IMPORTANT BLOG ANNOUNCEMENT

PLEASE CHANGE YOUR BOOKMARKS AND FEEDS TO THE NEW URL, YANKEEANALYSTS.COM. TYU IS IN NO WAY AFFILIATED WITH THE NEW YORK YANKEES OR YANKEES UNIVERSE.

This morning, John Sickels posted an article in which he suggested that sabermetric analysis has become too granular to be interesting and fresh:

The newest stuff is becoming so granular that I’m having problems making sense of it. I’m a humanities guy, and the most advanced math is beyond my ability to completely comprehend. 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…..

But I’m finding that as I read the most advanced sabermetric stuff regarding major league players, my eyes glaze over and I start to get the grad school feeling again: why am I reading this? I’m not enjoying it. I want to watch a baseball game.

So am I just entering my dotage prematurely? Or is advanced sabermetric analysis becoming so specialized that no one but physics and math majors can understand it, leaving us humanities majors behind, let alone the average fan? If that is true, what can be done about it? I don’t mean stopping research; obviously it needs to go forward. But I mean, how do we find ways to disseminate the new knowledge and make it comprehensible for the non-math folks among us? How do we integrate and explain the new knowledge?

This article has garnered plenty of interest in the sabermetric community, with two writers at THT responding. First, Pat Andriola:

So when you say that they are “more complicated ways to say the same basic truths,” you are, to an extent, 100% correct. However, the questions that remain are: 1) how much an improvement are we gaining over the basic truths and 2) how valuable are those marginal improvements? Maybe you find these advances boring and trite, but many others (such as myself) don’t. I’m sure there are front offices and analysts that clamor over the newest posts at Fangraphs and The Hardball Times, just like I’m sure you find the latest breakdown of a hot prospect’s swing riveting. These are, ultimately, questions of what gives us the most utility (or satisfaction), and are completely subjective.

Pat is right on the money here, as I have spoken to a number of people within front offices, including one GM, who said that they follow Fangraphs and THT religiously, attempting to get an edge in data analysis and evaluation. These teams find these marginal improvements important, hoping that they provide even the slightest edge over their competition. If the clubs find this sort of analysis important, then it makes sense for an interested fan to be interested as well.

The second article, from Dan Novick, does a fantastic job addressing the idea that sabermetric analysis is boring and too technical:

Baseball writing on the internet is a meritocracy. Sabermetrics isn’t spreading because we say it is. It is spreading because there is an increasing number of fans that find it useful. There is no such thing as “required reading.” If you don’t find a particular aspect of sabermetrics useful anymore, there won’t be any negative repercussions should you choose not to read it.

I could not have said it better myself. If you are a Yankee fan and do not like sabermetrics, you can skip over that sort of article here or at RAB, or ignore those sites entirely. There are so many options and forums for discussion that a fan could likely stick to the most basic of sabermetric precepts and still find a place where he or she can have a reasonable discussion about the sport, and have a fairly decent understanding of value and related concepts. If you are a creator of content, you can ignore sabermetrics as well, and cater to a less stat-obsessed crowd. No one is being forced to use sabermetrics. If you do not like them, just ignore them. It really is that simple.

Sickels is not “anti-stat,” and I doubt that he would suggest people ignore sabermetrics entirely. He was simply raising a reasonable point. Do you agree with him?

19 Responses to “Discussion: Have Sabermetrics Gone Too Far?”

  1. Newer stats interest me, but as a fan there just comes a point of diminishing returns. I appreciate (and hope) that front offices use them if they’re valuable, and I admire bloggers who keep them relevant and keep tweaking them to make them better, but for my own part I just don’t have the will to keep them all straight. And I’m only 26–part of a newer generation of baseball fan that is supposed to be more up to date with current, intelligent statistical analysis.  (Quote)

    [Reply To This Comment]

  2. I’m starting to understand Sickels’ point to a degree, but I’m not sure if its the analysis itself I find harder to comprehend or the manner in which its presented. Most new age metrics are “more complicated ways” but are necessary, for those hardcore fans amongst us to fully gauge what we are watching.

    Perhaps the issue arises in how it looks when presented, and that the list of equations given in an article makes some feel (me included on occasion) that they are “back in grad school” as Sickels put it. Take most new age stats and they aren’t really that difficult to understand the reasoning behind. Maybe its best we don’t fully explain how to get there, but generally outline the principle behind what we information this is generating.

