THE BOOK--Playing The Percentages In Baseball

Unfortunately, sabermetric writing doesn’t pay very well.  It’s actually something I love to do, but I spend far less time at it then I would in a “perfect” world.

Luckily, there is still some good sabermetric writing out there.  Probably more than there ever has been, in fact.  But there is also much more chaff to go along with the wheat.

Here’s what I wrote on Facebook in response to John’s status:

I tend to skim over or ignore a lot of stuff on run estimators, player valuation, in-game strategic decisions, roster construction, and forecasting for that reason. Not all of it, of course, some of it still interests me, particularly from the very best authors.

But I’m still fascinated by all the advances in defensive evaluation (including ... See Morecatchers), digital scouting, detailed hitter and pitcher analysis, batted ball studies, and umpire evaluation. I really love anything that gets down to the nuts and bolts of the pitcher-vs.-hitter battle, and we’ve had a profusion of that and an insane pace of advances in the last few years.

I don’t really give a rip about reading about SIERA (haven’t read any of the five pieces), but I do enjoy reading Tango’s dissection of SIERA.  Go figure.  To each his own, I suppose.

I think John has a point, and it’s a valid point. But I’m not sure what anyone is supposed to do about it or will do about it.  There are so many people writing about sabermetrics these days and so many high-profile sabermetric or saber-friendly web outlets that there is no one who even begins to control the voice of sabermetrics.  There will not be another Bill James of the 1980s because the amount of information now available blows the 1980s out of the water.

I do wonder what prompted John to write what he did.  Was it the SIERA series at BP?  Something else?  Yeah, there’s a lot of stuff going around baseball nowadays talking about regressions, and that kind of stuff is getting a lot of prominence lately at BP, but what else is there that is using advanced math?

I think that’s an excellent critique, Tango. However, I almost wonder with which metrics he struggles? FIP, tERA, and xFIP (and even the young kid, SIERA) are all scaled more-or-less to ERA (the presentation), yet each tells us something slightly unique (the presitige?). On the other hand, wOBA is scaled close to BA, and wRC+ is based on one of the easiest scales to understand: the base-100 format. For total value, we use wins (pretty easy to interpret) or FA value (which is in dollars).

Therefore, I think Sickels’ true struggle is not with the metrics—many of which are intuitively simple—but with the processes with which we create and sometime use these metrics. In that context, I should HOPE a common fan wouldn’t be able to understand the language or the methods—because the common fan doesn’t know enough of statistics or physics to truly appreciate the involved processes. The common fans, therefore, should look to the conclusion, partially believe what they have read, and resist to do as their predecessors did: use limited, anecdotal observations to deny what a sabermetrician has determined through much more stringent work.

Instead, readers of the Sickels-vein (though Sickels is certainly more than just common fan) should avoid the scientific lingo if it displeases them, but they should not dismiss the results if it treats them the same.

I don’t understand why the blogosphere should be atwitter. John is simply stating a preference - that much of what he reads now is not interesting to him.  Some people want to get down to the quarks and strings...John is happy understanding the atom and that level of knowledge will suit him fine. Why is that controversial? Different strokes and all that.

"Blame yourself”? Shouldn’t it be “Blame the person writing it”? That is a more logical conclusion based upon what you wrote.

And ‘blame’ seems harsh. Like many disciplines, many of the people writing “sabermetric” articles are not good writers. Maybe that has to do with the fact that a lot of them are “math guys” who either don’t have a lot of experience writing or don’t have a lot of interest in writing or maybe simply lack the ability to write well.

Being able to explain to someone else what is so interesting/important about what you’re doing is, to me at least, an essential aspect of what sabermetrics is about.

I can definitely understand these feelings.  I think there are basically two types of sabermetric articles written these days: 1) technical pieces explaining a metric or the value some metric provides and 2) pieces that analyze a player or team using those metrics.  I don’t have the technical skills necessary to understand a lot of what’s being done with regards to point #1 these days, but, if people whom I respect that DO have these skills vet and approve of a new metric, I will do my best to incorporate its appropriate usage into my analysis and understandings.  It seems (and this is entirely perception), that the ratio of articles in category 1 to category 2 has skyrocketed this offseason.  My guess is that Sickels is just looking for more baseball (and less technical) analysis.

