THE BOOK--Playing The Percentages In Baseball

He’s wrong on every single point. 

JC seems to have two very strong beliefs that influence all of his conclusions about the sport.

1. People of a certain position are essentially above reproach

2. That academic economic models can be easily applied to baseball.

He essentially makes the case for assumption #1 in his “buy low” argument, and we’ve seen him make it before it regards to why he won’t engage the sabermetric community about the flaws in his model.  He thinks the idea that a common fan knows as much about a certain topic as a baseball executive or a economics professor is ridiculous. 

The rest of his opinions seem to be shaped by the idea that models like MRP can be used to evaluate the baseball economy with few adjustments.  He misses some huge problems that are created by a closed market with significant inefficiencies, of course.  In his world, trades are good for both sides because otherwise teams wouldn’t engage in trade (a basic economic tenet), but misses the massive information gap between teams.

JC expanded on his points at his Sabernomics blog:

http://www.sabernomics.com/sabernomics/index.php/2009/11/the-size-of-the-free-agent-pool-and-competition/

http://www.sabernomics.com/sabernomics/index.php/2009/11/are-general-managers-myopic/

http://www.sabernomics.com/sabernomics/index.php/2009/11/trades-is-are-mutually-beneficial/

"In his world, trades are good for both sides because otherwise teams wouldn’t engage in trade (a basic economic tenet), but misses the massive information gap between teams.”

To be fair to JC, he repeatedly says that ‘trades are, by definition, accomplished such that both teams think they are getting what they want,’ which is tautological of course.

Now, just because I think I am getting fair value by trading my gold Rolex for your old pair of Levi’s doesn’t make it so, of course, other than in a philosophical sense (i.e. if I “think” that is a fair exchange, so be it).

The argument to which JC refers, I think, is, “To what extent do teams/GM’s make “mistakes” in evaluating trades/signings, etc.?”

That is a reasonable discussion, I think.  I don’t think that JC says the answer is “never” and I don’t think the saberist says that the answer is “always” so I am not really sure what or whom JC is arguing with/about.  And I don’t think it is an economic issue at all.  We ALL realize that in an exchange, in baseball or in any other context, each side thinks they are getting a fair deal and that many exchanged can have more than one “winner.” Again, that is tautological and we don’t need an understanding of economics theory to tell us that.

So while I think the article is silly, I am not really sure what is “wrong” in it.  Obviously JC does not think that every single transaction by every single team in the last 10 years does not involve some “mistaken” belief by one or more teams? That would be a preposterous and delusional assumption…

"GMs can buy low and sell high...”

This is one of the “myths” that JC attacks.

JC says that some people think GM’s are really stupid and JC says that they aren’t.

I wouldn’t really know how to respond to that one way or another.

It is true that GM’s know some things about players and about transactions in general that the fan and the saberist does not. It is also true that the saberist (and even the sabermetric-oriented fan) knows some things that the GM doesn’t.

That is probably the hardest thing for many people to grasp - that a person living in their mother’s basement knows some things that a GMw ho gets paid a million dollars a year or more and has been in baseball for 20 years does not.

But…

That is why many teams are hiring these guys that live in their mother’s basement, for cripes sake!

As far as the specific issue of whether some GMs/teams overvalue or undervalue recent performance, which is the specific issue that JC is discussing…

We should be able to look at the data to somewhat answer that question:  Compare the contracts for players after an up year with those of players after a down year and see if the former is higher than Marcel would suggest and the latter is lower.

I’ll bet my entire reputation (OK, that is not much) and a lot of money that that would be true.  To what extent, I have no idea.  And I would also bet that that has been diminishing over the last 10 (or however many) years.

Again, I am not sure what JC is arguing.  That that is NOT true to some extent on the average?  That that never happens?  That is hardly ever happens?  If the latter, that depends on your definition of “hardly.”

Discussions and arguments using qualitative words like “dumb, “stupid,” “hardly ever,” “usually,” “bad”, “good,” etc., are stupid (yes, “pun intended") arguments that cannot be won by either side, unless those words are specifically defined.

Let’s argue the question, “Are GMs/teams stupid?”

Dumb question.  Dumb argument. 

That seems to be the substance of JC’s article.

You love having a ballplayer like JC in your leadoff spot: he plays hard every day, does a variety of things, and he is not a streaky blogger. You gotta have a consistent force up at the top of the order! Quantity doesn’t slump! 

I expect every time I come across one of his articles that this will be the time he starts to figure it out- the Ph.D. might sound toolsy (pun not originally intended) but he continually puts out terrible work without improvement and it now seems almost amusing. 

Similar Value MLB Players: Aaron Miles.

I don’t read sabernomics at all, unless there is an interesting link to it, but the more I read it, the less I want to read it.  I’ll comment on one more thing that JC wrote which is the last link in #2 above:

“Every trade has a winner and a loser — Swapping resources only takes place if both parties are made better off. Therefore, when we observe trades taking place, it’s likely that both parties are doing so because they expect to improve their teams (see the weak axiom of revealed preference, or as I call it: “the useful WARP“). Mistakes happen, but as a general rule, all parties to trades are winners. Who says economists aren’t touchy-feely?”

Who EVER said that EVERY trade has a winner and loser?  You can’t talk about “myths” when the myth you are quoting is not a myth at all since no one would ever say it.  That is like saying, “Here is myth #1: EVERY GM is a stone cold idiot.” No one thinks that and it can’t possibly be true.

JC then says:

“Swapping resources only takes place if both parties are made better off.”

That sentence, again, is so preposterous that you can’t and shouldn’t even read further.  Obviously what he meant was that swapping resources only takes place when all parties THINK they will be better off.  Why wouldn’t he say that?  I don’t know.  He does in other parts of the original article.  I don’t like (I hate actually) when a person more than occasionally misrepresents what they mean to say, for effect.  That is highly unethical.  I don’t know that that is what JC is doing, but it surely seems as if that is the case.

Bradbury’s invocation of the “weak axiom of revealed preference” is a total red herring, as far as I can tell - I think he just threw it in so he could make a dig at replacement-level metrics.

The problem with Bradley’s arguement is that he assumes that the fundamental theorem of exchange holds for baseball trades without examining the question. I really don’t think it does - the fundamental theorem of exchange presumes a different marginal use value of goods being exchanged, if I recall correctly. That may sometimes be true in baseball trades, but I don’t think it’s necessarily so. (Bradbury also ignores things like transaction costs and the agency dilemna.)

I believe most of his points in this article are either trivial or wrong, depending on you interpret them.