    Its almost like people who don’t watch the news in favor or sports of reality TV, because they don’t want to think as much with something you can enjoy by just watching. Also, remember, within your blogosphere, many people think as you do, but that doesn’t mean a major of others don’t feel differently. It’s almost as if you have to step back and see a bigger picture.

    I promised myself as I got older (I’m 28) that I wouldn’t get set in my ways and start to curl away from advancement among the game. However, you have to present material in the easiest manner that doesn’t make people feel afraid to try and tackle it. Tables and equations of this magnitude can be overwhelming, but that doesn’t mean the information it gathers isn’t relevant.  (Quote)

    [Reply To This Comment]

  3. As a Ph.D mathematician I often have the opposite reaction. I find the combination of relative amatuerism (mathematically) and cock-sureness offensive.  (Quote)

    [Reply To This Comment]

    Moshe Mandel Reply:

    Here’s the problem- a number of “real” mathematicians have tried their hand at sabermetrics and have failed miserably, simply because they do not understand the game enough. The fact that sabermatricians have taught themselves regression analysis and modeling does not really bother me in terms of trusting the work, because the internet community is essentially a giant form of peer review. If an idea does not hold mathematically, it gets discredited, quickly.  (Quote)

    [Reply To This Comment]

    AndrewYF Reply:

    And it simply saddens us ‘regulars’ without PhDs that some of the ‘smartest’ members of our society look down on people’s work who don’t have degrees, simply because those ‘amateurs’ don’t have the piece of paper that they do.

    You don’t need a PhD – or a degree of any kind – for your work to be valuable.  (Quote)

    [Reply To This Comment]

  4. The stat approach has obvious value, and as time goes on, it can only get better. I used to only use BA, RBI, homers to judge a hitter. I think it has only been in the last decade, since “Sabermetrics” that OBA became popular. Slugging %, OPS, obviously expressive. There will probably be more that evolve into common usage, ie, beyond the stat pro’s and amateurs.

    Two caveats: limitations to formulae have to be made clear; and please check the forest when you are counting trees. also, help us out with a description when you flop some UZR on us.  (Quote)

    [Reply To This Comment]

  5. “Here’s the problem- a number of “real” mathematicians have tried their hand at sabermetrics and have failed miserably, simply because they do not understand the game enough.”

    Very apt comment reminding me of what Bill James had to say about Earnshaw Cook, a mathematician/ scientist who worked on The Manhattan Project and later tried to apply his knowledge of those subjects to baseball in the 1960s in his book “Percentage Baseball.” About the largely discredited Cook, James wrote:” “(he) knew everything about statistics and nothing about baseball.”

    Obviously the confluence of statistics and baseball has advanced markedly since Cook’s primitive attempts in the 1960s and James’ more enlightened efforts in the 1970s and 1980s and just as clearly efforts should be to continued from that aspect to challenge traditional thinking to better understand and measure both individually and comparatively the skill sets of various players, of managerial strategies, etc. This is no different than the use of increasing sophisticated visual aids to improve the techniques themselves and the recent use of replay to evaluate umpire’s calls and change when necessary. All of these “complications” are designed to improve, in various ways, how the game is played and presented and should continue to be upgraded, assuming they pass critical review.

    With the explosion of statistical analysis, I don’t think it’s imperative that someone knows how exactly how the “‘new age” stats are mathematically derived. Many people have watched baseball over the years and gotten tremendous pleasure and could tell you that an ERA under 3.00 for a starting pitcher is real good while an ERA over 4.50 is pretty bad without understanding how those rudimentary calculations were made. As long as we can communicate what are good, average and below average numbers for each of the advanced metrics, then a similar understanding and vigorous debate will necessarily follow.

    As for the argument that there is a diminishing returns in the new statistics, this is true but no different than any endeavor in life. Initially there is period of excitement which accompanies a new challenge and the ensuing growth and improvement. Then,after the initial charge, the rate of growth and understanding diminishes. At that point, one has to decide whether one wants to keep going and continue to pursue incremental improvements or spend that time and effort someplace else. As Pat , Dan and Moshe have pointed out,that is an individual decision for each of us to make.