I think part of Sickels frustration stems from the pre-spring training blues.  If he lives anywhere near my neck of the woods (Baltimore), the waist high snow drifts have warped his mind.  If I see one more SIERA article, I might scream.  I really can’t wait for games to start again, and for analysis to shift to slumps, hot streaks, breakout players, and pitching changes.  We all love the baseball, so those first images of millionaires doing PFPs will be very welcome.

Sabermetrics is a small field and there aren’t many jobs in it, but if you are making a living at it then you must keep up with new developments even if the reading is difficult. For most of us this is a hobby, and hobbies are pursued for enjoyment. I don’t see anything wrong with reading only the stuff that holds your interest. If you bought a boring book would you feel compelled to finish it just because you purchased it? Or would you say to yourself “I’m wasting my time, not getting anything out of it, I’ll go read something else”?

“Blame yourself”? Shouldn’t it be “Blame the person writing it”? That is a more logical conclusion based upon what you wrote.

You can’t blame a hobbyist for being involved in his hobby, because he can’t reach out well-enough.

Not unless the hobbyist himself is upset that others aren’t turned on as much by his hobby.

Great post. Unfortunately I got a similar feeling reading the SIERA posts, and I understand the math. I think where John is right on is that, in some cases, the guys doing the analysis lose sight of the love of the game that fuels the hobby.

If your passion for what you’re presenting/analyzing/writing about doesn’t show through, much if not most of the appeal is lost.

But don’t blame the stats.

Off topic, but in an interview with Batters Box Alex Anthopoulos said Tango will be a statistical consultant for the Jays this season. I just wanted to say on behalf of my fellow Jays fan welcome aboard and we’re ecstatic to have you, even if it will be in a limited role.

I thought it was a very strange article to be honest.  I first saw it through THT’s take(s) on it, so I was a little biased when I first read it, but on second read it just seems like John isn’t articulating his points very well. 

He mentions that certain aspects of sabermetrics aren’t enjoyable to him.  I agree.  Similarly to Mike Fast, I’m not going to read all of the articles explaining SIERA or some of Patriot’s stuff with Base Runs and calculus (no offense Patriot). 

However, he uses Pitch f/x as an example of that, which is just strange to me.  It’s strange because Pitch f/x is obviously something with limited, although not not-existent, applications towards prospect valuation, which is John’s niche.  However, I’m not sure why that should be an indictment of the sabermetric work with Pitch f/x.  It’s the same with Patriot’s work.  I personally don’t care a whole lot about how to Base Runisize SIERA to make it slightly more accurate; however, I recognize that it is being written for a very specialized audience, and so I shouldn’t get to worked up over not really enjoying that kind of stuff. 

Mitch/5 - If that’s your impression of Sabermetics, than you really need to read more stuff.  I don’t know how you could possibly gather that the only stuff being written about is technical articles on creating player valuation metrics, or other people using those metrics to value players.  Just look at the current THT staff.

I am writing primarily about Pitch f/x, and it’s application to jduging variance in a pitcher’s performance.  Harry is looking at how batted ball numbers translate from the majors to the minors.  Max is looking at which minor league outfielders have the best arms.  Josh is ranking teams based on their fan happiness, and ranking players and GM’s based on their value.  Jeff is looking at which players have the best opposite field power… and so on.  There is plenty of diversity in sabermetics, and it’s not even that hard to find.

My personal opinion is that the many of the newest metrics (at least in regards to hitting and pitching) are just more complicated ways to say the same basic truths.

I think this is the same reason I’ve soured on the more and more advanced metrics.  While simpler stats may have flaws, in a general sense, they rarely contradict the more advanced stats. Rather than contradict, they may just disagree on the level of impact (e.g. both agree Pujols is the best player, but may disagree on how far ahead of the pack he is (or isn’t)).

