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

[joepepsi]

An outfielder doesn't have to boot balls to be a bad fielder. Just not getting to as many balls would be enough....I don't think it would be much of a stretch to say that Braun lets an extra hit fall in front of him or get over his head once a week. An extra ball falling in here and there adds up to quite a big deal over an entire season.

 

I guess my point is that if a seemingly small event(one hit every week) on offense can lead to such a drastic difference(85 points of OPS) over the course of a season, a seemingly small difference in quality of defense would probably show up as a pretty large difference in defense as well.

Except the overwhelming number of plays in the field are routine, especially for outfielders. I agree a LF'er who missed one catch a week that an average (replacement level apparently), fielder would catch seems extraordinarily bad. And highly unusual. There just aren't that many opportunities to fail so miserably.

 

Braun had 314 chances in 1364 innings in LF. That's about 2 chances per game. Teams play 6 games a week, so that means Braun had 12.43 chances a week on balls he got to.

 

Ibanez is generally considered a poor LF'er. He had 224 chances in 1123 innings. That figures to about 10.77 chances per 6 games. Ethier is considered a pretty good LF. he averaged 11.55 chances per 6. Chase Hedley averaged 12.25. Soriano 11.75. Carlos Lee averaged 9.4.

 

Where's the evidence that points to Braun missing balls that other LF'ers in the league are reaching? Even if other LF'ers are getting more chances per game or per week, how do we know that's not due their teams' pitching staffs?

[sheethead]

According to you, that invalidates all the methodologies used. That's a weak argument.

 

If someone is arguing that I am trying to invalidate the methodology, they are missing the point. My argument from the beginning of this thread was the dollar value Fangraphs is placing on WAR. I am arguing the overall value of this methodology, and my feeling of its uselessness.

 

I go back to the point of this thread... $5.2 million for Gabe Kapler's .239/.329/.439 in 238 plate appearances??? Not even close, and I don't care if Kapler is 23 or 54. Age is irrelevant when given this amount of production. Lots of data with endless explanations and justification equals a headache and all this bickering. At the end of the season, the Yankees spend $200++ million and are in the World Series. Now that's Moneyball, and can be booked for the next half decade. The Phillies are no slouches with their payroll either. Will some bottom scraping payroll team (Tampa) get a crumb every now and then? Sure, but doesn't today's baseball economy just make all the SABR work seem fruitless?

 

Maybe Cashman will write a book... "Winning: Snatching Victory from 29 Other Teams." It will have nothing to do with geeked-up stats, and everything to do with explaining when he sees a top free agent, he signs him. A single chapter may suffice, and it should include an explanation as to why even the Yankees wouldn't pay Gabe Kapler $5.2 million for any of his seasons.

 

I still can't wait for the next reference justifying why Melvin paid who what because their WARP/WAR/VORP/WSP/WSAA is much higher than so and so's. Makes for intriguing discussion anyway.

 

Also, one thing zone rating doesn't take into account is how many excessively windy games a certain player plays in, how many games were played on a slick surface due to rain, how injured a specific player is at the time they are playing, or if its a contract year. I'm sure I could come up with more, but I'm done for now. Viva sabermetrics, for it gives many people something to do. We all need things to do.

[Ennder]

The entire purpose of a stat like WAR is to have an unbiased system of judging talent. It does the job fine as long as you understand the various flaws of small sample sizes. If every disucssion about baseball was just I think this vs you think that then we would never get anywhere. WAR at least is a method to add a little validity to the subject. But again you have to use your brain here, nobody is going to argue the value of Kapler based on such a small sample, if you are going to talk how good a player is we need at least 3 full seasons of data. Nobody is saying the system is 100% flawless, but with a large chunk of data it points to a relatively solid player value. Much better than just saying I like this guy he must be worth $10M a year or this guy stikes out a lot he can't be worth $10M.

[brewjihad]

Ok on the money thing. What they have done each season is add up the total WAR of players on the free agent market and add up the total dollars paid to those players in that year. They divide that to figure out the average amount paid for 1 win above replacement level. Kapler was 1.2 wins above replacement level, and on average, teams pay $5.2 million for that kind of production. If people don't like the dollar figure, they should just ignore it. I've heard the argument that the dollar figure should just be looked at as ranking and for comparison and not as literal dollars, but that's not a good idea because the amount a baseball team pays for a win is different each year. So a 2002 3 WAR season is worth less than a 2008 3 WAR season.

