measures. But I do think there is a huge amount we could learn about low level amatuer scouting and we just either don't have access to what we need or, in some cases, we have people who are so hung up on "measurable" that they don't even seem interested. Much to their loss.
   159. philly Posted: November 17, 2007 at 01:54 AM (#2618064)
Reaching way back to post #52:

Very nicely said, Russ. Yeah, I know I was sort of changing the subject. I would just really like to see that experiment done, because I'm curious about how much you really get for the tens of millions of dollars teams presently spend on their own scouting departments, as opposed to spending like $50,000 on three twentysomethings and a few books/magazines.

In other words, I personally agree with Voros on the general point that the structure of modern scouting departments is somewhat broken.


I would suggest that if you really think that teams spend "tens of millions of dollars" on their scouting departments that you are in absolutely no position to have an informed opinion about whether or not the structure of modern scouting departments is "broken", somewhat or otherwise.

I would be shocked if teams spent more than several million dollars on their scouting departments unless you're counting signing bonus money.
   160. Dan Szymborski Posted: November 17, 2007 at 02:44 AM (#2618092)
Sure, scouts missed in that he is talented enough to play regularly in the big leagues, but he's not going to be as good as many hoped just going on the stats he put up early in his minor league career.

Well, to be fair, his 2007 was actually better than any of his MLEs ever were - he played in some great hitting environments in the minors. Nobody ever said he was a God, but that he had done enough to earn a chance to play in the majors and show he could apply his talents over players who either demonstrated an inability to apply their talents in the majors or players who had done less in the minors to earn a shot.

We had sent a lot of vitriol towards the Orioles in 2002/2003 for not playing Cust, which is about when BL came up with the oh-so-clever "Jack Crust" nickname. That was a team, after all, that let Jose Morban play 14 games at DH.
   161. greenback likes millwall, they don't care Posted: November 17, 2007 at 03:04 AM (#2618103)
Re scouting: Has any study ever been done on the relationship between scouting reports and MLB performance, *after controlling for their statistics?*

Probably, but they are doing the work for a team and it will never be made public. People outside the game don't have enough access to scouting reports to do such a study.

Tangotiger's starting something with this that could yield results. You'd be counting on wisdom of the crowd, which isn't necessarily a bad thing.

As for taping games and sending them to analysts back at the home office, I've noticed a few minor league teams putting their offday pitchers to work to do just that. Well, technically the cameras are digital, in case that matters. Hire a few data entry grunts back home and you could build a nice index of your library as well.
   162. AROM Posted: November 17, 2007 at 03:16 AM (#2618110)
*for the purposes of projecting next year's performance* you are better off pretending that a pitcher has a league-average BABIP ability rather than trying to figure out whether his true talent BABIP is .287 or .293.


If you have an idea how to do it, you're better off figuring out the true talent. Its not a huge difference, but every bit helps.
   163. JPWF13 Posted: November 17, 2007 at 03:26 AM (#2618118)
but I did recognize it as real Klingon and not gibberish.


I think this is primey worthy
   164. AROM Posted: November 17, 2007 at 03:29 AM (#2618124)
You mean stick a radar gun on the bat?


Anyone else remember ESPN doing that on Sunday night baseball about 5 years ago? They'd pick a few players and report their bat speed a few times a game. The big hitters generally had their bat heads moving around 90 MPH.

Wonder why that didn't catch on? Maybe if you did that for several years and show it for every batter it would be a little too easy to see which players use methods to enhance their bat speed.
   165. Gaelan Posted: November 17, 2007 at 03:33 AM (#2618130)
The arrogance from the stats crowd is the refusal to acknowledge the limitations of statistics.

Everyone here knows the power of sample size. The larger the sample size the more reliable your statistics. It is because of sample size that things like MLE work. The problem with any sample is that it is based upon the assumption that for any sample that is drawn from events across time, the thing that is being measured (in this case ability) has not changed. In many cases this assumption turns out to be valid which is why statistics have any use at all. The problem is that there is no way of knowing, on the basis of statistics alone, whether this assumption is valid. Which is another way of saying that statistics are not self-grounding. It stands to reason then that an intelligent observer would want to ground those statistics in meaningful first hand observations.

