THE BOOK--Playing The Percentages In Baseball

It should be noted that Bradbury’s long awaited academic paper on his work with the sabermetric community is now available:

http://jse.sagepub.com/cgi/content/abstract/11/1/29

Costs $25 to read past the first paragraph of JC’s piece.

I really liked what Phil had to say. Tango has spoken often about being a ‘subject matter expert’. I’ve spent a few months wring MySQL code for the Oliver projections. Analyzing pbp for batting, pitching, fielding, stealing, on and on. I formulate a model in my head, write the code, look at the results and determine of they make any sense, how well do they conform to precedent, what implications do they bring up. Think about it some more, maybe make some adjustments. I need to know the output makes some sense. Then Dave Gassko sent me a formula to put in the code. I just didn’t know what the output was supposed to look like, so I really didn’t know if I did it correctly or not. I had to send it back and wait for his confirmation.

Colin/1: I think that’s the journal I have a print subscription to ... my copy will be arriving in the mail soon, I suppose.

Phil,

Don’t get too excited.  They attack you personally.

My favorite line is this:

“Kruger and Dunning (1999) find that many nonexperts tend to overestimate their abilities. ‘‘Not only do these people reach erroneous conclusions and make unfortunate choices, but their incompetence robs them of the metacognitive ability to realize it.’’”

Ironically, an economist is a non-expert in sports.  I wonder if the authors understand that.

In general, this paper is a jumbled mess written with the kind of condescending and patronizing tone that might have been reserved for talking to a misbehaving five-year-old two generations ago.  There are some really odd comments:

“Often the merit of ideas offered by this community is judged by a consensus of pseudonymous avatars, many of whom appear to lack training and experience with advanced research methods.”

“The metricians do not provide the same peer review that academic journals do. Consequently, just because a work is accepted among the metricians, it does not necessarily mean this
work would pass muster in the academic community.”

“When nonacademics discover their ideas are not being accepted, unpleasant behavior can result.”

There’s a part of +/- that’s total garbage:

“So, it appears that plus–minus is not a particularly powerful metric in hockey.” (Because year-to-year R^2 is low for individual players.  What b.s.)

“Turning to plus-minus, how many points are scored and surrendered when a player is on the court should be linked to current wins. However, efforts to separate the player from his teammates have not proven successful. For most players, the
adjusted plus–minus approach argues that the player did not have a statistically significant impact on outcomes. Second, adjusted plus–minus cannot predict the future very well. Consequently, despite the popularity of these approaches among the metricians, neither is an improvement over what has been published in academic journals.”

Hawerchuk/4: I’m not a big basketball expert.  Are the last two paragraphs false?  That is, does plus/minus in basketball still have too much random chance in it to be useful?

Last week, on Hawerchuk’s blog, Millsy said this:

Having read the article (now available at SAGE’s JSE page), I can say that the paper is very fair and claims much positive benefit from online communication. There are some explanations of how economists have fallen short in the metrics they use, as well as cautions as to ensuring academic validity of work developed outside academia, if for no other reason than to confirm it’s applicability within a paper to those unfamiliar with it. You can make whatever judgement you’d like.

I do think that there is some valid concern over accountability for comments and criticisms. There is a lot of power in one’s name. I don’t think you, Tango, fall into that category, given yours is pretty easy to decipher, and your book is prominently posted on your website.

I had pre-responded to him:

As for his upcoming paper, I can guess what’s going to be in there: posters with names not on their birth certificates. That we’re “shielded” or whatever, and benefit from the anonymity without the cost of exposing ourselves. And whatever other bullsh!t he wants to say. It’s easy to guess what’s in his paper, because at this point, he’s a 2-D online caricature of himself.

If I’m wrong, I’ll make sure to post so in a new thread on my blog.

Based on Gabe’s selective quotes, I think I may have been too kind in what to expect from their paper.

***

By the way, “Tom Tango” is an alias, as most of the regulars here know.  Not that it actually matters what actually is and is not on my birth certificate.  Every baseball site I’ve ever posted to (and that’s got to be upwards of 100, if not 200) has some form of “Tango” or “Tangotiger” in it.  I’m accountable for what I say. What matters is that I have ONE handle.

plus/minus, unadjusted, is a tough stat to accept on its face.  One you start to adjust, it makes more sense.

Basically, it’s like a pitcher’s won/lost record.  Greinke being 16-8 is more (or at least as) impressive than Felix going 19-5.  One is +4 wins, and the other is +7 wins.  Adjustments can turn Greinke into +6 and Felix into +5, or whatever.

Same thing applies to hockey or baseketball plus/minus: you simply need to be aware of the identities of all the other players on the field of play.

Whoever was Bobby Orr’s lucky defense partner when he was +124 (!!!) obviously is very aware of one man’s effect.

Phil @5 - depends how you define it.  I think Berri is still grinding an axe against Roland Beech, and he is just trying to show that the R^2 of some system he came up with is 0.15 with future wins while the R^2 of someone else’s system is 0.12, and therefore the other system should be rejected.  (This is an actual argument Berri made.)

Roland discusses his methods obliquely in this interview:

http://www.slamonline.com/online/nba/2009/03/changing-the-stat-quo/

But I think the basic point is this: why wouldn’t plus/minus be a useful stat?  It doesn’t need to be points in the NBA, it could be more micro events, like getting beat off the dribble.  Roland counts up all of the little events when he charts a game and finds massive utility from it.

Ironically, I think that Berri and Bradbury would call Beech a non-expert, even though he works for the Dallas Mavericks full-time and has a business doing game charting, while they would consider themselves experts.

Right, it depends, but I’m just trying to get a sense of how well the thing works, even without numbers.  Obviously, it will work perfectly if you have enough data, but how well does it work in real life?

Let’s say one of your subs, playing 12 minutes a game, has an adjusted +/- of 8 points a game (I assume the adjustment is for his teammates somehow).  How reliable is that?  Is it likely that he’s that good?  I guess I *could* use a number ... if you regress that to next year’s, how big is the r?  What’s a decent confidence interval for the +8?

Or just a gut description.  Is a +8 “almost always” going to be a good player?  Sometimes?  So insignificantly different from 0 that you can’t tell anything?

Or something like, “a +8 is like player of the week in baseball.  Probably a good player who had a lucky week, but sometimes an average player who had a REALLY lucky week.”

Anything will help, I really don’t know the answer to this question.

I love how Bradbury refers to “Solving DIPS” as “a measurement debate.” Yes, that’s what it was. A debate about measurement.

Which is I think where the real divide between sports economists and metricians lies. Bradbury and Berri cite this remark approvingly:

“It is incumbent upon sports economists to explain why their research can be generalized beyond the sports industry, something at which they have become increasingly adept.”

And you know what? The conclusions of DIPS itself have no general application outside the sports industry. They’re not even useful in other sports, like basketball.