I would look at it as being analogous to the differences between VHS, DVD and Blu Ray.  The jump from AVG/RBI/etc. to OBP/OPS/etc. is similar to that of the jump from VHS to DVD.  The quality change is huge and worth the added complexity.  The jump from OBP/OPS/etc. to wOBA/WAR/etc. is similar to that of the jump from DVD to Blu ray.  Yes the quality is greater, but the additional complexity in formulae from OBP/OPS/etc. to wOBA/WAR/etc. may not be worth it for most folks, just like the added cost of Blu Ray may not be worth the small upgrade.

/ramble

i propose every sabermetric writer read “the elements of style.” then read it again. then again. and when all else fails, use the bill phone. that said, it’s hard to simplify knowledge. there are only so many carl sagans/bill james to go around.

Yes the quality is greater, but the additional complexity in formulae from OBP/OPS/etc. to wOBA/WAR/etc. may not be worth it for most folks, just like the added cost of Blu Ray may not be worth the small upgrade.

AaronGNP - You are certainly correct that for most folks the added complexity is not worth a small upgrade.  But for a player or agent or team negotiating a player’s next multi-year contract a tenth of a run difference in projected performance in each of the next 4 years would be valued at over a million dollars total.  With that kind of money at stake there will always be enough incentive to try and make minute improvements in how we value past performance and project future performance.  Over time the metrics that work best for their intended audience will survive.  BrianK said it very well in post #3, what works for some doesn’t work for others, and there is no reason to try and make everyone agree on a one “best” formula or method.

#11/Aaron:

VHS is ~333x480 in video size, DVD tops out at 720x480, and Blu-Ray at 1920x1080. So, VHS->DVD is a 2.16x increase in quality, while Blu-Ray is 6x. Blu-Ray supports double the frame rate too, so that’s really 12x. Plus, they’re both disc-based systems and Blu-Ray players are backwards compatible. Thus, there is little if any increase in complexity for the end-user.

Unroll OPS and it isn’t that simple. wOBA at least is up-front about its complexity.

It’s hard for me to get enthused about a new metric when the difference in output between the new metric and the one it proposes to replace is dwarfed by the year-to-year noise/variation in player performance that exists in MLB. 

When this is the case, there is no way the difference between the new and old metric can ever be measured/resolved in real life (as opposed to via simulation), and thus my interest wanes.

I’m (obviously) much more interested in the proliferation of baseball analysis that goes beyond trying to wring the truth out of the same old box score counting stats, and generates new data and new information, preferably continuous-type data (mph, direction in degrees, hang time in seconds, etc.) rather than discrete-type data (LD/FB, zone 78, FF/CU/CH/...)

Insert your own percentage, but 10 years offensive stats got us 90% of the way. 5 Years ago we were at 95%.  Now we are at 97%.  Personally, I disregard a lot of the new offensive minded stats because that last % or two we could realistically attain is irrelevant to my enjoyment of baseball.

Nick: none taken.  You have no idea how amused I am that both Dan and Pat linked to one of my articles in their rebuttals to John on THT (with near opposite reasons to boot!)

A lot of time I write about esoteric stuff that doesn’t really have any real-world application.  I happen to be interested in that sort of thing; others are not, and that’s fine.

What bothers me about the tone of Larry’s post, though, is the notion that there is some obligation to present one’s ideas for the lowest common denominator.  It’s true that if no one can understand your articles, you must not be trying or are just a flat-out lousy writer, because this stuff ain’t rocket science.  But I have no interest in writing for the average baseball fan--I’m writing for people who are interested in the same topics I am.  If there are only 20 of us, that’s fine with me.

That’s a rambling post by John, and I don’t really know what his point is. Speaking for myself, I find the fielding and pitchf/x stuff very interesting and I do enjoy reading different mathematical takes on some baseball basics--SIERA is an example--because I think new underlying math can tell us things.

And I also think that the people who focus on this stuff are doing a pretty good job of intergrating pitchf/x and fielding findings into their writing and thinking.

On the other hand, I think there is such a thing as sabermetric fatigue.  I certainly get it sometimes, when I feel like I’m getting less and less out of new analysis and stuff I read.  Maybe that’s simply what John is expressing.

