Kapler re-signs in Tampa and further proof Fangraphs is ridiculous

[sbrylski]

One should look at the methodology, because if a metric was judged on how well it fit your current perceptions we would never learn anything new.

 

Nobody is accepting it as true fact and blindly following FanGraphs because, like Russ emphasized, we all know its an estimate. But its an estimate we can learn something from. How accurate that estimate is depends on your critique of the methodology and not how closely it matches your preconceived notion, else you're simply running in place.

 

Blindly dismissing WAR is close-minded. Accepting it for what it is does not equate to blindly following - the critical thinking filter is still in place.

[Hate those Rodents]

If somebody tells you that they outright HATE a certain ethnicity, do you let them go into all of the reasons why they hate that ethnicity, or can you simply make note of the end result (the hate of the ethnicity or for instance Kapler being worth over 5 million last year) and realize it's not worth your time?

[Invader3K]

The problem here is that some DO use those stats on this site like they're preaching from the Holy Bible. Do I need to bring up the much maligned and often repeated "JJ Hardy was worth $21 million last year," quote?

 

Like I said earlier, I wouldn't have a problem if they didn't stick that dollar sign in front the figure. Putting the dollar sign there insinuates that it has a correlation to real world economics and baseball salaries. They clearly do not.

[Thurston Fluff]

I feel bad for people who are so closed minded that they will just call Fangraphs ridiculous without trying to understand some of the information provided there.

 

as opposed to the open minded people who ignore the fact that it doesn't match up with what really happens. Call me closed minded but if reality and mathematical formulas don't match up I tend to think it's the calculations that have it wrong not reality. Using fangraphs to compare one player to another is different than saying fangrpahs has correctly calculated a players value in real life.

The other thing some of us question is how good can it be in comparing players if it can't accurately calculate players worth compared to reality? The comparison within the system is only as good as the accuracy of the system as a whole. I tend to think it's on the system to show it's value not the other way around. Since the system as a whole does not match up with real life then why should we believe it is good at comparing players worth in real life? I think it has some worth in comparative situations, though not as definitive as some open minded people like to think, but relatively worthless in calculating actual value.

 

 

Calling something stupid when you don't even know what it exactly is seems more like a defensive mechanism to me.

 

I think when you see teh results of it's calculations not match up with real life it can be seen as less than perfect without knowing the exact mechanisms being used. If you have a car that gets 9 mpg and you see it's rated as getting 25 mpg it doesn't take knowledge of the components of how they arrived at that number to know it's wrong.

Sometimes real life results trump paper calculations. When Gabe Kapler cannot get what the calculation says he is worth then it appears to me real life situations should be given at least as much credence as a mathematical formula that tells us what real life should situations should be.

Of course you are correct in saying if you want to see why it doesn't match up then you have to look at the formula used. I think you are also right to think the use of UZR in this situation might be the culprit.

But that gets right back to a fundamental shortcoming of one size fits all formulas. It doesn't take into consideration any intangible variable between sugjects. For example it uses age regressions but doesn't account for the difference between one player who takes care of his body and one who doesn't. I think players like John Kruk see age regression issues earlier and faster than workout warriors who eat right and don't drink or smoke. Lumping a 38 year old Mike Cameron in the same category as a 38 year old Andrew Jones is probably going to get you the wrong real world results.

 

In the end fangrpahs is a useful tool for use in gathering some amount of objective value but should not be seen as giving anything close to definitive truth. In either the real world or even within it's own mathematical universe. It is much better if used in conjunction with other information.

[thebruce44]

as opposed to the open minded people who ignore the fact that it doesn't match up with what really happens. Call me closed minded but if reality and mathematical formulas don't match up I tend to think it's the calculations that have it wrong not reality. Using fangraphs to compare one player to another is different than saying fangrpahs has correctly calculated a players value in real life.

 

I don't think anyone made such a definitive claim as "fangraphs correctly calculated a players value in real life." The only definitive claims I have seen in this thread are that Fangraphs is ridiculous, which is a close minded stance.

 

If you don't want to use Fangraphs or WAR at all, that is your decision and I won't try to change your mind. I do think you are neglecting a very legitimate source of information for the wrong reasons though since the player value is the least important and useful statistic on the site.