So it’s two competing purposes - academic economists are looking to use sports as a vehicle to make broader points about economics. Sabermetricians are looking to use economic and econometric ideas (among others) to answer questions about baseball.

These aren’t the same thing, and I don’t think it’s useful to act like they’re the same thing. If the goal is to look at competative labor markets as a function of marginal revenue, maybe you don’t care about DIPS - it’s a “measurement debate” where “sports economists prefer a
parsimonious metric to a more complex measure”. But if you want to know if pitcher A is better than pitcher B, then DIPS is hugely important.

I don’t really know how well the best points-based NBA +/- systems work.  I meant that you could define +/- in terms of one-on-one matchups or pretty much anything else that you wanted.  (Vic Ferrari has great examples of the predictive value of shot volume +/- and scoring chances +/-.) The assumption in the paper is that you have to stick with boxscore stats as currently collected. 

I guess I’m just surprised that a researcher would write off an entire area of study because the present state of the work there is unsatisfactory.  You’d think that would be an opportunity to make some headway…

Just subscribing to this thread because I love these things.

My favourite part is definitely the following:

“A growing trend in the transfer of ideas among academics is the publication of regularly updated Weblogs, more commonly known as ‘blogs.’”

Kids these days, with their rock and roll hair. . .

That said, I think in general this paper is complimentary - JC only offers examples of sabermetrician successes.  Even though the tone is somewhat dismissive, he at least acknowledges that most of the important theories in sabermetrics were developed by amateurs.

He’s a bit of a dick, but in content (as opposed to the feel) gets it right.

I tend to agree with Ian/13 that Bradbury’s examples of interaction with sabermetrics are much more positive than Berri’s examples of interaction with APBRmetrics.  I don’t know enough about basketball analysis to judge whether Berri has a firm foundation for his criticisms.

The paper does reflect one fallacy that I’ve seen from Bradbury before:

Birnbaum (2006) considers sabermetricians to be ‘‘no less intelligent than academic economists’’ and superior to economists in their understanding of baseball. This statement reveals a curious worldview. On one hand, the aspect that is universal across both groups—members of both communities have been devoted sports fans since an early age—is considered unique to the nonacademic sports analysts. On the other hand, when it comes to the aspect that is unique to academics—academia normally involves many years of advanced training and requires its participants to be judged competent by their peers in a ‘‘publish or perish’’ environment—metricians demand equal recognition. In our view, this mentality begets misplaced confidence.

Hobbyists (in general) are known for being trained experts at their hobby, often to a degree as great as the most dedicated academic.  However, Bradbury is mistaken about the hobby involved here.  The hobby is not sports event viewing or even dedicated following of the news on the local team. I know many people who are lifelong sports fans who know next to nothing about analyzing their favorite sport.  The hobby of sabermetrics is about obtaining an analytical understanding of the sport. 

As such, there is in fact training, experience, and credentialing involved in becoming well respected as a sabermetric analyst and writer.

Mike/14, I think that’s not their argument per se.  I think Berri/Bradbury would say that there is, in fact, value to the training one receives as an academic in terms of understanding methods.  The training/credentialing received from being a hobbyist may be valuable, but literally anyone can be a hobbyist; not everyone has the opportunity to get the training that some academics do.  Take, for example, Phil’s post of Oct. 27 where he argues that correlation does not imply causation.  This is true, in general, but I imagine if he had had academic/econometric training he would be familiar with causal econometric models where you can, in fact, make causal statements based on regressions.

Now, I’m pretty sure Berri and Bradbury don’t do any causal modeling, and most of the sports economists are not cutting-edge econometricians, but it seems to me that both sides have too much hostility towards each other.  The academics probably know better general techniques but may have less exposure to what’s going on in the saber world.  It would be nice to see some collaboration between the two worlds sometime.

Ben/15, I do believe that it is in fact Bradbury’s argument that a lifelong sports fan has as much subject expertise as an experienced sabermetrician.  This isn’t the first time he has said that.

I looked for the other reference where I remembered him saying that, but so far I haven’t found it.  I thought it was in defense of either the Kovash/Levitt paper or the Hanks/Sauer paper.  He said there, as he did here, that since the authors were lifelong baseball fans, sabermetricians had no expertise that the academic authors did not also possess.

"It would be nice to see some collaboration between the two worlds sometime. “

You mean like Andy Dolphin, PhD, with Tango and MGL to write something like The Book.  Yes, that would be nice to see.

Mike/16

JC said it in a reply to me in his interview over at that Braves blog.  Unfortunately, MVN has shut down so I can’t access the archives.

Basically, I said to him that he should respect saberists as experts in their field, as much as we should respect economists as experts in their field.  It appears that he doesn’t subscribe to that idea.

Tango/17

Yeah it’s a bit misleading saying that all saberists have no expertise in the academic world.  Mike is an engineer and I’m sure he has some high degree in Physics.  Max Marchi is a professional statistician.  Ditto AED.

The most important qualities a good sabermetric researching can have are a good imagination and tech skills.  And most good saberists I know have those things.

And I really don’t get why an economist would be considered an expert in things outside of economics!

Tango/17

Andy Dolphin, PhD in Astronomy?  I loved the astronomy courses I took in college, but it seems to me like an astronomer would be even less qualified to talk about baseball than an economist from a data-interpretation point of view.

Nick/18

I wouldn’t ever suggest that economists should be considered experts at something other than economics.  However, that would apply to engineers as well.

A subset of economists ARE experts at data interpretation in ways that are similar to what statisticians do.  Those are empirical economists and econometricians.  I’m not claiming that Dave Berri or JC Bradbury are the best of them—those are people like David Card who don’t ever get around to studying sports—but you guys seem awfully opposed to economists just because you disagree with a couple of them.  I understand that the interactions between JC and everyone else have not been friendly, but that seems like an awfully small sample size to dismiss a profession.

I’m just saying that economists may possess knowledge that saberists don’t (i.e. Phil apparently doesn’t know the necessary conditions for how to causally interpret regressions, but I’m sure Josh Angrist or most any microeconometrician could tell him those) and saberists possess knowledge that economists don’t.  This shouldn’t be the pissing contest that both sides seem to be interested in making it.

I should also add that a good saberist should also have a solid understanding of probability and randomness - qualities which a lot of outsiders to sabermetrics don’t really understand.

I’m just saying that economists may possess knowledge that saberists don’t (i.e. Phil apparently doesn’t know the necessary conditions for how to causally interpret regressions, but I’m sure Josh Angrist or most any microeconometrician could tell him those) and saberists possess knowledge that economists don’t.

Well that’s just fine, and nobody would disagree with you.  JC, however, has said he doesn’t consider saberists experts in their field, or at least not to the degree where they are more knowledgeable than a guy who has watched baseball for 20 years or whatever.  I would produce the exact quote in which he said that, but it’s lost in MVN’s archives. 