 

I've also seen people complain about the fact that about $5 mil per win is too much, but they're completely ignoring players in the first 6 years of service and only looking at how much free agents are worth with this column. And you also have to include the fact that players sacrifice money for long-term certainty, so the dollar number is like a 1-year deal fair value for the player last year.

[mletto157]

Is it possible that the ratings for Braun's defense is so bad because of how good Cameron is? Since Cameron in CF is so good, he may catch a lot of balls that are in Ryan's zone. If you replace Cameron with a poor fielding CF, maybe Braun's rating improves because he will be counted on a lot more to get to those balls in the gaps that Cameron used to be the one getting to. Or then again, maybe it decreases because in trying to get to those balls, he botches more plays. I don't know, I guess it's just a random thought I had in trying to explain this. I completely agree that Braun isn't a good defender, but how bad he is, is definitely still in question.

[sheethead]

But again you have to use your brain here, nobody is going to argue the value of Kapler based on such a small sample

 

What does sample size have anything to do with valuing .239/.329/.439 in 238 plate appearances at $5.2 million? The only thing I see, is that if Kapler put up the same numbers in 620 plate appearances he would be worth only that much more per Fangraphs.

[thebruce44]

If someone is arguing that I am trying to invalidate the methodology, they are missing the point. My argument from the beginning of this thread was the dollar value Fangraphs is placing on WAR.

Then why post this?

I also feel badly for people who believe they understand exactly what goes into computing WAR, when in actuality there is no way most people could possibly know all the variables Fangraphs uses to come up with their numbers. Key variables are proprietary information, and not available to the public.

Or this?

I simply am not in tune with having to multiply by 2.5, put to the power of the wind/continental divide, doubled and then taken as the reciprocal to end up at the conclusion that Nyjer Morgan is more valuable than Ryan Braun. Its comical to me.

How can you expect people to not defend or try to explain to you the methodology and usefulness of a metric you mock? Then, when someone provided information you asked for, you told them to stop insinuating that you didn't already understand what you asked for.

I go back to the point of this thread... $5.2 million for Gabe Kapler's .239/.329/.439 in 238 plate appearances??? Not even close.

And nobody is arguing with you on that one. Everyone seems to be in agreement that the value statistic on the site isn't very useful unless you understand it is to be taken in a vacuume. The debate started when you and several others claimed that WAR was arbitrary and proprietary information that was useless. That screams a lack of knowledge on this particular subject to me so I, like others, have tried to provide you with information to help you make a more informed decision.

I am not trying to call you out, and I think you make a lot of valid points here, but I am also not sure what you expect me and other posters to do.

[Baldkin]

Where's the evidence that points to Braun missing balls that other LF'ers in the league are reaching? Even if other LF'ers are getting more chances per game or per week, how do we know that's not due their teams' pitching staffs?

Chances per week are much more determined by pitching staff (GB/FB ratio) than by how many innings a player plays. The Brewers had a pretty high FB% staff this year (not many K's and no real sinker balling starter). The evidence is in UZR, the ratings of each "Pie slice" determining that the average OF makes more plays in given parts of the field than Braun does.

 

Here's my terrible drawing of what UZR tries to represent:

http://img519.imageshack.us/img519/965/uzr.gif

 

So, for a given type of hit (fly ball or line drive) an average LF makes the plays in each one of those zones a certain percentage of the time. Anytime an out is made, the fielder gets + runs, based on the inverse of that percentage. (An out vs a non out is .8 runs, so say the play is made 60% of the time, if the fielder makes the out, then he gets credit for .4 outs, which you can then multiply by the .8 runs/out to get the number of runs it is worth)

[Thurston Fluff]

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. 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.

 

I guess the point I was making is if a system is used to show a players value by attaching a $ figure to it and it doesn't match up with what that player actually got paid it appears to me that something is wrong. Either with the owners and GM's or fangrpahs.

To believe in the results of fangraphs when it goes counter to what reality is means one has to think fangraphs knows more about determining players value than the actual teams.

That very well may be the case at least in some instances. Then again if fangrpahs did field a team it may be a AAA level team with a Yankee payroll. The fact they rate players like JJ Hardy as being worth $21 million seems to me to suggest the latter but that is just me.

I think some of us get defensive when others suggest those of us who think actual GM's and owners might be the ones who are right (when there is a discrepancy) are ignorant or do not understand the methodology seems to me a bit off base. Maybe even missing something of significance fangrpahs missed.

Overall I think my problem is more with fan use as it pertains to $ value than the actual system itself. I think theBruce44 had it right and they should just eliminate that column.