I think we only glanced by one of his other myths, that one being: “The number of free agents at a position affects the price of free agents at a position.”

J.C. again either misrepressents what he’s trying to say or he doesn’t really get it.  He states, “The problem with this is that the free agents have come from somewhere. A high number of players looking for new teams means that there is a corresponding number of openings that teams need to fill.” While that might seem true enough, he forgets that players also arrive and leave the league unevenly.  So you might have 5 FA LFers leave a team, 3 LFers retire, but have 6 teams ready to plug in a rookie (or a former bench player from their roster or a maybe moving an older CFer over and a young guy into CF, etc).  So that’s 3 players leaving the system, 6 players entering and 5 looking for jobs.  Two people are getting left out or taking on a new possition.  Its possible to imagine all kinds of cases like this. 

Over all J.C.’s arguments come off as though he’s trying to sound smart, but really he’s simplified the issues to the point where he’s leaving out importants pieces.  Reminds me of an Einstein quote: “Make everything as simple as possible, but not simpler.”

"GMs have made mistakes in the past and will make mistakes again, but they’re not dumb enough to act on a meaningless hot/cold streak.”

I had to have a chuckle at that one.

Nothing that anybody has written here in the comments contradicts anything Bradbury said or would be anything he would disagree with. 

He’s pointing out that message board commenters discussing trades routinely ignore the fact that GMs generally know what a hot/cold streak is and that you don’t need to wait for a guy to get hot to trade him.  Does anybody disagree with that?  Do message board writers not routinely act like GMs are dumber than they are?  Of course, they’re getting smarter and they are subject to fallacies but the concept that they don’t end up learning from mistakes after they make them is silly.  And people do underestimate their intelligence all the time. 

Holding all else constant, does anybody disagree that since free agents come from teams that unless the free agents are free agents due to call ups that shouldn’t affect their value?  Yes, I can think of a thousand exceptions to that rule, but the point is that the price wouldn’t go down because there were free agents on the market directly. 

It’s also true that people who try to cook up trade ideas without knowing sabermetrics well frequently trying to think of trades where they trick the other GM.  The point is to find a trade where someone has a comparative advantage at one thing and someone else has a comparative advantage at something else, and they trade.  A win-now team trades prospects to a rebuilding team for an aging star.  Teams with surplus outfielders but no third baseman trade with teams with two third basemen and a shortage in the outfield.  That should be what people look for in trades. 

And people routinely act like older players are worthless, which isn’t true. 

All of these ARE hot stove myths, though maybe not to people who read The Book Blog obviously.

I don’t get the hate thrown his way or the way of economics.  Economic theory and sabermetric thought overlap tremendously.  I’ve been learning sabermetrics ridiculously quickly thanks to having the economic background that makes some concepts (such as replacement level) redundant.

I think Bradbury’s old discussions of replacement level were not even really dismissing them, but overreacted to and then turned into an argument where no one actually disagreed with anyone in reality.  His point was that you can think about opportunity cost without giving it a face, as near as I can tell, and that’s fine.  The analysis goes through just the same.

Swipes an “academic economic models” are silly.  An academic economic model can take any shape you want it too.  If you don’t have a model of the baseball labor market in your mind, you can’t write anything down about it and no one will read you.  Every single GM has a semi-conscious model of the labor market in mind.  Everyone has theories.  Economics tease those out.  Every trade analysis I’ve ever seen the people posting in this thread write up has a model in mind.  Economics is about writing down the models that people have semi-conscious when they try to explain human decisions.  Having training in how to write those models makes it easier to do so. 

Also, Colin, revealed preference is a ridiculously common thing that economists use amongst themselves all the time.  It’s not a red herring of any kind.  It’s a standard expression that economists use amongst themselves to explain a general concept quickly (one is logically self evident), and they do so about as frequently as they use the phrase “comparative advantage.”

Dave, MRP can apply with adjustments.  It’s just that wage=MRP cannot.  I don’t think that if you asked Bradbury, “does wage=MRP apply perfectly to baseball?” he would say “no, because there isn’t complete markets.” He may error on the side of assuming relatively complete markets more than I might being a search theorist by training, but that’s a legitimate possibility, and it’s a very good approximation.

I just don’t see the arguments here.  It seems like semantic arguments with someone you agree with and share similar models of baseball labor markets in mind.  If you don’t see the kind of people he’s talking about, go to mlb.com team message boards and read through the trade suggestions.  Those are typical myths and it’s good to get them out there.  I guess he’s somewhat using econ-speak, but I can’t see anything that any good sabermetrician could disagree with.

I think my mind is kinda blown that saberish stuff is being posted over at Huffington Post.

That’s all I got.

Matt, I can point out at least a couple fallacies in that article:

GMs have made mistakes in the past and will make mistakes again, but they’re not dumb enough to act on a meaningless hot/cold streak.

Are you really going to defend that?

Swapping resources only takes place if both parties are made better off.

Swapping resources only takes place if both parties BELIEVE they are made better off.  If one GM’s process and beliefs are fucked, I don’t see why we should consider the move good simply because he is a GM.  Last time I checked, being a GM is no different than being a fan, except they have more training/information/intelligence than the average fan.  That doesn’t mean that they are always right - some moves are really clearly the wrong move for all reason (like trading legimamite prospects and money to “add” Yuniesky Betancourt to a 60 win team).

Also, since when does production peak at age 29-30?

"does anybody disagree that since free agents come from teams that unless the free agents are free agents due to call ups that shouldn’t affect their value?  Yes, I can think of a thousand exceptions to that rule, but the point is that the price wouldn’t go down because there were free agents on the market directly.”

FAs are often due to call ups, so why even write this as some sort of “myth?” If a team doesn’t have an adiquate replacement ready in the minors or on the bench (or with some combination and changing possitions), they are more inclined to get players to sign extentions or pick up their option years (or players can opt out when they think the market favors them).  So yeah, I do expect the number of FAs on a market to have an effect.  Teams and players make choices based on their contract status that effects the market and the number FAs.  Simplifying the issue down to “FAs come from teams so supply will always meet demand” is silly.

Also, since when does production peak at age 29-30?

Well, speaking from personal experience, I surely ain’t what I used to be.

Matt, can you explain how the weak axiom of revealed preference explains Bradbury’s contention at all? Because I don’t see it. (And yes, I have an idea of what it is.)

I understand that I’m bringing a knife to a gunfight when it comes to arguing basic macroecon with someone like you, Matt, but let’s try it on for size for a second. Do you think the fundamental theorem of exchange applies to baseball team trades? If it was, at some point wouldn’t baseball reach equilibrium? Is there any evidence for equilibrium in the baseball labor market?