That is what this is about.  JC does not think that saberists are experts in the field of studying baseball.  He believes that an econometrician is more qualified to make conclusions about baseball through empirical methods.

If I am wrong about this, may somebody (preferably JC) correct me.

Nick, he says this:

“and saberists possess knowledge that economists don’t.”

Looks to me like he thinks saberists are qualified.

He has also said that saberists have no additionally knowledge or qualifications than being a long time fan, or something to that effect.  Again, it’s lost in MVN’s archives, so I can’t reproduce the exact quote now.

Nick/21

I agree that JC is out of line when he says that, and will gladly grant you that it is likely his opinion.  Hell, I’ve even heard Dave Berri say similar things about basketball people.

The second part of the quote from Mike/14, though, is what spurred me to comment.  As I mentioned, regardless of what you think about the training, credentialing, and experience of becoming a well-respected saber writer, that’s something that everyone has the opportunity, if not the ability, to do.  Arguing that economists should stick to the economy and let the engineers, lawyers and astronomers talk about baseball is nonsense and ignores the fact that formal economic training, which is not available to everyone, can be valuable for studying baseball.  And yes, sometimes an economist will even be right about something that a saberist disagrees with, if you take enough random draws.

Ben,

You should quote everything that Phil wrote, which is:

“No matter how expert you are in the technique of regression, you have to know something about the subject you’re researching to be able to reach the correct conclusions from the evidence. Because, as the saying goes, correlation doesn’t imply causation. But it doesn’t imply *non-causation* either. It could be that triples cause fewer runs, or it could be that there’s some third factor that’s positively correlated with triples, but negatively correlated with runs scored. Knowing something about baseball lets us argue for which conclusion makes more sense.”

I hope you’ll agree that it is far important to understand a sport well enough to be able to select the correct variables for your experiment than it is to be familiar with obscure conditions for causality.  (It’s not even clear that Phil’s unaware of the conditions for causality.)

Nick, here’s the interview:

http://74.125.155.132/search?q=cache:JL4W0kUOAsIJ:mvn.com/chopnchange/2009/11/13/interview-with-j-c-bradbury-economist-and-sabernomics-blogger.html+nick+steiner+site:mvn.com&cd=1&hl=en&ct=clnk&gl=us&client=firefox-a

As I mentioned, regardless of what you think about the training, credentialing, and experience of becoming a well-respected saber writer, that’s something that everyone has the opportunity, if not the ability, to do.  Arguing that economists should stick to the economy and let the engineers, lawyers and astronomers talk about baseball is nonsense and ignores the fact that formal economic training, which is not available to everyone, can be valuable for studying baseball.  And yes, sometimes an economist will even be right about something that a saberist disagrees with, if you take enough random draws.

Thanks Hawerchuck.  So JC didn’t exactly say exactly what I remembered, but he did imply it:

What does sabermetrics require? Watching a lot of baseball? Thinking about how the game works? Reading Bill James or other sabermetricians? Did it ever occur to you that these are things that economists do in addition to being an economist?

In hindsight, it was probably more of a pissing match thing, as I had previously slammed regression as “not that hard”.  Admittedly, that was an asinine thing for me to say.  However, I later rectified my comments in that same thread, and JC didn’t reply.

A few random reactions:

A Ph.D. Sabermetrician?  It would never work. 

People who use aliases are silly.

DIPS saved my dissertation.  Not that I was researching pitchers, but the idea of looking at year-to-year reliability came in really handy.

Peer review means convincing 2-3 anonymous people with degrees (not necessarily experts… there’s a difference).  Internet debate means that anyone with an internet account can claim to be an expert and chime in.  The thing is that even peer review just gets you published in a journal.  The real review comes on the ground.

In psych, if I propose a new theory or treatment or something and get it published in the Journal of Clinical and Consulting Psychology, that’s cool.  However, if it’s never used by anyone else, what have I accomplished.  There’s an inherent review process that goes with it, and it’s stronger than you might imagine.

Tango/17

Andy Dolphin, PhD in Astronomy?  I loved the astronomy courses I took in college, but it seems to me like an astronomer would be even less qualified to talk about baseball than an economist from a data-interpretation point of view.

That’s an opinion formed with no evidence.

but you guys seem awfully opposed to economists just because you disagree with a couple of them

Speaking only for myself, I learn little to nothing from JC, Berri, and Rodney Fort.

I learn from Ted Turocy.

Any other economist, I have no opinion on.

This shouldn’t be the pissing contest that both sides seem to be interested in making it.

There is no pissing contest, and even if there was, there is no “interest” in having one.

The only thing some of us are saying is that there are two types of experts: technical, and subject matter.  It makes little sense for an academic to have SOLELY other technical reviewers and completely avoid subject matter experts. 

There is no one, anywhere, that I would trust to verify a baseball matter more than MGL.  He not only watches far more baseball than anyone I know, he is also creative enough to form fantastic research pieces.  He’s as good a saberist as there is.  He can guess at things that I would need data on.  And how many academicians reach out to him (and he’s got no “avatar")?  None.  That’s embarrassing.

Yeah, I forgot Pizza had a PhD.  And, I preferred Pizza when he had an internet handle, and didn’t use his real name.  There’s only one PizzaCutter on the internet.  How many Russell are there? On some level Pizza would rather protect his internet handle than his real name.  That’s how I feel about me.  At this point “Tangotiger” is a brand, more valuable than anything else I have.

Actually, I learned from JC’s book.  That’s a worthwhile read, if you ignore the player valuation part.  I should say I learn zero from JC in terms of player valuation.  He’s out of his element there.

>“Phil apparently doesn’t know the necessary conditions for how to causally interpret regressions, but I’m sure Josh Angrist or most any microeconometrician could tell him those.”

That’s true.  I don’t know anything about that particular technique.  Those who do know it are free to use it.  So far, I haven’t seen any sabermetric studies where “causal econometric models” are used.

But that misses the point.  The things up for debate are not technical econometric algorithms.  They’re the most basic techniques in the book.  Selective sampling ... everyone learns about selective sampling in first year.  How to interpret a correlation coefficient.  How to interpret the results of a regression and their significance.

This is basic, basic stuff we’re criticizing these Ph.D. econometricians about.  How much does salary influence wins?  Geez, I can do that in about ten minutes, nine minutes of which is finding the data on a website and typing it in.  And I get it right.  David Berri, in his book, got it wrong.

This is not about fancy techniques.  It’s about basic statistical logic.  JC Bradbury can use the bestest causal econometric regression he wants.  He can use the Batangi and Wu (1988) random-effects method, which corrects for first-order serial correlation.  But, you know, if he’s going to selectively only those players with long careers, and then brag about how up-to-date his econometric algorithms are, then the problem is NOT that the rest of us don’t have enough statistical training.