[joepepsi]

Thank you for your explanation, Baldkin. Your graphic is much better than anything I could come up with!

 

Still, I'm not completely sold. Seems like if you play defense for a crappy pitching staff then your metrics are bound to suffer. Brewer pitchers may have given up lots of screaming line drives off the LF wall. Braun couldn't stop them; possibly no defender could, therefore he allowed 15 more runs than an average defender would. Especially one that plays with a better pitching staff.

 

It's relatively simple to come up with a figure for an average offensive player. Seems to me it would be very difficult to do the same for defense, and almost impossible to accurately compare a players' defense to this mythical average.

[brewjihad]

I guess the point I was making is if a system is used to show a players value by attaching a $ figure to it and it doesn't match up with what that player actually got paid it appears to me that something is wrong. Either with the owners and GM's or fangrpahs.

 

The owners are playing for projected performance. It wouldn't be right if the numbers did match up, because you're comparing salary to actual value in salary.

 

That very well may be the case at least in some instances. Then again if fangrpahs did field a team it may be a AAA level team with a Yankee payroll. The fact they rate players like JJ Hardy as being worth $21 million seems to me to suggest the latter but that is just me.

 

This is where the measure is misunderstood. Pre-arby and arby players are paid less than they would be worth on the free agent market. Market value for Hardy on a 1-year free agent deal putting up the year he did was about $20 million.

[logan82]

Braun playing a few more steps back so balls don't go over his head would lead to more falling in front of him. I don't think it is a stretch at all to say more balls fall in front of Braun than the average fielder.

 

Seems like if you play defense for a crappy pitching staff then your metrics are bound to suffer. Brewer pitchers may have given up lots of screaming line drives off the LF wall. Braun couldn't stop them; possibly no defender could, therefore he allowed 15 more runs than an average defender would

 

All bated balls used in UZR are designated as soft, medium or hard. Hard hit balls are accounted for.

[sheethead]

Then why post this?

 

I also feel badly for people who believe they understand exactly what goes into computing WAR, when in actuality there is no way most people could possibly know all the variables Fangraphs uses to come up with their numbers. Key variables are proprietary information, and not available to the public.

 

Because I still haven't seen anyone come up with Fangraph's version. I missed that word, and for the confusion causing sloppiness I apologize. But when I do feel somewhat vindicated when I see someone calculate Fangraph's WAR, then state:

 

Now I checked fangraphs, and their value says 43.0 runs above replacement, and 4.3 wins. So I was pretty close. Looks like the hitting is 2 runs different, don't know what would cause that.

 

Ahhh, we don't know what would cause that? THAT is what I am and have been saying.

 

This illustrates my point about how, in actuality there is no way most people could possibly know all the variables Fangraphs uses to come up with their numbers. If you can't get to their figure, then its a secret, right? But then the argument becomes, well its close enough and you are missing the point. Its a circular argument that those who bought in don't see what I am saying, even though I understand completely what they are saying. Someone even called it MIDDLE SCHOOL MATH!! I have no recollection of middle school math involving an infinite number of correct answers. Granted the multiplication, division and addition are middle school, but the variables are not given which is why everyone comes up with different answers. But that's okay, because its all close enough? Most of the formula is available, I get it. I argue Fangraphs has tweaked it and now we don't know why there are differences with their numbers versus what you or I can calculate on our own.

 

When I "mocked" it was not directed at the methodology or the parameter, it was the fact that we don't know why there is a difference with Fangraph's calc.

 

I am not trying to call you out, and I think you make a lot of valid points here, but I am also not sure what you expect me and other posters to do.

 

Lock the thread? Heh. I apologize for acting too defensively. I don't need any reason why people like the metric, I see why people like it.

[mletto157]

What does sample size have anything to do with valuing .239/.329/.439 in 238 plate appearances at $5.2 million? The only thing I see, is that if Kapler put up the same numbers in 620 plate appearances he would be worth only that much more per Fangraphs.

 

Sample size is huge here because of the calculation of UZR. Most likely his defense is way overvalued because he performed well over a very small sample. If you took his UZR rating over the next couple seasons and combined it with this one you would get somewhere close to his actual skill level (actually you'd probably need more than that with Kapler since he is only a part time player). However, none of this takes away from the fact that WAR (and therefore the $5.2M figure) is based on ACTUAL production from this last season, not true skill level. This is why it seems hard for people to grasp. And this is one of the flaws of WAR and the dollar value. Because the $ comes from only this last season of data, it doesn't accurately measure talent - only worth for that season. If you used the same type of formula, but over the last 2-3 seasons (or the next 2-3 seasons) you would get a better approximation of true talent level and true worth in the free agent market. I guess I don't understand why you are having a problem with this sheethead. This is a purely objective evaluation of a players worth based on what the average team is willing to spend for a win in the FREE AGENT market. It's very important to remember that this doesn't take into consideration pre-arby players, arby players, and players who never hit free agency, but sign long term deals.