For example there are many instances that we know, without a shadow of a doubt, that a player's abiltiy has changed. For instance:

1) when a player is young and developing their skills
2) when a player is old and losing skills
3) when a player is injured
4) when a pitcher learns a new pitch

In all of these cases the assumption that underlies statistical analysis is not reasonable. Circumstances have intruded and disrupted the sample size that gives statistics its power. Moreover if you are making a decision worth millions of dollars you don't have the luxury to wait and let the player build a new sample size. You need to make a decision now and the sample size you have is not reliable.

This should be obvious. It's well known that pitcher projections are much less reliable than the projections for hitters. The hard core stat bias is to say that this is due to luck. A more reasonable explanation is that the "true" ability that is being measured has actual variance (dead arm, hurt arm, new pitches, etc.). A smart man will send someone to watch that pitcher and not rely on unreliable projections.
   166. 8ball Posted: November 17, 2007 at 03:35 AM (#2618133)
Voros,

don't give up. Many people agree with you. I know I do.


Toby = Enabler.

If Mr. McCracken is going to take an extreme view and argue it in aggressive terms, then he shouldn't resort to "I'm just taking my ball and going home" tactics when people respond aggressively in kind.

But if he is, people shouldn't be begging him to return, so he can continue behaving like a petulant child.
   167. Mike Emeigh Posted: November 17, 2007 at 03:38 AM (#2618136)
*for the purposes of projecting next year's performance* you are better off pretending that a pitcher has a league-average BABIP ability rather than trying to figure out whether his true talent BABIP is .287 or .293.


The real problem comes at the other end - when you're looking at a pitcher with marginal major league ability and using a league-average hit prevention figure as his *real* level of ability. This leads you to overvalue the Glendon Rusches of the world.

-- MWE
   168. Jim Kaat on a hot Gene Roof Posted: November 17, 2007 at 03:41 AM (#2618139)
Gaelan, you're going to have to stop making so much sense. Heads are going to explode around here.
   169. 8ball Posted: November 17, 2007 at 03:42 AM (#2618142)
Sure Voros presents his arguments badly, and he comes off as arrogant, but he was on to something even if he tends to overstate things. Many anti-Dippers simply close their minds, eyes and ears, stick out their tongues and go Nyah Nyah Nyah pttph!

Don't you think that the badly presented arguments, the arrogance, and the tendency to overstate things contributes to the anti-Dippers responding in that way?

And don't you think it is likely that, if the conclusion McCracken initially reached hadn't been so overstated -- if it had simply been, pitchers have less control over whether BIPs go for hits than most people think -- that he, and his DIPS theory, wouldn't have gotten nearly the attention and notoriety they have?
   170. JPWF13 Posted: November 17, 2007 at 03:44 AM (#2618144)
For example there are many instances that we know, without a shadow of a doubt, that a player's abiltiy has changed. For instance:

1) when a player is young and developing their skills
2) when a player is old and losing skills
3) when a player is injured
4) when a pitcher learns a new pitch

In all of these cases the assumption that underlies statistical analysis is not reasonable. Circumstances have intruded and disrupted the sample size that gives statistics its power. Moreover if you are making a decision worth millions of dollars you don't have the luxury to wait and let the player build a new sample size. You need to make a decision now and the sample size you have is not reliable.


The problem from the decision makers POV is that many/most people when shifting from the objective POV (stats based) to the subjective let personal biases (wishcasting) take over. For example a young player who's shown nothing before (statswise) may have a hot week and the scout/decision maker who has always loved that player's potential may jump a little too quickly to the "1) when a player is young and developing their skills" scenario.