BP specifically asked me to use my real name.  I thought about fighting them on the basis that Pizza Cutter was now more a brand than anything (I don’t need it for cover like I used to), but decided against it.

Another thing: these fancy statistical techniques are very, very minor improvements over the less fancy ones.  There’s only so much information you can get out of a dataset ... little improvements can’t work miracles.

A fancy method might be able to get you a little more accurate an estimate.  It might be able to get you a smaller confidence interval.  But unless there’s a important reason that the first method shouldn’t work well, there really ain’t a whole lot of big deal there.

For instance: if I’m not mistaken, which I might be, correcting for serial correlation doesn’t change your regression equation, or your r-squared—it just changes your variance, so you’re more likely to get significance.  Sure, use it if you got it, but you can do decent sabermetrics without bothering, almost all the time.

Saying “you didn’t use the most advanced technique” is like saying “you didn’t adjust for park.” It doesn’t invalidate your results, and it usually doesn’t make a huge difference.  It just makes the results a bit less precise.

There are exceptions, but you can usually catch those by common sense.

This is the JC quote I was thinking of.  Thanks Nick for reminding me where it was from and Hawerchuk for the link.  I had looked at the chop-n-change.com version, but it didn’t have the old comments from MVN days.

Nick,

The subject is far too complicated for a tenth-grader to learn. Someone may have introduced you to the basic idea, but you did not learn it. Now, if I’m wrong (and I doubt that), please tell me who the author of your econometrics text was and which software package you used. Or I’d like to see some examples of past regression analyses that you have conducted. Right now it doesn’t seem to me that you do understand the subject. Excuse me for using my experience to make this judgment (Oh but now I’m going to be accused of credentialism, because credentials are only about making other people feel bad).

What I find strange is that you think econometrics is the easy part while the baseball component is the area where expertise is needed. Econometrics requires years of training and practice. What does sabermetrics require? Watching a lot of baseball? Thinking about how the game works? Reading Bill James or other sabermetricians? Did it ever occur to you that these are things that economists do in addition to being an economist? As I stated above (I get really irritated when people accuse me of not responding when those who respond to me don’t bother to read what I have written), both Skip and Jahn are highly knowledgeable of sabermetrics and life-long baseball fans. Jahn has an impressive library of a sabermetric books that would make most sabermetricians jealous. Jahn and Skip have been involved with sabermetrics longer than I have. Part of the reason I know them is that we were connected after learning of our mutual research interest. Why do you think they wrote the paper?

Regarding the questions about plus-minus in basketball:  the pure plus-minus that some boxscores have is next to worthless, but people here seem to be aware of that already.  More useful, but harder to evaluate, is “adjusted plus-minus”, first invented AFAIK by Wayne Winston and Jeff Sagarin (it’s also called Winval), and now calculated by a number of basketball analysts. 

There is still disagreement among hoops analysts about adjusted plus-minus, but I think that a consensus is starting to coalesce which says that adjusted plus-minus is rather unreliable (has high standard errors) when based on one year’s data.  Not worthless, not random, but not something that you’d want to hang your hat on.  (Winston, both in interviews and a presentation at the Harvard Sports Statistics Conference last fall seems to consistently underestimate the importance and implication of high standard errors, and I conjecture that this led to him losing his consulting contract with Mark Cuban, owner of the Dallas Mavericks).

Using two or three years of data reduces those standard errors, but at the cost of introducing other complications that need to be addressed such as aging curves and the inability to accurately measure rookies.

Adjusted plus-minus might thus be roughly comparable to advanced defensive baseball stats.  They give you a measure, and not a horrible one, but one which is still subject to noise.  (Depending on how reliable you feel that measures such as UZR are, you might prefer to say that adjusted plus-minus is less reliable than UZR, that might be legitimate.)

Winston likes to make all sorts of recommendations based on his Winval calculations of NBA players:  who should play more, who looks good but is actually worthless (as he infamously said of Kevin Durant last season).  And a tiny number of basketball analysts are also big backers of it.  I think most would say that it is a measure that provides some information, but not with enough detail or reliability to make useful strategic decisions when used by itself.  Additional information or analysis needs to be brought in.  E.g. that seems to describe what the Houston Rockets do, based on the NY Times article about them, their GM Darryl Morey, and their forward Shane Battier, who puts up mediocre box score statistics but evidently does well on the Rockets’ version of adjusted plus-minus.

Researchers continue to tinker with adjusted plus-minus, seeking to use more sophisticated regression models such as multi-level regression (using fixed effects or random effects) and ridge regression.  I think the resulting estimates are improved, but by how much is hard to say.

(Saved the links for the end because I don’t know how to embed them in the text.)

Best overall NBA statistical analysis site, but it is wide-ranging and varies in quality and relevance:  http://sonicscentral.com/apbrmetrics/

A video of Winston’s presentation at the Harvard Sports Statistics Conference:  http://video.yahoo.com/watch/6196780/16087711

NY Times article on how the Rockets use statistical analysis:  http://www.nytimes.com/2009/02/15/magazine/15Battier-t.html

Okay, I’m going to get a little snarky here on the subject of wins vs. salary.  Hyperbole ahead.

----

An economist comes along and writes a book.  He says that people believe Porsches are for people with money.  But people are wrong!  The “expensive Porsche” hypothesis is a myth.  In fact, the statistics show that you may not even be able to buy a Porsche!

Sabermetricians scratch their heads.  “Huh?” they say.  “How do you explain that sign in the window of the dealership, the one that says, “Porsche, $75,999?” How do you explain that my neighbor just bought one and is complaining about the price?” That’s just anecdotal evidence, the economist replies. 

He says, “Look.  I ran a huge regression trying to predict numbers of Porsches owned based on money spent last year.  The r-squared is very low, less than .01.  So, obviously, there’s no relationship there.”

The sabermetrician says, “well, the low r-squared is not because Porsches aren’t expensive, but because few of them are sold relative to other goods.  That’s normal.  It doesn’t mean Porsches aren’t expensive.  If you look at the equation, instead of the r-squared, you get a coefficient of about $73K per Porsche, with an SD of only about $5K!”

The economist says, “you obviously don’t understand how r-squared works.  You need to take some econometrics courses.  And I remind you that my study was peer reviewed.”

-----

And that’s why I say: the problem is NOT that the economists’ methods are too hard for us to understand.  The problem is basic logic.  It’s not quite as obvious as the Porsche example, which is why (IMO) certain economists still get good reviews for their work.  But, I submit, if you study the issues we talk about here, and you look at the arguments, and you evaluate them fairly, you will find that (again IMO, and I could be wrong):

(a) the criticisms are basic statistical logic;
(b) they are correct; and
(c) most of the time, they have nothing, absolutely nothing, to do with the technical details of the method used.