That’s why it’s important to constantly bring in new blood to places like THT, Fangraphs and BPro--it’s also why I admire people like Tango and Dave Cameron so much.  I don’t know how you guys have kept up the intensity and enthusiasm for so long.  But thanks for doing so.

#13/Peter:

I certainly agree that it’s in the clubs’ best interest to stay on top of these metrics and to ascertain whether or not any particular metric has value to them, but for the casual fan, and heck even some of us that consider ourselves “stat-heads”, the metrics may not have nearly as much value, considering the complexity.

#14/sean

I didn’t mean to imply that Blu ray or DVD were more complex than the preceding standard, merely that the value added by the new standard was returning less payoff considering the cost (cost being equivalent to complexity in the metrics example).

You ever have one of those math/science/engineering professors whose lectures consisted nearly entirely of proofs?

Those classes were useless to me.  Yes, I understand that it’s important that *somebody* proved those theorems, and that *somebody* peer-reviewed them to make sure they are valid.

But going through that process myself was nowhere near the best way for me to understand the conclusions.  I had no interest in *proving* the theorems, I just wanted to *use* them.

I’m very very grateful that y’all figure out wOBA and xFIP and UZR and all that stuff.  But if you try to explain how you arrived at those conclusions, I’ll just cut class.  Let me go to the Fangraphs lab and mix the chemicals together myself, though, and I’m there.

As BrianK/3 said, different strokes for different folks. I’ve been reading John since 2006 and there are elements of his writing & analysis that I love and elements that I hate. I think the same can be said for every writer/sabermetrican.

Personally, I disagree with John’s critique. My baseball passions have progressed over the years (from fan, to intense following of the minor leagues, to a student of sabermetrics) and I attribute that progression to my thirst for knowledge. I’m thankful that everyday I can read the discussions on this blog, Fangraphs, BPro, BtB, Baseball Analysts, THT, Baseball America, minor league ball (John’s site), with a sprinkle of insight from Patriot, Phil, and others. The wealth of the information on the internet is an invaluable resource, and I’m happy to absorb more and more everyday.

Its funny, I haven’t started blogging (yet) but am launching a site (3/1) with a few guys I met on John’s site. I’m sure we’ll provide more coal than the diamonds, but I’m excited to have an outlet to share my passion. For most of us this is a hobby and will always be. The beauty of a hobby is that each of us can choose how to emerse ourselves in it, there is no right or wrong.

The most interest element of John’s critique that isn’t mentioned is that baseball isn’t a hobby for him. It is a job. He writes his book (which I have purchased), has a radio show, and writes for rotowire. Whether he likes it or not, baseball analysis is John’s profession. But, the statistical element of things has passed him by (and he admits that, to a degree). Advancements in statistical player analysis are made all the time, and I think that bothers John because that isn’t part of his repertoire (and its growing day by day.)

Minor league analysis still is constrained by and overall lack of information (in comparison to the MLB). But one day soon it could catch up and provide more analysis for our collective sponges to absorb. And that day, there will be more “coal” infringing on John’s turf.

P.S. Sorry for that ramble.
P.S.S. I love minor league ball’s community and John

AaronGNP -

I have no idea what your complaint is.  Are you saying that you want us to start evaluating hitters by OPS because it’s an easier concept to grasp, even though it’s been proven that wOBA is better.  Or that we shouldn’t strive for more accurate metrics because the ones we have right now are good enough, and easy to understand?  Nobody is arguing that a slightly better Base Runs formula will be a breakthrough, simply that it will be better than the old one.  I personally don’t find that stuff interesting, but plenty of people do and you shouldn’t condemn them for it just because you don’t enjoy reading their stuff. 

Besides, there have been a ton of primers on WAR, UZR, FIP, etc. over the past couple of months.  Like a ridiculous amount, relatively speaking.  It’s not that hard to find people trying to convey these more esoteric stats to the uninitiated.