Not to sound like one of the BPro blowhards, but both the objective and the subjective should be kept in mind.
   171. BDC Posted: November 17, 2007 at 03:55 AM (#2618150)
A smart man will send someone to watch that pitcher

And a smart man or woman will always have to send someone to watch fielders. The strike zone is essentially a game: you don't have to have the greatest bat speed or arm speed to win the game; much of it involves fooling the other person, and reacting to how he's trying to fool you. Therefore a guy without great tools will sometimes put up excellent stats, and could do that at any level of the game. It's important to pay attention to the results of the game, no matter what your observation tells you about bat speed.

Fielding, however, is an athletic performance. The baseball is not trying to outsmart the fielder (the baseball would have a good chance of outsmarting some of the Rangers' outfielders, over the years, but fortunately it's not trying to do that :) You can quantify athletic performance, of course, but the more complicated the performance, the more uncertain the quantification (the winner of a 100m dash is the fastest runner; the winner of a decathlon is the compiler of the best arbitrary point total; the winner of a gymnastics competition is the compiler of the best arbitrary and subjective point total). And drawing conclusions from defensive statistics is extremely murky, for the reasons I've just mentioned and also because sample sizes are extremely low, and because opportunities and the difficulty of chances vary so widely within them.

Edit: speaking in the best BPro Blowhard fashion, of course :)
   172. snapper (history's 42nd greatest monster) Posted: November 17, 2007 at 04:28 AM (#2618177)
The arrogance from the stats crowd is the refusal to acknowledge the limitations of statistics.

I think the key is to go beyond stats, but stay with the quantifiable, or at least objectively measutable.

Things like video analysis of a pitcher's motion (a la CBW), or of bat speed, would seem to take the subjectivity out of scouting, while still getting at raw ability and development in a way that raw stats can't.
   173. Perro(s) Posted: November 17, 2007 at 04:31 AM (#2618180)
Video killed the radar-wielding scout.
   174. Mike Emeigh Posted: November 17, 2007 at 04:50 AM (#2618196)
Therefore a guy without great tools will sometimes put up excellent stats, and could do that at any level of the game.


True.

It's important to pay attention to the results of the game, no matter what your observation tells you about bat speed.


This doesn't necessarily follow, however.

What a stat line shows is that a player is capable of putting up that performance. What you don't know, just by looking at the stat line, is whether that stat line represents the absolute top of his range or something less than that. If the stat line represents the absolute top of his range, then that's the BEST he's ever going to do.

What the scout tries to determine is whether, and to what extent, the player has room to grow: he's looking for whatever visual cues he can find that will tell him "this guy is likely to get a lot better" or "this guy is already as good as he is going to get". Traditional statistical analysis doesn't tell you where a player fits on the development curve.

The Braves selected Cody Johnson as their first draft choice (24th overall) a year ago. Scouts were divided on him, but the consensus was that he was an overdraft. CBW wrote:

When I first reviewed him I thought he was a reach in the 1st round. After watching the video more closely, I think he’s a reach in the 10th round. Watching it today, his swing is just not good. He has a complete disconnect between his hands and body. Watching it with my swing guru buddy/cousin today, he expressed sheer disgust as we watched Johnson’s swing frame-by-frame. Most Sunday hackers slice because they come over the top of the ball by yanking the hands from the outside in, instead of turning your body quickly with the hands following until ready to unleash hell. My guess is Johnson, if he is a golfer, either has a severe slice or a pull-hook. I guess he has quick enough hands and is a good athlete, but, uh, not the right pick.


He was awful in rookie ball in '06, but was outstanding at Danville this year, popping 17 HRs and posting a 1.004 OPS while winning MiLB.com's Short-Season Player of the Year award. Which is more representative of his future growth? Did he make some swing adjustments, was he taking advantage of lesser pitchers in the Appy, or what? Where does he fit?