The basic idea behind Bradbury’s assertion is that the baseball labor market is perfectly efficient, that transactional costs are low to nonexistant, that information is symmetrical and that all actors are rational and self-interested. In that case, yes, Bradbury is correct. But I think that all of those are (to some extent or another) simply untrue.

It’s correct that baseball GMs are probably smarter than the typical MLB.com forum poster, and probably smarter than the typical MLB.com forum poster supposes. But it’s a false choice to say that leads to the conclusion that Bradbury is correct. I think it’s perfectly possible for both the MLB.com posters and Bradbury to be wrong here.

Nick:

The point is that you cannot just build a winner off tricking GMs.  The point is that all GMs have some private information that they are going off of, and you can’t just assume that you can trade with Dayton Moore and trick him into sending you Zack Greinke.

The point about believing you are better off was addressed in the comments section, and it’s not something that Bradbury would disagree with.  It’s semantic.  He’s seen Akerlof’s Lemons Model.  He knows about people getting sold raw goods.  That’s common economic ideas that everyone is aware of and Akerlof put a nice model to so that the ideas could be abstracted from further.  It’s a simple concept Bradbury knows.

Wally:

The point is that FA being on the market themselves (unless they are Cuban or Japanese players, for example, entering the labor market of the MLB) isn’t what cause the price to go down.  It’s the new call-ups.  That’s the issue.  And that’s just as easily associated with teams trading surplus talent.  Free agency is not the cause.  That’s true of him to say.

Colin:

Yes, it reaches an equilibrium.  No, it doesn’t look anything like the graph of perfect competition with the demand and supply curves.  It reaches an equilibrium, but it’s characterized quite differently.  It’s an evolving equilibrium with severe frictions.  That’s all I did in grad school.  Huge frictions, transaction costs, time costs, etc.  It’s real economics.  It’s just not econ 1 and it’s not Chicago School in the slightest.  It’s mainstream at this point.  Older economists lament that younger economists don’t know the math behind perfect competition, transaction cost free, perfect symmetric information.  We get taught game theory and the like, because it’s where new ideas are coming from.  But it’s all equilibria.  The fundamental idea that people are self-interested holds just fine, and it’s often the reason GMs shoot their teams in the foot trying to act out of their own self-interest.  Because 50 wins gets you fired even if it gets you the first draft pick too often.  That’s not out of the realm of economics.  It’s just not the graph of the X with the dotted lines.

The point is that all of this stuff can be modeled just fine, and we do it all the time in discussions here.  Everyone has a model of the world.  Bradbury’s is pretty clean and he sees things that non-economists don’t see the same way that a scout or a pitchF/X wiz have unique angles that help them see things clearly.  Bradbury speaks in broad brushstrokes and this article was clearly aimed more at the MLB.com poster than us, but the general points are something we’d all agree with if he’d tease them out, I believe.

Matt,

I’m an econ guy.  I’m not taking swipes at economics.  I took a very mild swipe at JC, because he does economics and sabermetrics a community by badly combining them. 

There are good ways to apply economic thinking to baseball, then there are his ways.

Matt, I am disagreeing with Bradbury’s point. I don’t think I have any trouble understanding the finer points, I just think he’s wrong. And I think so in ways that are, I think, perfectly ecumenical to economic thought.

I would argue that baseball trades (and the baseball player market in general), for the most part, resemble common-value auctions. (Obviously that’s not a perfect model - in the real world few things neatly conform to models as much as we’d like, but I think it’s pretty close.) That subjects them to the winner’s curse, among other things. This is not a unique observation, nor is it one unexplored in the economic literature:

http://www.hss.caltech.edu/SSPapers/sswp966.pdf

Let’s take a classic example - the Zambrano for Kazmir deal. (I am oversimplifying heavily here to make the point clearer.) The Rays had a poor record and Zambrano was going to get more money in the future because of arbitration, so they decided to “sell” him. If their “reserve” price wasn’t met, obviously they would simply hold onto him - the Rays would only make a trade in the event that it was percieved to benefit them.

The Mets, in this case, were one of several buyers (presumably). The Rays had an information asymetry, in that they were the only team that knew (for certain) all the bids for Zambrano. And they were free to choose the highest bid for Zambrano. It is true that the AVERAGE trade package offered for Zambrano should reflect his true value. But the Rays didn’t pick the average package offered, they picked the BEST package offered.

I hardly think that’s an isolated incident, either. Bradbury’s claims about baseball trades simply fail to describe the reality that’s observed.

JC seems to be speaking to a broader audience than the one gathered here.  Listen to sports-talk radio around this time of the year (if anyone’s still doing baseball) and he pretty much summed up what you will hear: people making completely foolish trade proposals based on tricking some other GM, and then when a trade is actually made, rushing to declare someone a winner and a loser, usually without the concept that both teams could end up winning on the deal (just in different ways).  Mostly the trade proposals are wishful thinking driven more by hope than logic, but that is the charm of the off-season.  You can dream like that.  The drive to declare a winner is just an annoying part of American culture (and usually a case of short-sightedness on the part of the caller to the talk show.)

Perhaps JC could stand to be a little more moderate in his conclusions.  There are information gaps between GMs and no GM has perfect information, and those differences can be exploited.  GMs are human, and therefore, subject to all the usual cognitive biases to which we all fall prey.  JC’s point is that within the baseball pop culture, there seems to be an assumption that those 30 men are all completely and utterly gullible (Side note: did you know that the word “gullible” is not actually in the dictionary?) fools with no clue what they are doing. 

It’s not that GMs are above critique.  No one is.  It’s that the type of critique that they normally get on talk radio is based not on a careful analysis of the data, but off of someone’s emotional reaction to whatever trade/FA signing happened.  I think JC was aiming this particular article at those folks.  And further, sometimes, you sacrifice detail for compactness knowing that you will get more effect from people reading an imprecise, short argument than from boring people to death with all the details.

"The point is that FA being on the market themselves (unless they are Cuban or Japanese players, for example, entering the labor market of the MLB) isn’t what cause the price to go down.  It’s the new call-ups.  That’s the issue.  And that’s just as easily associated with teams trading surplus talent.  Free agency is not the cause.  That’s true of him to say.”

First, that’s not how he put it.  He really did simply it to, “A high number of players looking for new teams means that there is a corresponding number of openings that teams need to fill.” This is obviously not necisarilly true, and probably only very rarely true never true.  Had JC orginally explained his point as you did, I may not have voice an objection, but what he said is clearly wrong.  That said, I still disagree with you, just to much lesser extent.