[mletto157]

sheet, did you miss this post from rluzinski?

 

Fangraphs tweaked Tom Tango's methodology and will come up with a slightly different number than him. They had something like a 7 part series outlining exactly how hey calculate it. Others are free to tweak, refine and improve as well. It's a model. If somone is arguing that one player was worth 2 runs/.2 wins more than fangraphs indicates, they missing the point.

[sheethead]

All bated balls used in UZR are designated as soft, medium or hard. Hard hit balls are accounted for.

 

Is defensive positioning factored in (where the outfielder was positioned prior to the ball being put into play)? I could see numerous close games causing a fielder's zone rating to suffer depending on whether they are playing shallow/deep or in/dp dept.

[sheethead]

It's very important to remember that this doesn't take into consideration pre-arby players, arby players, and players who never hit free agency, but sign long term deals.

 

So I take it a signing of a Sabathia or Teixeira type deal would significantly inflate the dollar values Fangraphs attaches to their player's WAR. Is that correct?

 

sheet, did you miss this post from rluzinski?

 

I got it, but I missed the seven part series on what goes into their calculation. It would need for it to be part of my job in order to care that much. Guess it was made available, my badhttp://forum.brewerfan.net/images/smilies/ohwell.gif

 

I'll quit now while I'm behind, and promise to stay away from this topic going forward.

[Thurston Fluff]

The owners are playing for projected performance. It wouldn't be right if the numbers did match up, because you're comparing salary to actual value in salary.

 

I fully understand that. But the value in the contract is based off of something and that something was what players really get paid. So there has to be some amount of connection between the two. I don't think the disconnect is the fault of fangrpahs as much as it is the use of it. To say Kapler should be worth more than he got because fangraphs says he is discounts the reality that he didn't get it. IF you want to say according to fangrpahs he will likely perform above his contract fine. To say he was actually worth $5 million is up for more debate.

 

I don't think I'm explaining myself well so I'll try one other tact.

We all know there are ways to find players and the most expensive one is free agency. If I understand it right Fangrpahs bases it's player values on FA market value as compared to performance. Yet market value is not set solely by performance. It is set by demand for that particular player and whether or not teams who need him have cheaper alternatives. His value is based off a much wider array of things than simply production. Thus to use only production is likely to get numbers that will not correspond to real value.

 

This is where the measure is misunderstood. Pre-arby and arby players are paid less than they would be worth on the free agent market. Market value for Hardy on a 1-year free agent deal putting up the year he did was about $20 million.

 

I understand. I used the Hardy example because that is one that many of us think was off base not because he was actually a lot cheaper than that. Part of the value of free agents is their true abilities are better known than a 3rd year player. CC Sabathia got his payday because he proved himself to be a durable top flight pitcher for 6 years. I'd venture to say he would be worth more money on the open market than Tim Lincicum because the jury is still somewhat out on him. Yet fangrpahs would ignore that in estimating what they are going to be worth next year.

 

JJ hardy might have looked to be worth 20 million but he would never have got that if he was one the open market after that season because nobody would have been sure enough of his continued production to offer it to him. He simply hadn't proven himself yet. If he couldn't get what fangrpahs said he was then why say he was worth that much?

 

The other aspect of it is some teams who have no better alternatives and are competing wit other teams for one player his value will be higher than his performance because of circumstance not ability. Thus using ability as the measuring stick for all players based off inflated players pay will almost always cause the calculation to be well over their true value.

That is going off what I read from TLB's post in the other thread a while back. If my memory is wrong and it bases it off all players contract, arby and pre arby included then my bad. Never mind.

[TheCrew07]

Brewjihad I admire your zeal to explain WAR and the dollar. However, the point about the dollar value that you're missing is that many of us think the FA market is not a good measure what every team in the league is willing to spend per win. As the league is top heavy, only certain percentage of the teams are paying max dollars for the best talent. Everyone else is trying to maximize production for the dollar... 1 fair market contract for a player can burden a small market team like Milwaukee as evidenced by Suppan, and when was the last time any team in the bottom half of the markets signed a marquee FA?