Mitch Einertson, as an 18-YO rookie in 2004, tore apart the same Appalachian League that Johnson flashed through this year, posting an 1105 OPS with 24 HR in 63 games. He actually got a mention in Baseball Prospectus - a rarety for a short-season player, as they admitted. He then spent two seasons at Lexington, neither of which were very good, and then hit well in Salem this year, winning an MVP award - but Salem's a decent hitting environment. In those three seasons he's hit a total of 30 HRs - although he did post 40 doubles this year. Where does he fit on the development curve?

I don't have any idea. More to the point, I don't think any statistical model can project either player forward with any reasonable degree of confidence, because the data points are so widely variant and the statistical models have no idea which of those data points (if any!) are most relevant to those two players.

-- MWE
   175. Honkie Kong Posted: November 17, 2007 at 04:57 AM (#2618200)
He was awful in rookie ball in '06, but was outstanding at Danville this year, popping 17 HRs and posting a 1.004 OPS while winning MiLB.com's Short-Season Player of the Year award. Which is more representative of his future growth? Did he make some swing adjustments, was he taking advantage of lesser pitchers in the Appy, or what? Where does he fit?

The raw stats may not be the answer here right? If you look at how many players in rookie leagues have a K rate of 25%+ and slugged 500+, and then seen what success they have, you might get a bigger set of comparables? If you are using stats, anythigng a scout can see on the field can be broken down into statistical components. Without looking at Cody Johnson ever, just looking at his stats, won't some statisticians be able to say he has an ugly swing?
   176. Toby Posted: November 17, 2007 at 05:06 AM (#2618208)
166:

I have scouted Voros and he has great stuff. His command is really iffy, but he can really bring the heat. His big problem is his "approach". By that I mean he wants to come high and inside all the time and strike everybody out. He challenges everybody and gets wild and out of synch. If we can get him to dial it back a little, less throwing, pitch to contact, stay within himself, that sort of thing, I think he can be a pretty reliable starter in this league.
   177. Mike Emeigh Posted: November 17, 2007 at 05:15 AM (#2618212)
The raw stats may not be the answer here right?


Right.

If you look at how many players in rookie leagues have a K rate of 25%+ and slugged 500+, and then seen what success they have, you might get a bigger set of comparables?


You might. Then again, you might not - especially when you factor Johnson's 2006 season into account as well, plus his age (there aren't that many 18-YOs in advanced rookie ball).

Without looking at Cody Johnson ever, just looking at his stats, won't some statisticians be able to say he has an ugly swing?


With any degree of confidence?

-- MWE
   178. 8ball Posted: November 17, 2007 at 05:25 AM (#2618216)
#176 --

Best post of the thread. Well played.
   179. G.W.O. Posted: November 17, 2007 at 10:06 AM (#2618304)
The District Attorney Posted: November 16, 2007 at 06:21 PM (#2617980)
No, I understand the pre-selection argument (although that's a good summary that's valuable in and of itself). I'm asking about the idea that (wait for it... wait for it...) "correlation between sequential time series points" is itself invalid.


This is a slightly hand wavy argument, and skips some complications, but I think it captures the general idea. CAUTION: Contains sums:

Suppose I've got two random number generators. One generates numbers from a standard normal distribution (mean 0, std deviation 1, i.e N[0,1]) and the other is slightly biased (mean e, sd 1, i.e. e+N[0,1]). I'm trying to detect if that bias exists, so I generate a series of M numbers from my biased generator. As my first test, I want to find out whether "0" is a good predictor for the mean of that generator, so I calculate the mean squared deviations of my time series. Now, by expected value of this is:

E[(e + N[0,1])^2] = E(e^2 + 2 e N[0,1] + N[0,1]^2) = e^2 + 1,

because the expecation value of the square of a std normal variable is equal to the variance, sigma^2, i.e. 1.

Now, as a contrast, I consider using the previous value as a predictor and calculate the mean squared error again. My expected value of this is

E[{(e+N[0,1]) - (e+N[0,1])}^2]

Now the 'e^2's cancel, which is good, but those N[0,1] are independent (i.d.) random variables so they don't cancel and what I get is

E[- 2 N[0,1] N[0,1] + N[0,1]^2 + N[0,1]^2] = 2.