Second, you’ll have to forgive me if I don’t see players staying in the free agent market effecting the price for their service.  Many of them have the choice to re-up with their team or retire.  They often don’t, or would rather not, and it drives the supply up.  Now players entering on the back end as rookies drives demand down, that’s true of course.  However, I just can’t agree that FAs have NOTHING to do with the market price.  There are too many competing and complicating factors to make such a blanket assumption.

Now I’m a biologist and not a economist, so let me put this slightly differently.  What you’re attributing to the cause of market changes is the consumption and creation of a certain product (the derivative of the concentration if you will), not the pressence of the product itself (the concentration itself).  To most people this is a distiction without a difference.  The loss of a FA through retirement is still a FA, so the FAs effect the market.  So while I know what you’re saying and understand it, JC did not get that across.  He just left it as the concentration and that’s all.

His whole post is riddled with false choices, strawmen, poorly-defined criteria, and some grandiose concept of objective reality.

If we look at “Every trade has a winner and a loser.” He doesn’t define winner and loser in this case. He is setting up the strawman of a zero-sum game. Nobody is arguing that a trade cant add wins to both teams, but it still doesn’t remove the concept of superior return. If 2 teams swap excess depth to fill positions of need they can both have better teams. BUT one team can still win that trade. In fact I would argue it would be impossible for a trade not to have a winner and loser, unless both teams exchange players of EXACTLY equal relative value. Believing you are getting value back doesn’t make it so.

“old players are worthless” Yeah that sounds like a myth that needs debunking. Lazy fallacious hyperbole.

The most preposterous claim of the piece has to be “You can’t sell high or buy low and profit financially because all GMs understand these things. You don’t have to wait for a guy to get hot to sell him, nor dump him before he gets cold” Because if their is one thing we’ve seen over and over again markets are always perfectly valuing assets! Is he attempting to disprove the concept of market inefficiency? I think he underestimates the length and depth a cold or hot streak can go. JJ Hardy, eh?

Bradbury’s point seems to be that message board commenters and talk radio folks tend to oversimplify and distort matters. This is true but not interesting. In a wider context it does make sense that such an article would appear on Huffington Post. It would’ve fit the outlet even better had he done something like call somebody at BPro a “birther” for questioning Albert Pujols’s age.

I think Pizza Cutter expressed what I was trying to say best.  The article was at Huffington Post and aimed at a broad audience.  It was intended to comment on radio call-in stuff.  We’ve all learned this stuff already, and agree with him.

David, I had thought you were an econ guy but your previous comment had made me think you were saying that you can’t applying academic economic models to baseball.  You can’t apply the big perfect competition graph with the X, but I think you really can apply it.  I think Bradbury is wrong sometimes certainly, but this all seemed pretty safe stuff.  Especially reading the comments section over at sabernomics and the expanded points, it all seemed pretty safe.

Colin, I’m interested in that Cal Tech paper.  Camerer always does interesting stuff.  I heard him speak once and was very impressed.  I do think that the average bid for Zambrano would be lower than his average value, though, because people respond to the winner’s curse by lowering bids.  That got Steve Phillips fired among other things, didn’t it?  I mean, the optimal bid is to underestimate the value based on the probability of winning and the winner’s curse should reflect that.  But Camerer may find something-- I’ll look at the paper-- that finds flaws in that thinking.  That’s pretty cool too.

Wally, I think in the comments section JC clarified what he meant.  It’s a good point when explained.  It wasn’t entirely clear to start, but I think it became pretty clear.  But it’s a blanket statement that people make sometimes and it’s fair for him to say that’s not the direct cause.

Phil, I guess that if you redefine winner and loser of trade to be who gained the most, you can say that.  But if you trade an ace to a team that uses him to win a couple World Series, but you get yourself a nice couple of prospect that help you win one World Series five years later, I’m not sure you can say that the first team won because they get two WS victories out of it.  The point is that there can be gains from trade as teams have different values.  People on talk radio clearly ignore that all the time.  Also, it’s not markets perpetually value assets accurately.  It’s that it’s very difficult to repeatedly identify the same market inefficiency over and over again and keep being able to draw from that well.

I think Bradbury’s old discussions of replacement level were not even really dismissing them,

Matt: That is 100% incorrect.  You obviously missed the 12MM Francoeur thread, where JC made his last appearance in this blog.

The one thing about JC is that he speaks for himself and he speaks very clearly, not to mention that he has a blog to clarify himself.

Matt’s interpretation of what JC really said is not even open to interpretation.  JC couldn’t have been clearer.

"I think Bradbury’s old discussions of replacement level were not even really dismissing them, but overreacted to and then turned into an argument where no one actually disagreed with anyone in reality.”

Matt, that defense of Bradbury makes me think you haven’t read the threads here or on his blog about the Francoeur valuation.  Quite simply, we are not on the same page, and so far off that it got to the point that further discussion (which he wanted no part of) was pointless.

The valuation models he used would show that a player who went out to right field every day and hit like a pitcher would have economic value to a team.  If you use any sort of replacement level theory, that is not the case, such a player would have negative value compared to the guy waiting for a chance in AAA.  But Bradbury would not hear of it.

Matt and Pizza:

As usual when I read Bradbury, I don’t get the impression that he’s going after message board posters.  I get the impression that he’s going after sabermetricians (witness the throwaway swipe at replacement level, or the comment about old players.  I get the impression that sabermetricians are much more concerned about age than the general fan is).  I think he believes that the myths he’s referring to are being put into practice on forums like The Book Blog, not just on talk radio and general message boards.

Now perhaps you guys are right, and he’s not talking about serious sabermetric analysis of these issues.  But JC has a track record of taking shots at the general sabermetric community, and it’s hard for me to read a piece like this without assuming the worst.  This possible bias on my part is in addition to the fact that, as Tango and Rally have expressed in 26/27, JC and the folks here do not have an agreement on player valuation.

By the way, JC posted this on his blog before he cross-posted it to HuffPo.  I think its quite the stretch to say that this was written for the message board/talk radio crowd.  Those people don’t read Sabernomics, and honestly, they don’t read HuffPo either. 

I’m all for giving someone the benefit of the doubt, but in this case, it seems pretty clear to me that this was JC taking another shot at the sabermetric crowd.  Or, basically, what Patriot said.