[sheethead]

If somebody tells you that they outright HATE a certain ethnicity, do you let them go into all of the reasons why they hate that ethnicity, or can you simply make note of the end result (the hate of the ethnicity or for instance Kapler being worth over 5 million last year) and realize it's not worth your time?

 

But there is value in understanding how Kapler was valued at $5 million last year, from what I am reading. The value in understanding is the knowing how not to value players in the real world, but how they are valued in a vacuum of all else being equals (every team has Yankee money to play with).

 

If you don't want to use Fangraphs or WAR at all, that is your decision and I won't try to change your mind.

 

Two years from now there will be a new hot stud stat to cite and use for or against certain players. Maybe I will get on board with that one. WAR doesn't impress me much, but I honestly see how it is liked. Its a fantasy stat, and fantasy stats are fun.

 

I posed the question earlier (using a point from another poster), who thinks Nyjer Morgan is more valuable than Ryan Braun? Any takers? I can see why that question is being ignored, if being done so on purpose.

[Thurston Fluff]

I don't think anyone made such a definitive claim as "fangraphs correctly calculated a players value in real life."

 

They certainly have on this site. The JJ Hardy was worth 21 million Mike Cameron was worth more than 10 million were both claims based on fangraphs. There are many examples of that type of use of the site.

 

The only definitive claims I have seen in this thread are that Fangraphs is ridiculous, which is a close minded stance.

 

I think you are overstating what we are saying. You can use closed minded if it makes you feel better about your position but I tend to think skeptical is a better word.

being skeptical about a system that doesn't match real world results with it's system of calculating value can and should be viewed with a healthier dose of skepticism than you seem to believe. I don't want to accuse you of being closed minded yourself but if you want to call the kettle black you should first check to make sure you are not a pot. Saying those of us who disagree are closed minded does nothing to show your position is correct. It just appears to be a cheap way to discount anyone who opposes the way you think. I've read enough of your stuff through the years to know it is not a common tactic of yours so let's just chalk it up as poorly chosen words. Lord knows I've been guilty of that enough myself.

[thebruce44]

Fair enough, I am not trying to discount people who oppose me. I don't think this argument is between you and me Backupcatchers, I think we are arguing about what other people have said and I think it sounds like both of our opinions are somewhere in the middle.

 

The problem I have is with people who think Fangraphs or WAR are crap because the value rankings aren't accurate. If you are skeptical because you understand WAR and what goes into it, thats fine with me. To discount a resource like Fangraphs because of the value statistic is "close minded" in my opinion and you are putting yourself at a competitive disadvantage in your search for baseball knowledge (sorry that came out so corny).

 

That's why I wish Fangraphs would just get rid of that column since people don't seem to understand it, especially if someone made those statements about Hardy and Cam (although I think the Cam one is a legit argument but thats for another thread).

 

Sheethead: You have found a legitimate place where WAR has overvalued a player because it accounts too much for defensive statistics. WAR isn't perfect, but there are plenty of other instances where statistics have showed one player is more valuable than another when that wasn't the case.

 

For example, Looper had some half way decent statistics this year, but WAR shows that he was basically the worst starting pitcher in baseball. I think WAR and FIP is more accurate there than ERA or other measures.

 

Taking all of these statistics together along with a reality check provides a pretty good picture. Discounting one statistic from your repertoire because of a few examples or because the "player value" statistic is ridiculous means you have one less source of information to go from and I can't get behind that stance.

[Thurston Fluff]

I agree. One should not discount any stat or measurement system, fangraphs included. On the flip side I don't think one should use any single system as a definitive indicator of true value, fangraphs included.

[joepepsi]

From my perspective the problem with WAR, VORP, Win Shares and other complicated intricate measurement systems is they don't really tell us much more than simple stats like OPS, *OPS+, K:BB ratio, and other handy, easy to get to measurements. I realize the more complicated systems are attempting to measure defense, but it seems they always grossly overvalue it.

[TooLiveBrew]

How much of the problems stem from just using small samples (aka one season) of UZR data? Iirc 3 or 4 seasons of UZR data is roughly as reliable as one season of hitting data.

[rluzinski]

OK, just going to ramble for awhile. No spell check, no nothing.

 

as opposed to the open minded people who ignore the fact that it doesn't match up with what really happens.