So, I've eliminated the 'e's, at the cost of an additional sigma^2. So, if the bias I'm trying to detect is small, in comparison to the natural variability of the unbiased data, using the previous biased figure as a predictor is bound to be worse than using the unbiased mean, because the biased figure introduces an additional term of statistical noise.

Conclusion: The sum (or difference) of two random variables used in autocorrelation has a naturally greater variation than the difference between one random variable and a constant.

EDIT: one equation for "clarity"
   180. GGC don't think it can get longer than a novella Posted: November 17, 2007 at 02:24 PM (#2618326)
G.W.O., is that what Bill James was said in "Underestimating the Fog"? I had totally forgot about that essay when I posted #101.
   181. Scoriano Flitcraft Posted: November 17, 2007 at 05:03 PM (#2618399)
Calling that which you disagree with BS or don't think has yet to be proven or disproven is not science. It is, however, a form of arrogance and unbecoming to a serious scientist.

Of course, the notion that sabermagicians stand in the shoes of scientists is a bit much, too.
   182. canoe Posted: November 17, 2007 at 05:58 PM (#2618430)
I checked the “Hall of Merit” threads seeking but not finding an argument like this:

“I’m putting Will Clark at the top of my ballot – see link to video below. Look at that bat speed! Check out the position of his hands at contact, that stiff front side…what beautiful weight shift. And talk about a good approach! He was the best!”

You want an unprejudiced demonstration that a guy can hit? You go to the statistics, not the video, not to what some old timer thinks. Johnny Damon has an uglier swing than 90% of the players in the NCAA, but who cares? Hitting stats provide wonderful detail based on hundreds if not thousands of opportunities (rookie ball, Class A, wherever) for success or failure. And people who do personnel evaluation for a living have a little saying, “Past performance is the best indicator of future success.”

On the other hand, there is fielding and baserunning, for which the reliability of low minor league statistics is highly questionable. I think traditional scouts can add more value there. Remember in Gladwell’s “Blink,” where he describes the unusual and uncanny abilities of Vic Braden to instantaneously recognize when a 120 MPH tennis serve will be a fault? It is an automatic and almost unconscious reaction, based on the fact that Braden has watched several hours of tennis almost every day for 50 years. When a team sends a Birdie Tebbetts out to scout, they are relying on that kind of assimilated knowledge (unquantifiable, unconscious) from 10,000 hours spent watching the game. A traditional scout can provide good qualitative input on this basis for difficult-to-measure components of performance like fielding.

Caution is needed, since scouts can have significant biases that petrify with age. I don’t think I’d want Bob Feller evaluating pitchers down in San Pedro De Macoris.
   183. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 17, 2007 at 06:28 PM (#2618455)
Mike E--The Glendon RuschES of the world? There is only one Glendon Rusch, my friend. He's the exception that proves the rule. For a minor leaguer who is coming up, you absolutely do want to look at his minor league hit prevention numbers to see whether he has MLB BABIP ability or not. But for a guy with a MLB carer of any length, you have an 0.0001% chance of running into Barry Zito, an 0.0001% chance of running into Glendon Rusch, and for everyone else, the stdev of the ability is so small that your projection won't be off by much if you use league average. The study I did for my NY Times piece on the subject found *less* variation in BABIP over the last 6 years than one would expect from flipping a coin--I can send you the data if you'd like.
   184. Dan Szymborski Posted: November 17, 2007 at 07:02 PM (#2618487)
It's also worth noting that outside of the freak season, Rusch's career ERA was only like a quarter of a run worse than his career DIPS ERA.
   185. Dag Nabbit: secretary of the World Banana Forum Posted: November 18, 2007 at 04:15 AM (#2618847)
It's also worth noting that outside of the freak season, Rusch's career ERA was only like a quarter of a run worse than his career DIPS ERA.