I have to disagree with Matt S. and Pizza and side with the majority here.  Once upon a time, J.C. provided good value to the baseball community back in the day when he wrote his article about strikeouts and DIPS.  I haven’t seen him contributing to knowledge about baseball in a good long while, certainly since Colin took him to the woodshed over MRP.  He’s propagating false ideas through the vehicle of being an educated economist, and he seems to take pride in the scorn he gathers from other baseball subject matter experts.

He’s below replacement value as a baseball analyst at this point.  The world of baseball ideas has grown, and he refuses to grow with it.  Of course he probably thinks that as long as he continues to write, that his production has marginal value to the baseball community by its very existence regardless of its quality.

"He’s below replacement value as a baseball analyst at this point.”

JC would be fine with that, though, as he’d still deserve a large pay check.

Matt: Additionally I think his point is circular, inasmuch he claims trades don’t have winners and losers because they wouldn’t be consummated if they did. Can a trade have a winner/loser? Only as long as it hasn’t happened. The moment a trade is finalized both teams win because they got returns. That idea borders between delusionally idealist and purposely misleading. Also I think he’s is missing some big differences between a market writ large and the baseball market which operates with a set of different rules/restrictions. By his circularity he can’t say any GM is better than any other, they are constantly justified by their position of authority. They know more than you, therefore they can’t lose.

5 stages of grief:

1. denial
2. anger
3. bargaining
4. depression
5. acceptance

Life is good at #5.

I didn’t have a big problem with JC’s first 3 points.  Assuming they were in fact directed at the crude second-guessing of teams/GMs made on many message boards, they provide valid counterpoints.  They are obviously stated as generalities, and at some level are truisms, so I won’t quarrel with anyone who says they aren’t very interesting.  But I wouldn’t say he’s “wrong.”

But I’m surprised no one has commented on the one point on which he is unambiguously wrong, which is his last one:  that players peak at age 29-30.  This is based on a journal article he published, purporting to show that both hitters and pitchers peak at age 29-30, not the 27-28 that virtually every other study has found.  (Unfortunately, there doesn’t appear to be an ungated version of the paper available anymore.) It’s really quite extraordinary:  he claims that hitters are basically as good at age 32 as at age 27, and as good at 35 as they are at 25. 

This is quite clearly not true.  But it’s what your regression tells you when you start with a sample of players who have at least 5000 PAs from age 24 to age 35, which virtually requires that they remained a productive player in their mid-30s.  It’s like studying the impact of family income on SAT scores, and studying only kids who scored over 700—you’ll find a weak relationship, just as JC finds an extremely flat aging curve.

This thread is worthwhile for, if nothing else, Pizza’s side note about the word gullible and Tango’s #33.

Matt, and for those who missed it, here is the famous Francoeur/JC thread:

http://www.insidethebook.com/ee/index.php/site/comments/francoeur/

JC makes an appearance.

Matt, from post 13 of the linked thread:

JC said this in the comments as well:

“If there was one concept that I could remove from sabermetrics it would be “replacement level.” It is a concept that adds nothing to our understanding of player value, and it causes people to believe that marginal players are worth the league minimum. I have explained why this is incorrect, and I have seen no argument to change my thinking on this.”

Tango:  FWIW, JC appears to have revised his valuation model.  He now says Francouer is worth much less, and acknowledges his earlier calculation was incorrect.  He hasn’t described the new model (that I know of), so it’s hard to evaluate other than asking if his newer valuation estimates seem to make sense.  Don’t know whether he’s changed his mind on replacement value (though I’d be shocked to hear him say so).

Guy, correct.  Colin made a post at the end of the linked thread to that effect (revision in valuation model).

***

If Nixon could go to China…

So now Francoeur is worth 6 million instead of 12?  That’s still about an order of magnitude wrong (in some subjects that’s perfectly acceptable, but not here).  My gut says JC’s model is probably assuming absolute zero production (having a scarecrow bat and take the field) is worth zero dollars.

I’m all about making models, but your models have to reflect reality for people to care about them.  So while JC’s model might illuminate some sort of principle (player’s are not efficiently allocated or something, I don’t know all the jargon), it doesn’t really matter.  The system we’re operating in is what it is, and you have to make your model to reflect that.

He’s at it again:

http://www.baseballthinkfactory.org/files/newsstand/discussion/chop_n_change_interview_with_jc_bradbury_economist_and_sabernomics_blogger/

Matt: That Bradbury response is why you might perceive “hate thrown his way or the way of economics”.  If it exists, it’s a response in kind.

Bradbury (and some others in the academic community) refuses to acknowledge the fact that there are people outside of economics and academia that are experts on baseball and might be capable of critiquing research focused on it.

I have posted a response to JC’s interview.  Feel free to use that thread:

http://www.insidethebook.com/ee/index.php/site/comments/jc_v_tango/

Guy, from what I recall, there was some evidence that during the steroid era the peak age for position players may have been closer to 29.  But other than during the steroid era, yes, the peak age for offensive players is clearly around 28 or even a little younger.

There is only one way to do it (figure out peak age), and the way JC did it is not the way.

If JC is writing FOR the average message board poster, they would obviously have no idea what he was talking about.  They couldn’t possibly understand his discussion, whether he was right or wrong.  The idea that he is writing for that audience is preposterous.

More importantly, though, regardless of to or for whom he is writing and regardless of who is propagating those myths, his discussion and his comments are filled with all kinds of falsehoods and misconceptions.

Talking about what audience he is directing his comments towards or what segment of the population he is drawing his “myths” from is a complete red herring, as those things have no relevance to whether his comments are accurate or not.

Bradbury is a pseudo-intellect.  So what?  There are tens of thousands of them.  He just happens to have some notoriety in the sabermetric community. If not, he would be just another Anne Coulter or Cal Thomas.  I’m sure he has lots of knowledge, credentials, and expertise with respect to economics.  That does not make one a smart person nor does that make one an expert in other fields that may or may not be related to economics.  It doesn’t even make one qualified to speak authoritatively about some or even all aspects of economics.

MGL

From #44, what is the right way to figure out peak age? Would you post a link to a good study on that?

Part of this discussion has been about trades. What are the good studies on trades that address some of the questions here like do both teams benefit and are some GMs better than others?

Cy

"That does not make one a smart person nor does that make one an expert in other fields that may or may not be related to economics.  It doesn’t even make one qualified to speak authoritatively about some or even all aspects of economics. “

The key is knowing what you don’t know.  Then you don’t get into this kind of thing and come off way smarter.

Cyril, for peak age, simply take the difference between one age and another for all players and get a weighted average of all those numbers for all age intervals.  This is the “delta method.”

How you weight them is a matter of contention.  One way is to weight by the lesser of the two PA (for each age pair).  Another way is to weight all age pairs and players equally.