 

You mean, teams don't pay players after the year is over, for how much value their performance was worth? I think we already know that. http://forum.brewerfan.net/images/smilies/smile.gif And if they did, you would see a significantly higher variance in what players currently get. That's because we expect performance to have a higher variance than projected performance. Players are paid for projected performance.

 

Sheethead,

 

You are still completely confused. You are claiming that the methodology is propietary, despite it being pointed out over and over again that it isn't. You cite that the damning evidence against it is that end result doesn't match up with YOUR expectations. That is the WORST reason to ever reject something. You keep referring it as WAR, which which is simply baseline, not the metric itself. You claim it assumes that every team has Yankee money, which is completely false. You say this:

 

"I posed the question earlier (using a point from another poster), who thinks Nyjer Morgan is more valuable than Ryan Braun? Any takers? I can see why that question is being ignored, if being done so on purpose. "

 

People continue to point out the distinction between "is" and "was". Fangraphs claims that Morgan's and Braun's performances were essentually equal. Let's go through quickly why. First, offense:

 

Morgan: 5.2 RAR (run above replacement)

Braun: 46.0 RAR

 

I think everyone would agree that the difference was huge and this agrees. Morgan isn't in the same universe as Braun.

 

Now defense. We have to remember two things. First, UZR is trying to estimate the value of performance, not true talent. Now obviously, if a player gets a 10 RAR every year for 10 years, we are reasonably confident that his performance was equal to his talent. Keep that in mind when you see this:

 

Morgan: 27.4 RAR

Braun: -15.5 RAR

 

UZR thinks Morgan is an elite defender, not only for 2009 but also in his short stints in 2007 and 2008. But it's only about a year's worth of data, which is roughly equal to a half a year's worth of hitting data. Maybe he's really a 12 RAR defender who happened to luck himself into a true 27.4 RAR of value in 2009. Maybe he's the next Andrew Jones and really is worth that much defensively. I really don't know and neither does UZR. Just an ESTIMATE based on the best data available.

 

Is Braun really a -15.5 RAR defender? I don't think he but we only have 2 years of data to look at. We was a terrible 3B but looked decent last year and his UZR thought he was just a smidge below average.

 

Maybe he played below his true abilities this season? Maybe? I watched almost every game and I can't really offer up a strong opinion. I remember some bad plays I guess. But short of reviewing every play he made in 2009, I guess I would trust UZR more than my memory. Of course, I don't really trust that my memory is clear and unbiased anyway, so that's not saying much. It will be interesting to see what the fans thought when tangotiger.net publishes the fan scouting report (maybe he did already?)

 

Finally, we have positional adjustment:

 

Morgan: -2

Braun: -7.3

 

I hope their isn't too much argument over why a positional adjustment is necessary.

 

Now, how fangraphs converts RAR to $ is another topic altogether. I won't get into that but it's based on work by Tom Tango and you can find thousands of words about it on the web. It is based on real life data, though. So, it wasn't just "made up", as some people claim. insidethebook.com had a LONGGG thread that looked at every off season signing and checked the calculated value vs. what they actually got. It wasn't perfect but it was pretty darn good. Of course, they were using PROJECTIONS as the foundation, not simply the unregressed numbers from 1 year.

 

I hope that this stat is forgotten someday because that will mean that it was replaced by a stat that better estimated the value of performance. Right now, for better or worse, this is the best publicly available one. I'm sure many teams have their own proprietary way to estimate what players are worth.

 

I hope this helps!

[sheethead]

Looper had some half way decent statistics this year

 

Other than wins (slightly useless "stat"), what was decent about Looper? I don't need Fangraphs to tell me that he was a bottom scraping SP in the league in 2009, most of his stats will tell you that.

 

That's why I wish Fangraphs would just get rid of that column since people don't seem to understand it...

 

I agree, and I think that might be the point of this thread.

 

Discounting one statistic from your repertoire because of a few examples or because the "player value" statistic is ridiculous means you have one less source of information to go from and I can't get behind that stance.

 

WAR is not an official statistic in my opinion, its a parameter. If it was a statistic, it could be observable and be a function of a sample where the function itself is independent of the sample's distribution. Since only the proprietors know exactly how WAR is calculated, how can anyone consider it a statistic. The methodology is available, but no one has yet to show me HOW ITS ACTUALLY CALCULATED.