Szym, I have no idea whatsoever if you're referring to 2001 or 2003 when you say "the freak season." Both were freakishly high. Heck, you could mean 2005 for all I know. I remember figuring Hit Delta for every pitcher since 1986 with at least 150 IP in a season (this was before finding out prospectus already did the work on their own). Two of the three most extreme were Glendon Rusch seasons. Neat, huh? So I really, truly have no idea what you mean when you refer to freak season in the singular. (FWIW, the other really high season was Jaime Navarro).

Mike E--The Glendon RuschES of the world? There is only one Glendon Rusch, my friend.

No one that extreme, but there are others. Just ask Andy Pettitte. Or Jim Kaat. Or Tommy John.
   186. Voros McCracken of Pinkus Posted: November 18, 2007 at 04:43 AM (#2618864)
Why does everything I say have to be about DIPS? It's been eight years. A less interesting subject to me does not exist.

It's frustrating when I try and explain just how many DIPS studies I actually did, and then two posts later I get an argument saying that the only study I ever did was flawed.

Is it that you don't bother to read what I'm writing, or you don't believe me?

That's what's so frustrating.

"Why did you do A?"

"I didn't do A?"

"You shouldn't have done A?"

"I just told you, I didn't do A?"

"Doing A was wrong, you shouldn't have done A?"


Do you have any idea what having to deal with this is like? Yes I got up and left. Any sane person would. Eight years of complete strangers hating me over pitching statistics. Why do you care so damn much?
   187. GuyM Posted: November 18, 2007 at 05:02 AM (#2618878)
The study I did for my NY Times piece on the subject found *less* variation in BABIP over the last 6 years than one would expect from flipping a coin

Well, in that case there's likely something wrong with the study. The true talent SD for BABIP, among MLB pitchers, is around .009. That doesn't sound like a lot, but it means about .20 in ERA. So a good BABIP pitcher (+1 SD) has about a 0.40 ERA edge over a poor one (-1 SD). That's pretty substantial, similar (I think) to the spread of talent for BB/9, less than for Ks and HRs. Identifying which pitchers are good early in their career is indeed a challenge. Using both their minor and major league data should help increase sample size.

There's virtually no chance that the variance over six seasons would be less than the binomial variation.
   188. Voros McCracken of Pinkus Posted: November 18, 2007 at 05:15 AM (#2618885)
"However, I don't think you're wrong here. If the league average totals are adjusted to represent
league average totals for GB, FB, LH, RH, SP and RP we might gain some accuracy in the DIP numbers. Also to night is that, anecdotally, I believe pitchers with trick deliveries (e.g. Knuckleballers) might post consistently lower $H numbers than other pitchers. I looked at Tim Wakefield's career and that seems to bear out slightly... I believe that GB/FB adjustments to league average figures would make the system more accurate."

11/9/99

"The key from here is to refine the DIP measures (assuming now that they are a valuable tool) to be more accurate. We can assign different "league average" values for various pitchers depending on whether they are left handed or right handed, starter or reliever, groundball pitcher or flyball pitcher and other designations which could cause subtle shifts in the overall evaluations."

12/7/99

"There may, however, be instances where a hits allowed total may be representative of a special
ability of a certain pitcher, and the best candidates there would be trick delivery pitchers like knuckleballers and sidearmers. STill more work needs to be done there."

2/29/00

All more than a year prior to the Prospectus article. If I ever said pitchers had ZERO ability in this stat, it was clearly an accident since I've been saying quite the contrary since the very start of this.

But as I mentioned before, I'm sure the above will fail to sink in and I'll have to do this all over again.

This is my point in a nutshell, I'm forced to clairfy my point of view on DIPS over and over and over again, and when I ask a scout for an explanation once, I'm now an arrogant jerk,

So pardon me if I'm a little testy.
   189. The District Attorney Posted: November 18, 2007 at 11:24 PM (#2619300)
Okay, I have absolutely no idea what #179 meant. If James wrote about it, maybe that'll explain it in a way I can understand. Yes, I am not the least bit mathematically inclined. (OTOH, I'm not sure that I should have to be in order to understand points about baseball.)
   190. DCA Posted: November 19, 2007 at 12:46 AM (#2619363)
I don't really understand 179 either. And I understand statistics pretty well. I can follow the equations but I don't see how this is relevant to the question at hand -- we are not trying to detect any individual's bias, just whether or not all individuals have the same true talent, and that is not an autocorrelation problem.