The biggest issue to deal with is the selective sampling issue that occurs when a player only gets to play in year X if he played well (and got lucky) in year X-1.  That especially applies to young players and players near retirement.  There are a number of ways to handle that, none of them being perfect.

I think Tango has a nice explanation of this “delta method” on his web site.

http://www.tangotiger.net/AgingSelection.html

So, for example, if you have 3 players who each played for 10 years:

For all 3 players take their difference between age 22 and 21 (assuming they all played at those ages) and average all 3, weighting by the lesser of the PA for each age pair.  So if player A had 100 PA at age 21 and 200 PA at age 22, the weight for his difference would be 100.  If player B had 250 PA at age 21 and 150 PA at age 22, then the weight for his difference would be 150.  Technically, rather than the lesser of the two PA, you can use the harmonic mean of the two sets of PA.

Anyway, you get a “delta” for all age pairs.  Plot those numbers on a curve and you get a graph of age versus performance for an average player in the major leagues.  You will get around 27 years of age for the peak, give or take a year.

It is possible of course that the best players or the players who get the most playing time have a true peak of 29 or 30 since using this method, you are figuring the peak age for the average of ALL players, weighted by playing time (which puts more emphasis on players who are good and play more of course).  But if you only limit your data sample to players who are either good or play a lot, as JC did, you really get a selectively sampled group of players who not only may have a true peak age of 29 or 30, but who probably peak at a later age by luck (because of the selective sampling effect) as well.

MGL: 
Bradbury’s aging paper is based on 1921-2006 data, so not limited to steroid era. Unfortunately, it is no longer available free online.  But to give you a sense of his findings, here is his predicted OPS in terms of percentage change from a player’s peak:
22 -7%
23 -5%
24 -4%
25 -2%
26 -1%
27 -1%
28 0
29 0
30 0
31 0
32 -1%
33 -2%
34 -3%
35 -5%
36 -7%
(Linear weights peaks at 29 and 30).
My view is both that the peak is too late and the post-peak curve is way too gradual. 

In terms of a recent rise to 29, are you thinking of Ray Fair’s paper?  Or something else?

MGL

Thanks I will take a look at that. What do you thing are the best salary studies out there?

Cy

Okay, so after coming home from work and finally reading through all of it, I think I see the problem.  Bradbury made some mistakes in recent years, and everyone interpreted his relatively tame post as aggressive or wrong.  I’ve disagreed with Bradbury at times.  I hadn’t really remembered the replacement level post all that well, but reading back through the parts where he talked about it here, it just seems like he’s being defensive.  I don’t think Jeff Francoeur is worth $12MM, but I think Bradbury makes strong points sometimes and uses economics effectively.  He also takes sabermetrics more seriously than other academics.  But regardless, these are all historical things.  He made good safe points here, that are true if you don’t try to interpret them a different way.  I recognize he’s made mistakes before, but I just don’t see this article as a mistake, especially given his explanation.  I think the reaction to this post is clearly more about the replacement level issue than anything, and it shouldn’t be.  That makes a lot of the attacks here ad hominem attacks.  And I can see the frustration when someone using their real name is subject to ad hominem attacks by people who don’t.  I don’t really think his reaction was appropriate, either.  I’m not sure I really agree with either side here.  I do think that Bradbury’s outrage at somebody saying “Tangotigers hates the H&S model” is completely justified, and I would think Tango would find that to be a ridiculous statement too.  Bradbury has a unique perspective, and it helps him.  Sometimes having that background does.  The typical model used by sabermetricians to value players is used in-season sometimes and that’s just wrong because the value of a player changes when the odds of X wins affecting their playoff birth can go up.  That’s a perspective that I came to with an economic background, and I think it’s important to criticize the models.  The same perspective also leads me to erroneous conclusions at times.  Bradbury’s perspective leads him to erroneous conclusions at times.  Also, that poster responded further by making an absolute attack on economics and economists in general, and made his point by citing three papers by Steve Levitt, the most controversial and probably the most disagreed with economist in our era.  We’re talking about someone who was not invited to Penn’s Econ Department on principle and who I don’t even know if I can get a fair perspective on because his papers were treated as straw men in my classes.  To dismiss an academic discipline as invalid when looking at baseball and without unique insight at looking at baseball is just as stupid as dismissing sabermetricians.  Bradbury does that better than Levitt, as near as I can tell.  He’s right when he says he does it better than most.  He’s out of line sometimes, but I don’t get the “responding in kind” like that somehow makes it better.

I do think that Bradbury’s outrage at somebody saying “Tangotigers hates the H&S model” is completely justified, and I would think Tango would find that to be a ridiculous statement too.

No, I don’t think it’s justified at all, because that person provided a link to why he was agreeing with me.  I don’t know what the reader said that was ridiculous.

That reader did not offer unsubstantiated opinion (i.e., b.s.), but a qualified opinion.  JC admitted to reading my thread.

***

I really don’t have much opinion about JC’s article at HuffPo.

I hadn’t really remembered the replacement level post all that well, but reading back through the parts where he talked about it here, it just seems like he’s being defensive.

If you go read the conversation at Sabernomics on that topic, you will see that J.C.’s basic response to the issue was “La, la, la, I have an economics PhD and a book and you don’t, so I can’t hear you.”

I think the reaction to this post is clearly more about the replacement level issue than anything, and it shouldn’t be.  That makes a lot of the attacks here ad hominem attacks.

I’m with Tango in that I don’t particularly have a strong opinion one way or the other on the myths JC was exposing and the evidence he used in the HuffPo piece.  I do have a strong opinion about his behavior during and since the Francoeur discussion.  I guess you can call that ad hominem if you want, since it is personal rather than about the evidence.  But my issue with Bradbury is personal.  He handled the Francoeur matter very poorly, and since then he’s taken every opportunity to make coy insults against my friends in the sabermetric community.  Discredited on the evidence + acting like an ass to everyone = the respect that should properly be accorded a PhD?  No.

And I can see the frustration when someone using their real name is subject to ad hominem attacks by people who don’t.

His main critic (Colin Wyers) uses his real name.  That didn’t garner him one iota of respect from J.C.  I’m using my real name.  For many of the folks around here who uses pseudonyms, their real names are easily available, they have blogs and email addresses where they can be corresponded with, etc.  Criticizing Tango as somehow suspect because that’s a pseudonym is poor form at best.  Tango is the most widely respected sabermetrician there is.  He’ll answer your email.  He has a blog where you can correspond with him.  He works for the Seattle Mariners.  He’s published a Book where you can examine all his models and methods.  What more does J.C. want, really?  It’s ridiculous

To dismiss an academic discipline as invalid when looking at baseball and without unique insight at looking at baseball is just as stupid as dismissing sabermetricians.