    The newer metrics by continuing to isolate and to refine statistical evaluation have the effect of cutting out the superfluous to give us clearer picture, much like a pitcher abandoning an ineffective pitch and going with a more focused assortment.or less invasive therapies cutting out side effects from various medical treatments( ( I realize the analogy may be a little heavy).. They should be applauded even if not always fully comprehended in all their intricacies..  (Quote)

    [Reply To This Comment]

    Michael Reply:

    Your comment shows a problem with the basic stats right there. You say that a 3 or lower ERA is really good, but that’s only by today’s standards. In the dead-ball era, league average ERA was in the neighborhood of 2.7. So basically, a pitcher with a career 2.7 ERA now would be a superstar, but in the early 1900s a pitcher with a career 2.7 ERA was just average. ERA doesn’t take things like this into account, but more advanced statistics, like ERA+, does. A pitcher with a 2.7 ERA in the dead-ball era would have an ERA+ of around 100, which a pitcher with the same ERA today would have an ERA+ of around 160, meaning he is 60% better than average. The basic stats like ERA can give you a quick glance at a player’s abilities, but if you want to seriously compare players, you need the advanced stats to do it.  (Quote)

    [Reply To This Comment]

    classicsteve Reply:

    Understood. I think that people who watch the games on a regular basis are sophisticated enough, in most instances, to understand that an ERA under 3.00 pre- WWI or in 1968 is less outstanding than the same ERA during the late 1990s or 1930 and can adjust their thinking as to what’s good , average or below as overall run scoring fluctuates up or down. Obviously, ERA+ by comparing a pitcher’s ERA to the average of his peers for a specific season enables us to look at Walter Johnson in comparison to Greg Maddux better than looking at just their raw ERAs.  (Quote)

    [Reply To This Comment]

  6. Two of the bigger problems are presentation of the new stuff (usually done so that only those who love numbers and formulas can read through it) and the usefulness of the new stuff – what is the ratio of complications/improvement? The newest example is SIERA – SIERA is almost certainly better than FIP and xFIP. So is BaseRuns-DIPS. But is improvement big enough to justify moving from metric whose formula can be easily explained to anyone to metric that will be black box for most of the general public, because of its complicated formula? For BaseRuns-DIPS it isn’t, for SIERA we’ll have to see.  (Quote)

    [Reply To This Comment]

  7. I think that there are two different concerns. One is bringing sabermetrics into the colloquial, and the other is finding the truth as best we can.

    Bringing real statistics into the colloquial really needs to happen. It needs to happen so that color commentators don’t sound like asses all the time, so that awards and Hall of Fame voters can make the right decision, and so that to keep me from being driven insane by writers talking about the value of how many runs a player scores. To use the language of another comment, we’re not close to the point of diminishing returns here. Simple, understandable statistics are essential here. wOBP is not going to hit the mainstream.

    A friend and I are working on some new statistics right now (I don’t want to give it away) for a Hall of Fame project we are attempting. We are deliberately making them as simple as possible so as to make them understandable. We’re sacrificing some “truth” at the expense of colloquialism. As long as we understand the limits of that statistic, I have no problem with doing that.  (Quote)

    [Reply To This Comment]

  8. Except for a very small minority if us, baseball, at it’s core, is entertainment. If you enjoy deconstructing the game in 100 different ways, go for it. If, like me, you prefer to stick to BA and ERA and just watch the game with a beer and a buddy, that’s there too. It’s all just part of what make baseball the greatest game on Earth and being a Yankees fan the best place to be.  (Quote)

    [Reply To This Comment]

  9. I disagree that wOBA can’t hit the mainstream. If QB Rating can be popular among mainstream NFL fans and analysts, things like FIP, wOBA, EQA, etc. can definitely be popular among mainstream MLB fans and analysts. The key, IMO, is making the stat “recognizable,” as both wOBA and EQA are; they may be calculated differently, but when they’re “spat out”, they look like OBP and BA, things that the average fan will recognize.  (Quote)

    [Reply To This Comment]

  10. Perhaps I was terse to a fault for some reason this typing interface doesn’t work well on my iPhone. What I miss are the papers. Can anyone point me to original WAR paper containing a study of it’s effectiveness? I feel as though sabermativians are incredibly confident in there estimators (perhaps as a reaction against themainstream reluctance of the old guard?) to a point of failure. Certainly it is hard read so much unrefereced work.

    And please try to check your baggage at the door: I mentioned my background to establish where I was comming from, I didn’t realize I’d expose such vigourous hatred based on one anecdotal story.

    Thanks  (Quote)

    [Reply To This Comment]

    Moshe Mandel Reply:

    I don’t think anything here qualified as vigorous hatred, but I am sorry if you felt that way.