Here's a simplification of the correlation problem: 100 pitchers, with true BABIP talent distributed normally with a mean of .300 and a standard deviation of X. For each pitcher, BABIP in a particular season is distributed as normally with a mean of "true BABIP" and a standard deviation of .020

For X = 0 (the strong version of the DIPS hypothesis), the 95% percentile of observed correlation is about 0.17 -- I didn't calculate it exactly, I'm running a simulation with 1,000 repetitions.

Now I set to three conditions, X = .010 variation in BABIP true talent smaller than annual sampling variability, X = .020 equal to annual sampling variability, X = .040 greater than annual sampling variability. The power of the "Voros test" -- observing a correlation above 0.17, what is likely by random chance if there is no variability in true talent -- for each condition is:

X = 0.010; true correlation = .20; power = 65%
X = 0.020; true correlation = .50; power = 100%
X = 0.040; true correlation = .80; power = 100%

The real-life set of subjects will be more complicated, and will probably tend to reduce power, but the correlation between year N and year N-1 data points for a sample of 100 would almost always be able to identify an effect size equal to or greater than the natural variability of the statistic within a season due to sample size. Even if the effect size is only half the natural variability of the statistic, using the correlation as a test statistic is likely to find it.

With a sample size of only 50 pitchers, the 95th percentile under the null hypothesis is 0.24, and the power for each effect size is:

X = 0.010; power = 42%
X = 0.020; power = 99%
X = 0.040; power = 100%
   191. John (You Can Call Me Grandma) Murphy Posted: November 19, 2007 at 01:24 AM (#2619386)
This is a fun game. We could have a whole thread stating what celebrities people compare us too.


I have been compared to Greg Brady, Tom Hulce and HOFer Buck Ewing. IOW, I'm Rollin Hand from the IMF.
   192. John (You Can Call Me Grandma) Murphy Posted: November 19, 2007 at 01:26 AM (#2619387)
Johnny Damon has an uglier swing than 90% of the players in the NCAA, but who cares?


Lou Gehrig's swing was compared to a rusty gate.
   193. GGC don't think it can get longer than a novella Posted: November 19, 2007 at 01:47 AM (#2619392)
I have been compared to Greg Brady, Tom Hulce and HOFer Buck Ewing. IOW, I'm Rollin Hand from the IMF.


If you've come across someone who remembers Buck, I'm impressed. I usually confuse him with Buck Weaver.
   194. John (You Can Call Me Grandma) Murphy Posted: November 19, 2007 at 02:16 AM (#2619416)
If you've come across someone who remembers Buck, I'm impressed. I usually confuse him with Buck Weaver.


Heh. Craig Burley was the one who pinned Ewing on me recently. Some of the lithographs of him don't look too bad, but his actual photo ain't so hot. I'd rather accept comparisons to Greg Brady (that was when I was a teenager, BTW) instead. :-)
   195. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 19, 2007 at 02:42 AM (#2619440)
GuyM--how are you calculating true talent? My study used MGL's PZR data, which shows how many hits a pitcher would have allowed with average fielders, taking defense out of the equation and focusing in on pitcher hittability. I can email you the data if you'd like.
   196. Dan Szymborski Posted: November 19, 2007 at 02:53 AM (#2619454)
Wow, Buck Ewing is an incredible match for Murph.
   197. Chris Dial Posted: November 19, 2007 at 02:56 AM (#2619457)
All more than a year prior to the Prospectus article.