Patriot,

if that’s the tone you got from my post, i must not be a very good writer. if you or others are not interested in writing for the average baseball fan, that is fine. i find that to be sort of sad since i think the majority of baseball fans are/could be interested in the insights that sabermetrics offers. i just don’t want to see such people then taking a condescending attitude towards those same average baseball fans who they are unwilling to educate, which is something i see all too often.

perhaps this all stems from this “hobbyist” thing that gets thrown around. i find the distinction a bit odd, as a lot of these articles - and i’m guessing some of the stuff that set sickels off - is coming from outlets where it appears people are doing it for a living, or at least as part of a living or with the aim of making it a living. seems the idea there would be to reach a few more than 20 people and otherwise show an ability to communicate thoughts.

at any rate, maybe that’s where things are now. the advanced stuff is so “granular” that only the hobbyists do and should care about it so it’s perfectly appropriate for it not to be tailored for a more mainstream audience by those same hobbyists.

Sabermetrics has plenty to offer the average baseball fan.  The optimal treatment of intentional walks in the Base Runs A and B factors is not one of them.  In order to appeal to average fan, I would have to stop working on areas that I find interesting.  That’s not going to happen until somebody pays me to do otherwise (and that’s definitely not going to happen).

The topics that I am talking about are not ones that anyone condescends to the average fan about.  I’ve never seen one of these too-esoteric sabermetricians condescend to the average fan because they used Pythagorean with an exponent of 2 rather than Pythagenport.

I don’t think that everything I do, or everything [insert sabermetrician here] does is inaccessible or irrelevant to the average fan.  Some of it is, though, and has to be.

Analogizing sabermetrics with hard science is always dicey, because it makes it seem that one is taking sabermetrics way too seriously and overstating the complexity of the field.  Accepting that risk, though, do people demand that top physicists communicate their findings in a manner accessible to the layperson?  They’d never get anything done.

In any event, if the people who are attempting to write for a living are unable to communicate their ideas to a sufficient number of people, then the market will speak and they will be living in cardboard boxes.

patriot, again, i’m pretty sure john sickels was not talking about you or your esoteric-ilk. he doesn’t seem the type to be reading your site.

and, yes, i think the analogy to advanced physics is both dicey and premature. but i think many physicists would be able to tell someone of reasonable intelligence what it is they are doing, why it is important, how it advances knowledge, etc. that doesn’t seem particularly like a task that would take up so much time that they couldn’t get anything done.

one can probably just ignore all of what i’ve written because this guy said it better than i can.

http://www.beyondtheboxscore.com/2010/2/16/1313631/a-little-saber-rant-or-dans

Of course I don’t think he’s talking about me, specifically.  That would be remarkably vain and completely unrealistic.  But if he’s not talking generally about people doing esoteric or technical work, then who is he talking about? 

With respect to Dan T’s piece, I just can’t agree with:

I think there’s some staleness in the community right now.  There seem to be three major trends in sabermetrics at the moment: small advances in areas (projections, for example), trying to figure out how to make Pitch F/X data useful, and valuing a player’s contribution using WAR.

I think there’s as much interesting and innovative sabermetric work now than ever.  Much of it (specifically the Pitch F/x stuff) is way out of my area of knowledge.  I don’t recall the existence of a golden age in which the field was “fresh”.

Assuming there was such a moment in time, the new work of the day would have had to have been on things like run estimators, park factors, baseliness--areas which have been well-covered and which any improvements are more likely to be of the esoteric type rather than the type that fundamentally changes the way one thinks about the game.

#22/Nick

No condemning going on here.  It’s a free world, and if one wants to work with advanced metrics, so be it.  I was merely agreeing with John’s position that it was extra complexity to come to essentially the same conclusion.

Five years ago, sabermetric types worshipped Adam Dunn style players.  Were you around for the years and years of love that was handed to the Rob Deer types? Or the ridicule that guys that was heaped on teams who were willing to play guys like Tom Goodwin?

These “new metrics” have shown that almost all of that was wrong.  The current appreciation of defense by statistical analysis is not “the same conclusion” that has always been reached.  Rey Ordonez was the whipping boy of statheads for most of his career.  Now, we’re among those who find value in Adam Everett. 