I don’t think it has to do with having a PhD, as I’ve had great experiences with Jinaz, Pizza Cutter, Matt Swartz, Alan Nathan. Maybe some of it has to do with teaching in a university, but Jinaz does that, Pizza has, I believe Ted Turocy does.

Some people just have frakking attitudes. I posted a pic on my facebook of me, Pizza & Eric Seidman. Will Carroll commented “There’s 500 points of IQ!” and Pizza says “They’re all in Brian’s head”. Now my son thinks that is the most hilarious thing, especially coming from a guy with a PhD, but at least when I left a comment on his article as BP today, Pizza didn’t tell me “it’s too complicated for a tenth grader to learn”. He knows how to respect people.

I’m sorry, but the idea that the sabr crowd has a problem with econ guys is just wrong. 

Nate Silver is an econ guy.  His stuff was well received because it made sense and was founded on the basis of fundamentally sound arguments.  He didn’t waive his academic credentials around - he just wrote good stuff and a community of smart people recognized that he was right. 

Matt Swartz is an econ guy.  He’s doing just fine.  I’m an econ guy, and I don’t think I have a problem with myself. 

There is no sabr/econ divide.  There is a JC/logic divide, in which he stands alone arguing stupid things and everyone else with any sense of rationality says “hey man, you’re being an idiot and a jerk.  Knock it off.”

#32, thanks for the props Tango.  I often feel like an old dog trying to play a new game with a bunch of young dogs.  There are a host of great young sabermetric researchers out there.  I also have always been a little frustrated at not having the technical expertise that some of the more statistically-trained folk have.

I don’t think there is any pissing contest here. To be a good saberist requires the same thing as being a good anything. That is, a great understanding of the material, lots of experience regarding the subject matter at hand (that may include being a “baseball fan for 20 years” but not necessarily), an inquisitive and imaginative mind, a critical eye, an ability to be objective, and some tools to analyze and evaluate the material, questions, and problems within the field.

It is great if you are a statistician, economist, or a research psychologist (or you have the expertise that they do) like Pizza (and others) AND you possess the baseball requisites as well, but it is by no means mandatory.  There are many niches in sabermetrics, just as there are in all fields.  Some require more statistical or technical acumen than others.

An economist may or may not be a good econometrician and he may or may not be a good sabermetrician. It has nothing to do with whether he is an economist or not. I have no idea whether JC or Levitt, for example, are even good economists/econometircians.  But whether they are or are not (and I’ll assume they are), I don’t think either one of them is a good sabermetrician, although JC is probably much better than Levitt.

I mean, most sabermetricians, even if it is not their “real” jobs, spend countless hours thinking and writing about sabermetrics.  I doubt that Levitt has spent as much time in his entire life thinking about sabermetrics as Tango spends in a week.  He can’t possible be an expert in that field, or even close. And while JC does spends much more time than Levitt on sports analysis, I doubt that he spends nearly as much time as most of the regulars on this blog and the ones who write for BP, THT, BtB, etc.

Anyway, I lost my train of thought, so I’ll sign off…

I don’t have a problem with JC or the saber community (and I would align myself firmly the camp of the latter, however, I’m neither a sabermetrician nor an economist), but I feel that both sides just get too snarky for my tastes. Then again, I suppose that comes with the amount of times the two sides have clashed over the same subjects.

Secondly—and this is astray from the topic—but is MGL a pseudonym? I’ve been reading/lurking around here for a couple of years but can’t recall if it was alluded to or not. It doesn’t make a difference to me one way or the other, but was just curious.

And while on the subject of it, Russell, why did you choose Pizza Cutter as the handle?

Sorry if this takes this off topic a bit. I do like reading these disagreements a ton.

MGL are my real initials.

It all went downhill once JC said that Frenchy’s 2008 performance (the performance, not the talent level we can infer from that performance) was worth 12MM$.  And, that JC stuck to it, with feet firmly in the ground… like quicksand.

We gave him lifeline after lifeline, and he refused it.

As others have said, it’s JC v world here.  We’d welcome him with open arms, if he would acknowledge he’s in quicksand, not on a great surfing wave.

Pizza Cutter was my DJ name back in college.  I was working as a therapist when I started writing publicly and needed a way to keep myself from being easily googled.

If I may also point out a rather interesting study that we were fond of in grad school.  Someone looked at therapy outcomes (whether people got better) as a function of the level of education of the therapist (Ph.D., Masters, B.A., no degree in psych) and found no differences.  As the authors said, there is no empirical justification for grad school.

Wayne Winston, who sources from Phil, Book Blog, BP, and many others in his book, is a professor with a math degree and phd, and worked for NBA team.

Cyril Morong is a professor of economics.  He’s been nothing but congenial in conversations with me and has contributed a fair bit to sabermetrics.

I’ve also had a good experience with other PhD’s and teaching professors in other fields.  Pizza Cutter is responsible for getting me started in the world of sabermetric writing at StatSpeak.  Alan Nathan has been a huge help to me in the study of the physics of baseball and PITCHf/x, and similarly, I’ve found Dr. Dave Baldwin and Dr. Mont Hubbard to be very easy to correspond with and willing to help.

I also use a great deal of my training in physics and engineering in studying baseball.  Not only the technical tools like linear algebra and Perl scripting, but more importantly, the rigorous training in methods of thinking.  Attention to detail, careful observation, problem solving, dealing with measurement error in physical systems, perceiving underlying principles, etc.

I also attempt to stay as current as I can on the academic literature in my niche of the field, which covers things like the aerodynamics of balls, bouncing balls, pattern classification, and the like.  I very much appreciate the help of those who are more plugged in to the academic world than I am in finding the relevant papers.

Academics versus laypeople is as false a divide as scouts versus statheads.  Sabermetrics is all about learning about the game.  The source doesn’t matter.

For that matter, if you starting counting up the PhD students, the divide between academic and sabermetric starts looking even sillier:  Dave Allen, Dan Brooks, Chris Moore, Sal Baxamusa.  I’m sure that list could go on for a while, but those are a few off the top of my head.

And I forgot the sabermetric pillar Dr. John Walsh when I was listing professors earlier.

Actually, I think Sal has his PhD now.  My apologies Dr. Baxamusa.

Ike Hall, who has been a PITCHf/x contributor, has his PhD.  Josh Kalk was a PhD student, also; I don’t remember whether he has his PhD now (nor do I particularly care--look at the quality of his work, but I digress).  Rany Jazayerli has an MD.

Shall we go on?

I think Sean Forman has a PhD as well.  And maybe Dave Smith at Retrosheet?  They do good saber work when they are not providing fishing rods for everyone else.

I would definitely count Dr. Forman and Dr. Smith as key parts of the sabermetric community.