 

A statistic is an observable random variable, which differentiates it from a parameter, a generally unobservable quantity describing a property of a statistical population.

 

I don't look at just one or two metrics to compare players, but I generally like to use statistics that have clearly defined calculations. All the other metrics appear to be fantasy type stats that are beloved only by hardcore baseball geeks. Thus, I understand why actual calculations of such metrics are held proprietary, because they have flaws (as admitted by proponents here). Stats don't have flaws. Limitations? Of course.

 

My point? I have never needed a parameter like WAR to tell me who a better baseball player is, and I doubt many GM's need it either.

[TooLiveBrew]

It will be interesting to see what the fans thought when tangotiger.net publishes the fan scouting report (maybe he did already?)

 

Not yet. Fwiw, the '08 version was posted on Nov. 25th

[thebruce44]

From my perspective the problem with WAR, VORP, Win Shares and other complicated intricate measurement systems is they don't really tell us much more than simple stats like OPS, *OPS+, K:BB ratio, and other handy, easy to get to measurements. I realize the more complicated systems are attempting to measure defense, but it seems they always grossly overvalue it.

 

I am not going to say what you do or don't know, but your post makes me think you don't know enough about these metrics to make a fair assessment of when and how they could best be used (I don't mean that to sound condecending so I apologize if it comes off that way). I don't have a great handle on them either so I am not the best person to explain it. I will just point out a few instances.

 

OPS is just raw offense. You wouldn't want to use it to compare JJ Hardy to Prince Fielder. It has nothing to do with how well they play their respective defensive position; it has to do with the fact that a SS isn't expected to hit as well as a 1B. JJ should be compared to other SS's and Prince to other 1B for the offense they provide.

 

OPS+ suffers from the same deficiency, except it at least adjusts for the year. If you wanted to compare just Yount's offense in '82 to Prince's offense in '08, you could use OPS+. However, in doing so would you say that Prince's '08 was as good as Yount's '82 when looking at everything? No, that would be foolish since Yount provided nearly the same offense (at least adjusted for the time period) from SS. VORP would be a better way to measure that since it takes position and year into account.

 

Now I feel like I am rambling too, but hopefully there is a point in there somewhere.

[TooLiveBrew]

The methodology is available, but no one has yet to show me HOW ITS ACTUALLY CALCULATED.

 

This might be the thread Russ was referring to... Link. You'll have to read the comments, too. It took me doing a google search for "WAR calculation baseball" to find this.

[sheethead]

I guess I need to read The Book (extensively) to find out why the baselines are what they are, and how the "boosts" allocated to certain positions, AL vs. NL, etc. are determined. I understand SS is harder to play than 1B (methodology). What I don't understand is why its +1.5 harder. It just seems arbitrary to me right now. Now, I am sure I can read the explanation as to why that should be that number, and it probably makes sense. But I will need to put faith in that person's perceptions over another person's perception. Maybe I will dig in this winter to find the root of what bothers me with Tango's numbers.

[logan82]

If you are looking into something to try and find something wrong, I think you are going about things the wrong way.

[Hate those Rodents]

If you know the end result is wrong, your only motivation for even "looking into it" would be to find out why/where it is wrong. Anyone without the time on their hands to read up on such a thing would just simply say it's a bad formula and not worry about it. You won't hear about WAR or Fangraphs from players, managers, analysts, even rotoworld. You'll hear about it on the site and mlbtraderumors.

 

I will say that if they take the player dollar value off of their formula, it might make more sense. Might. But until that happens, it's hogwash that is frustrating to read. I get sick and tired of every single SS and CF being rated so highly by the site and then posters who I respect here coming and telling me that so-named brewer is worth that much and we had better pay him that much. But that's just one man's opinion, and one who admittedly does not like obscure data.

[rluzinski]

There is so much about baseball that I was CERTAIN about that was later proven wrong. In 2003, I was spouting off on this very board that I didn't think the difference between good and bad defenders was large enough to make good defense very valuable. I had no evidence of course; I just KNEW it was right. How pathetically wrong I was. I'm sure there are things about baseball right now I take for granted as true that isn't. If you are never proving yourself wrong it's only because you are never trying to.

[TheCrew07]

Why is the "you don't know enough about the metric" card always played in these discussions? If that's the base assumption then it simply isn't possible to discuss the matter intelligently.