That’s true.  But the two economics professors who routinely comment on baseball player valuation, Bradbury and Fort, don’t do a very good job of it and come across as very arrogant and close-minded.

Contrast that with the discipline of physics where the professors who are involved in discussions about baseball (Nathan, Bahill, Hubbard, Adair) have been leaders in the field, easy to work with, and willing to learn from others.

Cyril, sorry, I don’t follow the salary model issues very much at all.  Tango would know more about existing research in that area than I.

Guy, I agree that the post-peak numbers look too small (too gradual a decline).  Again, if you presuppose a minimum number of years or career PA, of course you will get more “encouraging” numbers (later peaks and more gradual declines).

I vaguely remember that I used the “delta method” on players in the last 20 years or so, and came up with a later peak than when I used more data (more years); thus I came to “conclusion” that something, perhaps PED use, may have changed the dynamics of aging versus performance. 

Unfortunately the data set from the “steroid era” is not large enough I don’t think, to reach any conclusions with much certainty.

Plus, it could also be that in the last 20 years or so, the selective sampling issue has changed as well.  For example, if for whatever reasons, older players with bad years or younger players with bad years are not allowed to play much or at all in the subsequent year, as compared to in the “old days,” or vice versa, the aging curves might look steeper and the peak might change.

In any case, there could be reasons other than PED use that could change the texture of the aging curve, given the method ("delta") we mostly use to develop it.

Tango

Since salary studies are a concern here, can you suggest some that are good? Thanks

Cy

Tango:

I didn’t realize there was a link in the post and can’t find it between all these threads and different links.  Regardless, it’s not wrong because you say so; it’s wrong for the reason you say so. It’s wrong because it treats PA linearly and separately.  The attitude is wrong.

Mike:

I agree he handled the Francoeur thing poorly, and I think it’s fine to hold a grudge.  I just didn’t really remember the Francoeur thing, and this thread seems really, really out of place otherwise.  I didn’t get the outrage, and now I understand it has to do with Francoeur and now hot stove myths.

I personally don’t mind people using fake names or real names.  It’s not a big deal.  I used initials for a while on message boards.

BTW who is Fort?  It doesn’t really change anything, because he’s not face of economics either and Nathan has been very nice the time I talked to him but I’m sure there’s mean physicists out there too.

This thread just doesn’t feel like it’s about what people are talking about and I’m just finding myself confused in what is clearly a previous feud that I didn’t have anything to do with.

I still think all four of his myths listed in the actual article linked at the beginning are mostly true with some exceptions and I don’t get the impression he’d disagree with those exceptions.  I guess I need to read his book, the Francoeur argument, and a hundred other things to join the argument I’m not really a part of.

MGL

A few years ago I posted something called “Player Aging Patterns Over Time” One thing I found was that the % of players 36 and older who reached 400 PAs over time seemed to rise alot in the last 30 years. I don’t think I know why. 

http://www.beyondtheboxscore.com/story/2006/9/8/144018/2113

There was a part 2 at

http://www.beyondtheboxscore.com/story/2006/9/15/84847/5713

It seems like players are aging better than they used to in terms of playing time and performance.

Cy

Cy: the best one that has been published is by Studes at THT 2007.  (I think it was 2007).  It is by far the best thing out there, and was the initial basis for my salary model.  Look for “Net Asset Value” I think he called it.  He also has an article at THT, and that book I believe he said is available online.

I also have referenced that article here in this blog, so you can find my thoughts on it here as well.

That is an example of something constructed by an SME.

Tango

Thanks.

Cy

"But the two economics professors who routinely comment on baseball player valuation, Bradbury and Fort, don’t do a very good job of it and come across as very arrogant and close-minded.”

I think it’s a stretch to say Fort ‘routinely’ comments on player valuation.  He had a back and forth with Birnbaum about VORP, but I haven’t seen much in that area otherwise.  Not sure where it routinely comes off as arrogant.

Otherwise:

I think Matt makes some good points here.  The attitude toward an extremely tame article seems to have skewed the view of its actual presentation.  Rather than focus on the big picture, commentors seem to be directing their criticisms at semantics in the article.  I don’t find it misleading, and the general conclusions are ones I would think most people in this community would agree with.  At this point in the game, it’s not outlandish to think you can continuously get swindled by everyone else AND keep your job.  Is that so outlandish to state in a news article to a more general public?

Is JC extremely vocal?  Yes.  Is Tango?  Yes.  Does that mean one is always right and the other always wrong?  I don’t think so.  Bradbury is extremely opinionated, as are many people here, but I think his response to the notorious ‘commentor’ (now posted in the comments at Chop n’ Change) was extremely reasonable.

I’m not trying to take one side or the other here.  I’m sure there are economists that disregard sabermetrics (wrongly), and I’m seeing there are plenty of saber-minded people that seem to think economics have no application to baseball (that’s just silly). 

I think Matt does a fine job in explaining why JC’s generalizations about Hot Stove myths make sense.  I think people here respect Matt as a saber-minded folk.  Of course there are anomolies, which is why an economist’s answer to just about everything is “well, it depends”.  There’s always a “yeah, but”; however, those are usually a smaller sample size or multiple anecdotes...in which case I would think people here would be sympathetic to the case of “as a whole and on average”.

As for the pseudonyms, I understand where that can cause problems.  I don’t have a huge problem with it, and I don’t think the Tangotiger name is a problem, given the knowledge here at this site about the name.  However, many of the names can lead to having no reprocrussions for statements made (I would say mostly at a site like Baseball Analysts).  A name ties responsibility to things said, and some of that can be lost in a pseudonym and it’s something I’ve wrestled with myself even using a simple one like Millsy.

Sometimes, I do get a feeling of blind trust in some of the work done on internet sites.  That’s not to say much of it isn’t interesting or worthwhile.  I think even Bradbury even admits there have been numerous advancements through the sabermetric community, and he has embraced these in much of his work.

I don’t know, just my two cents on everything.  Most of the people involved are very intelligent.  Of course there are people without a PhD that can analyze things critically.  The problem, I think, lies in the fact that many who read it don’t have the knowledge to critically analyze conflicting analyses (getting redundant here).  I think that really was exemplified by the “commentor’s” misguided response at Chop n’ Change.

"At this point in the game, it’s not outlandish to think you can continuously get swindled by everyone else AND keep your job.”