And that adds another teaching professor and former teaching professor to our list.

Foreman was a tenure track professor at one of the local colleges.  He’s certainly a PhD.

Isn’t Smith a prof at a Delaware university?  He’s very likely a PhD for that reason as well.

Also Dan Rosenbaum was an Econ professor who was part of ABPR community and critiqued Berri.

Sounds like this community is not lacking in credentials.  The difference is those who use credentials as part of the argument or those who let the results of the studies be the argument.

For some reason this has generated a lot of traffic lately.  We’d probably be better off ignoring him until the time JCB actually produces something that enhances our understanding of baseball.  Though I guess he serves some purpose, showing how not to value a player’s MRP or the aging process, maybe proving some examples for others to avoid.  Though before his Frenchy2008 = 12 million sytem I don’t think even the most novice saberist could have conceived of such an obvious and fundamental mis-evaluation.

Rally/56, I think we were well on our way to ignoring him, or at least ignoring his thoughts on player valuation and aging, until BP decide to stir the pot by giving him their front page.  That’s what’s generated most of the recent traffic, IMO, what Will Carroll would call “leading the discussion”. 

Let’s cut to the chase.  I know a number of friends who are professors—(and many of themhave tenure, so they aren’t the bitter ones) who think the peer review process is overblown.  Simply put, it the peer review process wasn’t directly tied to tenure decisions, it would die a slow (or maybe not so slow) death. 

The main function now of the review process is precisely for tenure decisions, and to a much lesser extent some snob appeal for credentialing to the outside world, but that’s about it. 

Given the free flow of information and sometimes data, the cream would naturally ride to the top if academia moved to an internet debate methodology and that is indeed what happens in the sabermetric community.

And P.S. count me in (if you think it’s justified) as someone on the fringe of the sabermetric community who has a PhD

The fact that many of us have no idea as to the credentials and degrees levels of many of the otherwise well-known sabermetric researchers and writers is a testament to the fact that there is very little ego, chest beating, and appeals to authority in the sabermetric community, at least as compared to other fields.  One of the reasons is that it is well-known that much of the best sabermetric work has been done by laypeople over the years.  It is probably only recently that a lot of great work is being done by people with advanced degrees and wide swathes of expertise. Part of the reason for that is that the data (like pitch f/x) has gotten so much more “technical” that it requires more advanced techniques to analyze it.  As well, the questions that we are trying to answer have gotten more complex and difficult and thus also require more advanced and sophisticated methods, knowledge and tools to answer.

The two advantages of peer review are that (in theory) it screens out the worst and forces improvements to other papers. It also serves as a permanent quotable record of where “good” papers are. Which helps if you are trying to research and learn about an issue.

I guess one issue is would sabremetrics benefit by having a recognized repository of “good” papers where only the best would make it?

Are good papers being ignored because they dont catch the eye of the community?

Will older papers be lost if their websites etc arent kept going?

Is much time wasted writing and reading poorly analysed studies or those that repeat what has been done before?

A peer review approach would be too slow and cumbersome and goes against the culture that has evolved. I like “By the Numbers” but I would be interested to know what proportion of papers are rejected and what proportion are re-written.

But taking the wisdom of crowds approach papers could be put on the site for say, three months and then anyone could vote on the quality and those that get a high enough score would be kept permanently.(A Sabermetric paper Hall of Fame!!!!)

Alternatively maybe you have to have a paper accepted before you can vote like the oscars.

This would act as a type of QA and any paper could be submitted for the community’s approval it could be a one stop shop to see what new papers are out and also where the good ones are.

James

Hmmmm… wisdom of crowd approach,eh?

If we can make it more nuanced, like tagging the papers/articles, linking them to the reviewer/voter, and then creating some sort of rating system based on the reviewer himself, that would go a long way.

Basically, the way the Google search engine works.  And I think Google is more successful than the academic review system.

I think it would be great if we had a sabermetric librarian.  Unfortunately, I think it could be a nearly full-time responsibility.

I started out doing that for PITCHf/x-related research, but the cataloging eventually exceeded my ability to devote time to it.

I think Tango’s sabermetric wiki has the potential to fill most of this role of “sabermetric library”.  We’ve got a few people tackling the enormous responsibility of adding content and links of the most noteworthy research on all sorts of sabermetric topics. 

The wiki is a community effort to condense the best research out there into one place for reference, much like the library you’re describing.  Another nice thing about the wiki is that it can be updated easily with new research (via text and links) as it is published.

And of course, the more people who can help out, the better!  You don’t even have to be an “expert” to help- if you can spare a minute to add a link to a topical article, that helps a lot!

I find the wiki format hard to use for library-type research.  It’s more useful as an encyclopedia--hitting the high points and having a brief mention of everything--giving you a place to start. 

I wanted to expand the PITCHf/x entry in the wiki, and I did write some, but to really do the topic justice would be a significant time commitment.  Even the article I wrote for the Hardball Times Annual was just scratching the surface of information on PITCHf/x, and that extended to nearly 4000 words. 

I applaud Dave Cameron’s effort to collect links over at Fangraphs.  Similarly, the sabermetric awards at BtB are a very good idea.  But neither of those two ideas really functions very well as a sabermetric library for future research, unless somebody takes them and spends some time thinking about the best way to organize and index the information.

That’s why I say we need a librarian.  I know that’s not realistic.  But neither is the Internet’s favorite solution, barfing up long lists of links that are loosely organized at best.  Those things fade away because they’re not usable.

A library doesn’t really condense (like a wiki does), nor is it primarily a list (although it is a “list”, among other things).  It’s primary function is to organize in a manner that will facilitate education and research.  Among the responsibilities of a good librarian is effective culling and ranking, which the Internet doesn’t do well.  Another responsibility is linking, and the Internet does that amazingly well, but it doesn’t completely make up for its lack of the other features.

Agreed, you need a librarian to have success.

I agree that a long comment thread with links is not the ideal solution.  My hope (apparently a long shot, given the tepid response so far) is that we could collect the information that we would want to categorize as a first step. 

I also agree that there should be a “librarian” position, though I am not sure it needs to be filled by one person.  If, for instance, we ended up hosting something like that at FanGraphs, we have enough writers on staff to spread the workload around (this is purely hypothetical, of course - I’m not suggesting that the library should be or will be hosted at FanGraphs).

I’m not trying to discourage you, Dave, by any means.

However, I will caution that one thing I found most difficult during my effort as a PITCHf/x librarian was deciding what to exclude.  I think I included too many articles, reducing the usefulness of the catalog and increasing the difficulty of maintaining it. 

It is hard to tell someone, even implicitly, that their nifty new article on your topic is not good enough or novel enough or helpful enough for further research to be included in your library. Or to rank the ones that are good enough to determine whether they are just plain-old-good-and-informative or you-must-read-this-if-you’re-doing-any-research-in-this-field. 