 

Most of us understand perfectly well how WAR is calculated and how they arrive at the average $ figure per win and simply don't agree with the methodology involved. Just like UZR, no matter how they tweak the metric it will never be fool proof because it's based on a zone concept where fielders share zones and there are simply too many "what ifs" to definitely measure defense. It's not that metrics don't have value, they all have value in their own way, it's just that too often they are thrown around as facts or absolutes when they are nothing more than complicated opinions. They are a piece of the puzzle, not the puzzle itself.

[thebruce44]

Most of us understand perfectly well how WAR is calculated and how they arrive at the average $ figure per win and simply don't agree with the methodology involved.

 

It seems like half this thread was people asking how it is calculated, so thats why the "you don't know enough" card was played.

[LouisEly]

UZR is just a "best guess" at defensive value.

'zactly - it is a "guess".

 

I'm not saying they factor defense in perfectly, but because a metric doesn't match up with your perception is not reason enough to reject its findings. That's simply being close-minded.

 

I feel bad for people who are so closed minded that they will just call Fangraphs ridiculous without trying to understand some of the information provided there.

 

First, UZR is trying to estimate the value of performance, not true talent.

 

And that is my issue - it's not that I don't understand UZR, it's that I think that lumping centerfielders and left fielders together for UZR ratings is a bunch of carp. Allow me to re-address my close-mindedness with specifics that I posted on October 9th.

 

"Most hitters are right-handed, and when a hitter pulls the ball it gets hit hard and gets to it's destination very fast. (There's a reason why 3B is called "the hot corner"; balls pulled towards third base by RH hitters get there very, very fast - this I know from experience.) There are fewer left-hand hitters, but when they pull the ball it is to RF. The harder the ball is hit the less time it takes to get to where it lands, and the less chance that the fielder will catch it. Thus why I think UZR is biased against left-fielders and biased towards CFs; I will agree that LFs are generally the worst fielders overall but due to righties pulling the ball they have the least amount of time to get to it to catch it. Where I'm going with this is if you watched a lot of baseball games and calculated the average time that a ball was in the air when it was hit to LF or RF versus CF, I would be willing to bet that it would be a significant difference. You see very few screamers up the middle to CF; centerfielders have a lot more time to react to the ball and get under the ball and catch it because it is in the air longer. Thus I don't think that a good CF saves as many runs as UZR and Fangraphs thinks they do."

 

My understanding, and correct me if I'm wrong, is that UZR basically calculates what percentage of balls hit to a fielder's zone are caught by that fielder relative to others at that position. With the specifics posted above, I don't think it's fair to compare LF to CF because of differences in the average length of time that a ball hit to LF is in the air versus CF. CFs have more time to catch it because it is in the air longer. I agree that generally LFs are the worst fielders, but I don't think it makes that much of a difference. You could put those CFs in LF and I don't think they would catch a lot more balls than most LFs do, with the exception of Carlos Lee and Manny Ramirez, and I don't think that the few more balls that they would catch would make that much of a difference in runs. Rickey Henderson and Tim Raines played LF, and I would be curious as to what their UZRs would be. Bottom line, I don't think that the % of balls caught by a CF is comparable to the % of balls caught by a LF, and thus their calculations of lumping in all OF positions is a bunch of carp.

 

So I hope you see that I am not being "close-minded" - I am flat out disagreeing with their methodology with specifics.

[rluzinski]

Most of us understand perfectly well how WAR is calculated and how they arrive at the average $ figure per win and simply don't agree with the methodology involved.

 

You are one of the posters who's been willing to do his homework and actually learn about this stuff. I don't agree with all your criticisms (some certainly point out significant weaknesses) but at least you've armed yourself with much of the knowledge needed to carry on an intelligent discussion about the topic. Despite your claim, most of the detractors posting here have no idea how it's calculated. They don't agree with the conclusion, so that's the end of that. Hell, at least one poster still seems convinced that the methodology is a secret.

[thebruce44]

I feel bad for people who are so closed minded that they will just call Fangraphs ridiculous without trying to understand some of the information provided there.

 

So I hope you see that I am not being "close-minded" - I am flat out disagreeing with their methodology with specifics.

 

Agreed. That comment was not addressed at you at all.