    I do not have the time to search for a paper on WAR, but I am pretty sure I have seen a number of them on the accuracy of WAR and its correlation to actual wins. You could try and google it, and if you cannot find it, I will be glad to search for it later. An example can be found in SIERA, which was just released in a 5 part series that included an extensive study of its effectiveness.  (Quote)

    [Reply To This Comment]

    Michael Reply:

    To be honest, your comment was rather condescending. But that aside, I don’t think anyone is claiming that WAR is perfect. It has some very obvious flaws.

    The offensive side of WAR is pretty good. It uses Runs Above Replacement, which is based on wOBA. In case you’re unfamiliar with it, wOBA is a pretty precise stat that uses linear weights for each possible outcome of an at-bat that have been determined through years of data analysis. Basically, it does a similar thing as SLG, but the weights are more accurate, i.e. a HR is not really 4 times better than a single.

    The main problem I have with WAR is the defensive side. Evaluating defense is an inexact science at best. Things like errors, assists, and putouts are easy to keep track of, but things like range are much harder to account for. Maybe player A has a lower fielding percentage than player B, so you might think he’s a worse fielder, but maybe he gets to a lot of balls that player B can’t get to because he is faster, reacts quicker, or takes a better route to the ball. These kinds of subjective qualities are very hard to quantify. The stat that WAR uses to do this is UZR, a defensive metric that looks at how many balls hit in a player’s defensive area does he get to, error rate, for infielders how well he turns double plays, and for outfielders how well he can either gun down or prevent runners from advancing bases.

    UZR is probably the best defensive stat available now, but it is very flawed. First, it doesn’t take into account how hard a ball was hit into your area. If Albert Pujols smokes a line drive by an infielder, that will decrease his range factor, but if Julio Lugo hits a weak grounder to the same spot, it will increase the range factor of the fielder if he grabs it and throws Lugo out. One of those plays is very hard to make, while the other is easy, but UZR treats them equally. The same is true for outfielders, the balls hit in your area does not account for if it was a flyball or a line-drive, or how far you had to run to get to it. Because of things like this, you can see wild fluctuations in UZR’s for players from year to year, especially for outfielders. This is mostly due to the sample size of defensive chances for outfielders being much lower. Finally, UZR has nothing for catchers. Quantifying a catcher’s defense is even harder than the other fielders, and no one has come up with a really good all-around catcher defense stat yet.

    UZR, however, can give you a rough estimate of a player’s defensive ability. If Player A has been averaging a UZR/150 of +20 over the past few years, and Player B has been averaging -20, you can feel safe to say that Player A is much better defensively. When two players have very similar UZRs, that’s when I’m not comfortable making a judgment using it.

    So that’s my main gripe with WAR, the offense side is good, but the defense side is shaky, and catcher’s don’t even get their defense counted. It does account for the position a player is at, which is important. Basically, if you take two equivalent players, one of whom plays SS and the other plays 1B, the SS will be more valuable to the team and have a higher WAR because he plays a more important position.  (Quote)

    [Reply To This Comment]

    Moshe Mandel Reply:

    One correction- UZR accounts for ball speed and type. A slow grounder is scored differently than a hard liner.  (Quote)

    [Reply To This Comment]

  11. It’s not that those of us who like advanced statistics don’t sit back and enjoy the game too. Believe me, when the game is on, I could care less about any stats, the only stat I care about is my team winning the game. At other times, though, sabermetrics lets me inform myself on the game so I better understand it. If my team is looking to sign a free agent and I want to find out which of those available is really the best player, sabermetric stats like wRC+, UZR, and WAR are much better to look at than BA or OPS. Obviously, I’m not the one making the decisions, but I talk a lot of baseball with other people, and I like to feel like I know what I’m talking about when I suggest one player over another. Saying something like “Ichiro is a better hitter than A-Rod because he has a higher BA” has less validity than “A-Rod is a better hitter than Ichiro because he has a higher wRC+.” (I’ve actually had people try to argue that Ichiro is the best hitter in the league because he hits for high average…) The more precise the stats you are using, the more validity there is to what you say, and I, like most men, hate to be wrong, so I use the best stats I can find.  (Quote)

    [Reply To This Comment]

Leave a Reply

(required)

(required)

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

© 2011 TYU Suffusion theme by Sayontan Sinha