Setting aside the personal feelings that may be involved, it’s often hard to tell what someone else will find valuable and how they will find it valuable.  Voting in some manner, whether by experts or community, can be helpful, but a good library finds a way to index materials in better ways than just by popularity or rating.  Sorta like Google PageRank does, I guess.

Just by way of a brief example.  Let’s say someone publishes a PITCHf/x article about A.J. Burnett.  Maybe it’s sorta ho-hum in terms of what you learn about Burnett, although it might be useful for somebody who’s particularly wanting to study him.  And it’s got some useful concepts on pitch classification buried inside, although it doesn’t really address them directly.  And it’s got a real nugget about something completely tangential, maybe some nifty SQL queries that could have broad application.  How do you vote on its overall worth and how do you index it?  And how do you help people looking for those various things to find them?

I’d love to hear answers.  I eventually had to give up the PITCHf/x catalog because I couldn’t satisfactorily resolve them for myself.  Instead I end up making a mental note and relying on Google to find what I want later, but that’s temporary and very imperfect system and of little value to the community.

Also, it’s probably easier to catalog in an established “field” like linear weights or run estimators.  It’s more difficult to measure the value of a publication in a rapidly evolving field like defensive evaluation or PITCHf/x.

Ideally, it’s the authors themselves that catalog their work.  That’s why I tag all my blog posts here. Then, you just need to merge all the catalogs into one catalog.

The incentive system just isn’t there for this to happen any more than it already does.  If it feels like work, then you need to get paid for it.  If it doesn’t feel like work, you’ll do it because it’s part of your hobby.

As it stands, being a sabermetric librarian is pure work.

Mike/others,

For the wiki, you DON’T have to write more than a sentence or two of summary of past work.  Writing a brief recap and a link is a thousand times more efficient use of time, and it gives the same information (just via link rather than reprinting whole articles, which is probably a plagiarism issue as well)

Two minutes to track down the right link, two minutes to write two sentences, and two minutes to add the correct wiki syntax (if you don’t know it already).  Hopefully that’s not too time-consuming for anyone.

Mike,

What do you think of a library for articles that detail a pitcher’s repertoire or simply have quotes in which a pitcher says that he throws two types of fastballs? Just for the sake of making pitch classification easier on current and future pitchf/xers. Would you help contribute to it?

By the way, “Tom Tango” is an alias, as most of the regulars here know.  Not that it actually matters what actually is and is not on my birth certificate.  Every baseball site I’ve ever posted to (and that’s got to be upwards of 100, if not 200) has some form of “Tango” or “Tangotiger” in it.  I’m accountable for what I say. What matters is that I have ONE handle.

I don’t think this analogy is correct or fair.  Phil and JC put their name out there in their professions.  If you go on ESPN, and you go on as “Tom Tango”, then I can buy it (more).  The Internet is *mostly* anonymous, so saying that you use a consistent anonymous name in a place where few people see it (baseball message boards) will make you a “brand” on baseball internet message boards, but it isn’t the same as JC.

For instance, Pizza Cutter has a brand, and it will change when he moves beyond “internet message boards”.

I am NOT saying you cannot “become” Tom Tango, but merely you aren’t in a position to be that just yet.  Close, but not quite.  MGL is absolutely known as MGL in all the places you are known as Tom Tango.  But he’s not MGL in person/on TV.

It’s different.

Also, a PhD means NOTHING outside of whatever your particular dissertation was on.  You take the same classes and do the same grad school work that people with Masters, or even those that didn’t get a degree did.  *Perhaps* you wrote a really long book report on the Vietnam War. 

Having a PhD doesn’t mean you have any claim to understanding what is being done.  Hell, you can get a PhD and a C in an area while someone who doesn’t get a PhD gets an A.  Who understands the material better?  Who sucked up more to the Division head? 

There are more politics to getting a PhD than you can believe, and let me say that it’s about studying the subject matter and the tools.  Having a PhD or not doesn’t give you any claim that you understand their utility any better.

No offense to any PhDs reading - many of them will tell you that the smartest guy in their grad school never finished.  they *can* be the smartest, but a PhD in and of itself has no specific knowledge outside of his specific area of study (above any non-PhD person that took the classes/studied the field).

/climbs down from high horse

I also agree that there should be a “librarian” position, though I am not sure it needs to be filled by one person.

Not to spill beans, but Ted is working on this.  And I think he’s reasonably close.

If you go on ESPN, and you go on as “Tom Tango”, then I can buy it (more).

http://insider.espn.go.com/mlb/insider/news/story?id=4750792

“By Tom Tango
Inside The Book and special to ESPN Insider”

http://online.wsj.com/article/SB124000966755730671.html

“—Tom Tango
Printed in The Wall Street Journal, page W4”

Dude, you are not being fair now.

/climbs down from high horse

Actually, you are likely in the majority opinion here.  You are at sea-level.

Not to spill beans, but Ted is working on this.  And I think he’s reasonably close.

Turocy?  I’ll ask him, but I will guess he’d only have at most 10% of his catalog complete, and likely under 5%.  I say that only because in email exchanges I’ve had with him on other unrelated matters, his time availability is as limited as mine or anyone else’s.  If I’m wrong, then I’m in awe of his time management skills.

Dude, you are not being fair now.

Well, I disagree.  That’s still online content, is it not?

And I said you were “close”.

Ted’s is going to be wiki.  You’ll have to add a bunch of your own....

Actually, you are likely in the majority opinion here.  You are at sea-level.

No, then we have to adjust for position, so I get like a +14.

The WSJ article was printed in the WSJ.

All my sports-related dealings are as “Tom Tango” or “Tangotiger”.  My entire reputation is based on my brand.  If, let’s say, I decided to go crazy and blemish “Tangotiger.net” or “InsideTheBook.com” causing irreparable damage to the “Tangotiger” brand, and therefore, I have to shutdown both sites and have no more sales on my book, yes, I can still live my normal life outside of that.  No one would know.  I’m like two separate people.

BUT, for me to get back into sports-related dealings, my real name is almost useless.  And, I’d have to build a new brand up from scratch.  And do it in such a fashion that I was completely separate from “Tangotiger”.

This is what your argument comes down to, that I could just pack up and leave and come back with some other handle, like “Frank Culligan” or something.  But, the way I write, the way I present things, the way I interact is all fairly unique.  Someone would eventually say Frank Culligan??  Uh, you’re Tango aren’t you?  They guy that went ballistic that time?  No thanks “Frank”.

So, I’m not close.  I’m out there as much, if not more, than JC and Phil.  I’m more exposed, I’m more involved, and I have more to lose.  I’ve jumped over the line where I’m not “anonymous” like most internetters are.

I’m like two separate people.

This is why I say it is different.  JC doesn’t have this safety net.