Addendum to this: “It IS outlandish to think you can continuously get swindled by everyone else and keep your job.”

Woops.

Addendum 2: I meant BTF, not Baseball Analysts above.  It’s late for me…

"I still think all four of his myths listed in the actual article linked at the beginning are mostly true.”

Matt, I really don’t think that’s true for the 4th myth, that players peak at 29-30 and decline very slowly afterwards.  That finding is a function of very severe selective sampling, where JC looked only at players who played a lot into their mid-30s.  But players must have a variety of aging curves, with some guys declining more rapidly than others.  And if so, those players will be severely under-represented.

And in fact, many players are productive full-time players in their 20s, but decline rapidly and are bench players or out of the game by age 32 or 33. For example, here are a few recent players who logged 5000 PA (JC’s minimum) from age 22-33, but not from age 24-35 (and thus would be excluded from his sample): Derek Bell, Carlos Baerga, Dean Palmer, Gregg Jefferies, Jeff Blauser, Benito Santiago, Ruben Sierra, Lenny Dykstra, Mike Greenwell, Alvin Davis, Jesse Barfield. If you include players like this, you will get an earlier peak.

Even using the delta method, as MGL suggests, you still face this problem.  When you measure the decline from age 33 to 34, you can only look at the guys still playing.  Those who started dropping like a stone at age 30 or 31 are gone. (If you’re deciding on a 1-yr contract for a 33-yr-old, that doesn’t matter.  But if deciding on a 5-yr contract for a 29-yr-old, it does.)

If you reduce the PA minimum to 4000 or 3000 PAs, what you find is that the average starting age grows only a little, but the final year age drops a lot.  Shorter career players are not symmetrical around age 29-30, but rather 27-28.

Guy, that’s a fair point.  Not really my area and I guess I only focused on the second half of the statement.  But the aging thing is a fair point; 30 does seem old for the average peak.

For those interested in aging studies, there’s a bunch of links here:
http://www.insidethebook.com/ee/index.php/site/comments/guym_on_jcs_aging_study_at_phils_site/

And I strongly recommend where I have a link near where it says “payoff chart”.

Tango & MGL

How about if you guys write another book with articles for a general audience on aging, WAR, salaries and trades?

Cy

Cy:

One book was enough for me, though I appreciate the interest.

I think the best setup is the “anthology” type of setup in the Hardball Times Annual.

Is a player most likely to be in MLB or be a full-time MLB player when he is at his peak?

Tango

Okay. I think the reason I suggest that is that there are so many studies out there on the web and people mention them, sometimes without links, and you never know what the comments were on it elsewhere. Some articles are alot more technical than others, so it can sometimes be hard to keep up. It was so much easier in the 80s when all you had to do was read one Bill James book a year

Cy

Bradbury’s claims on aging also ignore fielding and position. Say you take a 22 year old who’s a good-fielding center fielder. Is he likely to be a better hitter by age 28? Yes. But is he likely to still be in center field? And if he is, do we think he’ll field the position as well as he did at 22?

Now of course this falls in a tremendous blind spot in Bradbury’s player evaluation methods (he still, so far as can be discerned, does a bad job of adjusting for position - and if he has improved in this regard he hasn’t publicized it).

To Colin’s point in 70 - overall productivity does peak aorund 28 or 29 (using the delta method) but different skills ahve different curves.

Speed (measured by triple%, steals, infied hits%) peaks about 21. I haven’t specifically looked at defensive aging yet, but I would find it a fair assumption that range would start declinging very early as well.

BABIP peaks around 26 - speed it decreasing, while power is still increasing at this age.

HR% peaks later, 29-30.

BB% is the latest peak, about 32, while SO% minimizes around 27-28.

So before 28, speed is decreasing, but power, contact and strike zone judgement is still getting better.

Does anyone remember that Kevin Mitchell came to the majors as a shortstop? In less than 10 years he morphed into that generation’s Pat Burrell.

I would guess overall productivity peaks closer to 27 than 29, since hitting peaks at 27-28 and both baserunning and defense likely peak earlier than hitting.

Even using the delta method, you have a big problem of selective sampling in figuring out the post-peak aging curve.  Here’s an easy way to tell:  calculate the number of player seasons at some high performance level, then check the age distribution.  Let’s look at seasons of OPS+ of 130 or more, and 500+ PA.  If Bradbury is right, these should be about as common at age 26 as at age 33.  But there have been 203 at age 26, and just 94 at age 33.  Now, career-ending injuries might explain some difference.  But this gap is huge.  It’s just preposterous to say players are, on average, as good at 33 as they were at 26.  (And this doesn’t even include defense).

I took a quick look at the age distribution of the best 900 RCAA seasons since 1920. Runs created above average from Lee Sinins. It is park adjusted (it actually turned out to be 920 guys due to a tie for last at 44). It is a counting stat or raw total, not a rate stat, so no PA minimum. As guys age and have a harder time playing every game, this will go down at some point. Here it is. Not sure what it means. Maybe it will make sense to someone or someone can explain if it means anything. Seems like a nice structure except for age 27.

20 ** 5
21 ** 13
22 ** 16
23 ** 47
24 ** 50
25 ** 71
26 ** 89
27 ** 82
28 ** 94
29 ** 87
30 ** 72
31 ** 71
32 ** 71
33 ** 50
34 ** 31
35 ** 33
36 ** 17
37 ** 10
38 ** 5
39 ** 3
40 ** 1
41 ** 1
42 ** 1

It will be interesting to see in the future what kind of sabermetricians get hired by ML teams. The ones with college degrees, taught by the academic researchers or the ones from the inernet/blogosphere who teach each other and/or are self taught.

Maybe those groups will or have started to overlap. Will there be problems in classrooms when a prof teaches something that conflicts with what the blogosphere thinks. What will profs say when a student says “but so and so on the web says the opposite?”

Does anyone remember that Kevin Mitchell came to the majors as a shortstop? In less than 10 years he morphed into that generation’s Pat Burrell.

Brian, a small correction in this vast array of critical comments: Mitchell didn’t come to the majors as a shortstop.  Davey Johnson played him there at times, even then he was out of place.  It was fun to watch, though.

On the other hand, that has nothing to do with your larger point.  Just couldn’t resist…

Guess that’s why nobody remembers it. Checking B-Ref, yes Mitchell was a 3b all the way thru the minors, but did start 20 games at SS in his first full season in the majors. I guess that’s what stuck in my mind.

Point is, players bulk up, slow down, lose range and move down the defensive spectrum.