All hail USENET.
   198. GuyM Posted: November 19, 2007 at 03:21 AM (#2619485)
Dan:
I was using Tango's data comparing pitchers' career BABIP to that of their teammates: http://www.insidethebook.com/ee/index.php/site/comments/career_dips_numbers/ While not a perfect control for park and fielding, I think it's pretty good. The SD for the gap between a pitcher and teammates is .0144, and when you reverse-engineer for the true talent SD, given these sample sizes, you get .0085.

I'll have to think about using the PZR estimates, rather than actual outcomes, in this way. Seems to me that it may in a sense be removing some of the random variance.
   199. J. Lowenstein Apathy Club Posted: November 19, 2007 at 03:30 AM (#2619493)
Wow, Buck Ewing is an incredible match for Murph.

See John? I told you!
   200. David Concepcion de la Desviacion Estandar (Dan R) Posted: November 19, 2007 at 03:34 AM (#2619497)
I do not think comparing BABIP to teammates is a good approach when PBP data is available. First, it assumes that the rest of the pitching staff has neutral BABIP ability, which may not be true for teams like the 90s Braves, but that's a minor concern. The big issue is that a pitcher's BIP distribution is heavily influenced by his handedness and GB/FB tendency, and so teams that have excellent fielders at some positions and poor fielders at others will likely have a high stdev of pitchers' BABIP even if the pitchers' actual BABIP ability is equal. I could see a case for using pitchers' BABIP compared to teammates of the same handedness and GB/FB tendency, but that gets you into worthlessly small sample sizes.
Page 2 of 3 pages  < 1 2 3 > 

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
robinred
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogOTP April 2014: BurstNET Sued for Not Making Equipment Lease Payments
(1748 - 11:41am, Apr 21)
Last: Rickey! In a van on 95 south...

NewsblogMorosi: MLB must evolve to let players express themselves without rebuke
(23 - 11:39am, Apr 21)
Last: SoSHially Unacceptable

NewsblogPrimer Dugout (and link of the day) 4-21-2014
(30 - 11:39am, Apr 21)
Last: BDC

NewsblogDoug Glanville: I Was Racially Profiled in My Own Driveway
(423 - 11:38am, Apr 21)
Last: Jolly Old St. Nick Still Gags in October

NewsblogJ.R. Gamble: Albert Pujols' 500-Homer Chase Is A Bore, But That's Baseball's Fault
(7 - 11:38am, Apr 21)
Last: Avoid running at all times.-S. Paige

NewsblogA’s Jed Lowrie “flabbergasted” by Astros’ response to bunt
(26 - 11:37am, Apr 21)
Last: theboyqueen

NewsblogVIDEO: Brewers, Pirates brawl after Carlos Gomez triple
(122 - 11:34am, Apr 21)
Last: Sunday silence

NewsblogOMNICHATTER for APRIL 21, 2014
(7 - 11:34am, Apr 21)
Last: CFBF Is A Golden Spider Duck

NewsblogOT: NBA Monthly Thread - April 2014
(385 - 11:33am, Apr 21)
Last: King Mekong

NewsblogJohn Torres: Baseball must bag sickening farewell tours
(51 - 11:33am, Apr 21)
Last: Traderdave

NewsblogBryce Harper benched for 'lack of hustle' despite quad injury
(110 - 11:29am, Apr 21)
Last: bunyon

NewsblogIvan Nova’s season in jeopardy after tearing elbow ligament
(13 - 11:22am, Apr 21)
Last: bunyon

NewsblogOT: The Soccer Thread March, 2014
(964 - 11:12am, Apr 21)
Last: ursus arctos

NewsblogMinuteman News Center: Giandurco: This means WAR
(97 - 11:08am, Apr 21)
Last: tshipman

NewsblogGleeman: Mets minor league team is hosting “Seinfeld night”
(162 - 11:00am, Apr 21)
Last: You Know Nothing JT Snow (YR)

Demarini, Easton and TPX Baseball Bats

 

 

 

 

Page rendered in 1.3154 seconds
52 querie(s) executed