I’ve jumped over the line where I’m not “anonymous” like most internetters are.

Fair enough.

"This is why I say it is different.  JC doesn’t have this safety net. “

I don’t think that’s true.  JC could write virtually anything he wants on his blog and it would not affect his academic career.  He’s a tenured professor and his promotion to full professor is not going to be affected one whit by what he writes on his blog barring a complete meltdown, and even then it probably wouldn’t matter.  Nor is it going to be affected by the quality of his non-scholarly publications.

There are people in this world who know exactly who Tom is in real life.  If he did something completely nutty, the people who employ him in the sports world might not take too kindly to it.

Wow lots of comments
when i first suggested this I did not have a massive library of every paper in mind. Rather I thought initially a list of maybe the 100 “best” articles/book chapters etc. These could be lodged in the “library” or where they are in commercial sites or book chapters linked to or listed.

I guess I am trying to answer the criticism from some academics of the lack of peer review by setting up an equivalent system based on quality as judged by peers rather than the largely unreviewed open publishing system we have now. Because in science you are a judged as a “good” scientist if other “good” scientists think your work is good enpugh to be should be published. I think the same can be done in Sabrmetrics.

i also think the system needs to be automated as much as possible and one way is to keep it small and high quality (at the risk of missing things) rather than universal.

Admittedly even selecting the best 100 papers would provide endless (enjoyable?) debate and you might want to have a Bill James 100 and an everyone else 100.

After this is set up anyone can submit a paper for grading to be reviewed by people who have a paper in the library. For every paper someone has they get one point for voting purposes.

Ther papers would all be public and anyone could comment on the paper

Reviewers have 3 months to grade a paper as

A) Accept- Ground breaking work everyone should read
B) Accept- very interesting work people will refer to over the years
c) Accept - Interesting work worthy of being highlighted.
D) Needs revisions- Comments supplied
E) reject.- reasons supplied

After the three months the paper is accepted/revised/rejected depending on which category got the most votes. YOu could also score accepted papers on the number of As Bs and Cs so that not all papers are equal in the library.

Revised papers come be resubmitted with either accept or reject options no more revisions.

I would have the review being doubly anonymous ( reviewers and authors )until the scores are revealed.  So a Bill James paper would be marked the same as one by an unknown person. Although you could argue that any comment of his is worthy of serious consideration

I work on peer reviewing scientific applications a lot in my job (and I have a PhD) so I can see the advantages and disadvantages. But having a standing panel with open peer review could work and once your paper is accepted you are then a member of the panel entitled to vote.

One draw back could be the devlopment of the bitchiness we see in academia and the moaning and accusatons of rejected authors, and a tendency to salami slice.

But I think it could improve the quality of research avoid duplication and allow people a chance to make a name for themselves.

Commercial sites could submit papers (although I guess only the abstract could be viewable to the non-voters) so that if they were accepted they could then advertise this on their sites similarly if one of Bill James books has lots of good chapters in it then I would be tempted to buy it.

Now I know nothing about how difficult this would be to set up technically and also would people be interested in assessing others work. Also how many papers would be submitted every month 1, 10 or 100 ?

James

I had some thoughts which have many things in common with what James said.

I believe some blogging software would handle my proposal, probably similar to FanGraphs.

There would be a few people who would be the gatekeepers, moderating, blocking spam, and maintaining membership lists.

We would decide on some criteria for membership in the ‘Panel of Experts’. These are the people allowed to post articles and vote on them. Posting articles would not consume the time of only a handful of people.

Entering an article in the library would include a link to where the article is available online, with an indication of whether this needs a subscription to read in it’s entirety. These can be old or new, we are building a reference library.

Standard and ad-hoc tags for the article so that they may be grouped topically for searching. Stuff like BTB just used for their awards ‘Novel Research’, ‘Applied Research’, ‘Commentary’ etc, plus the actual subject such as ‘DIPS’ ‘Pfx’, ‘Player Valuation’, etc

Then zero to many links where the article may have already been discussed (such as here on ‘The Book Blog’.

A poll, one to five stars, which can be associated with the A-E James listed above. Who voted how on the poll is not know to the public, but possibly is recorded by the software. Results shown in a histogram, plus a numerical average of the vote (possibly a Bayesian regression such as IMDB for smaller vote totals). Authors could also have an average score for all their articles.

Then comments and reviews by any member. Let the author on to defend his work if he is not already a member. Why it’s good, why it’s bad, how it can be improved.

This way, a library user can receive a list articles on a particular topic, how well that article was scored by the ‘Experts’, read reviews, and have links to any long discussion threads about the article.

Hmm…

http://www.lookoutlanding.com/2010/1/26/1271512/on-the-shoulders-of-giants-a

Just wanted to say that a big part of the reason that I proposed the sabermetric awards at BtB (results are in, awards write-ups are in progress and should be out next week) was directly related to my feeble efforts to assemble a library of sorts for this baseball class I was prepping for.  I was struck at how difficult this was to do.  We rarely as a community seem to make a concerted effort to compile our best work, and so I thought an annual “awards” thing of sorts might be a good way to encourage us to do that.  Naming a “winner” was always a very secondary goal, but it does make it a bit more fun and gets people interested.

This year, my plan is to make monthly posts to help us take note of our favorite studies throughout the year--I think a lot of us had trouble remembering many of the great studies we’d seen over 12 months, so hopefully people will be willing to share their favorites.  We could then mine these lists for next years’ awards thing.

Also, I have added a great deal to my list of sabermetric studies since I originally posted this thread:
http://www.beyondtheboxscore.com/2009/12/17/1200459/want-to-help-me-plan-my-baseball
I keep it in a google wave right now, but I can post it as a resource somewhere if folks would like me to. 
-j

Here’s the updated list:
http://www.basement-dwellers.com/2010/01/sabermetric-library-resource.html

Re Jeremy/70:

What do you think of a library for articles that detail a pitcher’s repertoire or simply have quotes in which a pitcher says that he throws two types of fastballs? Just for the sake of making pitch classification easier on current and future pitchf/xers. Would you help contribute to it?

Yes, I think it’s a good idea, and yes, I would be willing to contribute.

In my head it’s something a little like what Rob Neyer and Bill James did with the Pitcher’s Guide (see “All the Pitchers Who Wouldn’t Fit”: http://www.robneyer.com/book_04_extras1.html), except that due to the availability of PITCHf/x information and more interviews on the web, we should have more references and more detailed information on each pitcher than Neyer and James had when they did their book.

The wiki format might fit this project pretty well, unless you had something else in mind?

Also, if you haven’t seen it, Sven Jenkins scouting reports are a great resource.  He’s not perfect, but he’s far better than any other scouting source I’ve found, and he’s pretty complete in the number of pitchers he covers.
http://www.60ft6in.com/