I agree with the statement that sometimes, the cost of explaining a new metric to the masses is not worth the effort.  It’s one of the reasons why I never pushed tRA in favor of FIP, even though it is slightly better and was developed by a fellow Mariner blogger. 

But I just can’t agree with the idea that we’re coming to the same old conclusions we always have been.  The current understanding of player valuation is light years ahead of what it was five years ago.

I should be clear in saying that I feel defensive metrics have come a long way, and that is a spot where real value has come.  The clarity gained going from fielding percentage to UZR is gigantic, while the clarity gained going from OPS to wOBA isn’t nearly as great.

The thing I don’t like about many of the advanced metrics is the merging of offensive and defensive values into a “mega stat” (Win Shares, WAR, etc.) I’d rather look at the offensive stats and defensive stats separately and come to my own conclusions, especially since I feel advanced defensive statistics are really just in their infancy.  Plus, with these merged stats, you need to buy into the balance (of offense and defense) that is presented by the stat’s creator.  Again, I’d rather see values separately, and make my own choices on who’s better.

Aaron: seeing that Fangraphs shows the components in an easy to add form, as does Baseball Projection, then I don’t understand the objection.

You are not beholden to the stat’s creator.

A personal anecdote: I have had several statistical blogs over the years for various websites.  There has always been constant pressure for me to write what I would call “entertainment” pieces - that is, articles that are more “accessible” to the low-information fan.  You know, like Malcolm Gladwell: nicely written popular pieces that ring true but are fundamentally incorrect.

I suppose I could be good at doing that (particularly getting things wrong), but it’s not something I enjoy.  It’s not like I want to be a professional writer who churns out content for low-information fans - I want to produce technical content for people who like that kind of work. 

Maybe three people out there were excited about some of the subtle improvements I made to several stats this year.  So?  The value of those improvements greatly exceeds the value of me churning out accessible pieces. 

It’s not as though somebody is paying me any sum of consequence - so why shouldn’t I have complete autonomy to analyze things that interest me?

I fully appreciate this site (and others) and the ideas they put forth, regardless of the style.  I don’t spend much time thinking about offensive metrics, but I went through all of the calculation involved in UZR and I look forward to the benefits the various F/X systems will bring in the future.  Plus I love regression to the mean.

Greg Rybarczyk (#15):

I’m (obviously) much more interested in the proliferation of baseball analysis that goes beyond trying to wring the truth out of the same old box score counting stats, and generates new data and new information, preferably continuous-type data (mph, direction in degrees, hang time in seconds, etc.) rather than discrete-type data (LD/FB, zone 78, FF/CU/CH/...)

Absolutely, Greg. This is exactly the point I made twelve years ago--man-hours and technology is the way to truly crack the defensive walnut-- and there is the prospect of all that happening. When it does, and I think it is now a “when” rather than an “if,” it will be interesting to see how much of what has been developed in the latest frenzied period will turn out to be a bridge to that level of knowledge, and how much will prove to be detritus.

I think certain people find a staleness, not because the community has slowed down, but because the new knowledge being taken in for them specifically has slowed down.  I’m talking specifically about people like me, who have been introduced to sabermetrics within the past few years.  Although it may have taken 20 years to come to many of these conclusions, I learned them all in 6 months.  So for me, the intake of knowledge has slowed considerably. 

Personally, I find articles interesting when they present ideas that are new to me (probably a lot of the “granular analysis” John is talking about), use methods that are time consuming, and I couldn’t do myself, or are about the Twins.  I’ve personally grown tired of the “Player a is overvalued because of… Player B is likely to break out due to...” type articles.  This is all stuff I can look up myself.  Although some of them are tough reads, I enjoy the technical articles that increase my knowledge of the game, even if it’s just a small increase. 

I guess what I’m saying is that I don’t want Patriot or others like him to water down his analysis to reach a wider audience.  There are plenty of websites (fangraphs, btbs) that offer well-written sabermetric pieces that are easy to digest.  If you don’t like the more advanced pieces, then do what I do to some of the more basic ones: skim over them and read the conclusion.  If it isn’t interesting to you, read something else.