 

Just looking around the site at the leaderboards, many FA players are signed to deals that are around 60-75 percent of their fangraphs' value. Every year 3 or 4 players really get paid, and they get paid so much that it pulls the average up so high that the value fangraphs is putting out there is actually significantly higher than than player's FA contract value, even for a guy that performs like Sheets does when he's healthy. No one is giving him $18 million per year, just like no one is paying Kapler $5.2 or whatever. I'm just talking players that performed well here and excluding players that have had down seasons since signing their contracts.

 

Mostly though you're right, I just ignore the dollar value until someone comes here and points to it as justification as to why someone is worth X because that's what their fangraph's $ value is. That sort of analysis really gets under my skin, because it's flawed from the start... given the sliding scales of economy from the Yankees all the way down to the smallest market here in Milwaukee, and the disparity between salaries of players prior to hitting FA and after hitting FA. The entire system is on shakey ground at best given all of the "adjustments" that really can't be proven to be true. In the end, it's just an opinion, there's simply no way to prove the adjustments, park factors, etc to be true in a classical geometric or calculus type proof, it's all just theory.

 

WAR is fine for what it tries to accomplish, but it shouldn't be a stretch to see why so many of us have a problem with the methodology involved in arriving at the $ value, even a casual glance between Cots and Fangraphs should be a enough to bring pause to a thoughtful fan. I'm not usually one to gripe without offering a solution, but in this case I'm not sure how they'd arrive at a more accurate value each year, which is why I've posted in the past that I wish they would just take it down. I've thought about this quite a bit and they'd probably have to adjust each signing as a percentage of the teams payroll and adjust for market size, the end result being some reduced rate for contracts signed by the Yankees and such while allowing the full value for teams like Milwaukee. There's just so many ways and so many holes in trying to analyze the contracts in that manner that I'd actually just rather guess and hope to be close or settle for being wrong.

[Thurston Fluff]

sheet, did you miss this post from rluzinski?

 

Fangraphs tweaked Tom Tango's methodology and will come up with a slightly different number than him. They had something like a 7 part series outlining exactly how hey calculate it. Others are free to tweak, refine and improve as well. It's a model. If somone is arguing that one player was worth 2 runs/.2 wins more than fangraphs indicates, they missing the point.

 

This shows a little why some do not buy into any one place or system for definitive answers. The simple fact that tweaks and adjustments can be done shows there is some amount of subjectivity in the whole thing. One person thinks one aspect deserves a little more weight than another and prove it through a 7 page formula while another does another 10 page formula to prove why their version is better. Next thing you know you spent half your life reading different arguments about which formula should be used, confuse one for the other a few times, reread them all just to be sure, and try to figure out which objective measurement is better.

The upshot of it all is one person forms their opinion of a player based on one figure and another a different one then argue the merits and everyone agrees/disagrees they are roughly worth X. Which in the end is the exact same thing all of us lesser in depth stats calculating guys do. We look at how well they played, their relative age and health and decide if we think they are worthy of a spot on our team or not.

[rluzinski]

Comparing a model that is constructed to estimate reality to adding and subtracting? My goodness. Engineers and scientist create models all the time. Some are better than others. A crude model got men to the moon, so prehaps their is utility in a model, even if it doesn't account for all variabes.

 

UZR considers positioning part of the skill of being a defender. Now, positioning is often dictated by the coaching staff, so that isn't always fair. As soon as we can get pre-play positioning data, it won't be an issue.

 

A lot of UZR's weaknesses are diminished with increasing sample size. For instance, the ball speed data is very course and somewhat subjective. Over 4 seasons, it sort of corrects itself. Over 70 games, not so much. It's like trying to measure the value of a ball in play when all youknow his hit location and whether the ball was hit soft, medium or hard. Give me 1500 balls in play and I bet I could get close. 100? No way. So the UZR based defensive value calculations have a whole extra layer of uncertainty than the offensive calcs.

 

And I'm dismayed thata people continue to say that it's obvious that fangraphs can't predict what players get paid. They can't because they aren't even trying to. If we want to start predicting contracts for this free agents this offseason, we'll need to generate projections first. What player shall we do it for?

[rluzinski]

Hits and misses but great reads:

 

http://www.insidethebook.com/ee/index.php/site/comments/sabermetric_moves_of_the_2009_pre_season/

 

http://www.insidethebook.com/ee/index.php/site/comments/sabermetric_moves_of_the_2008_pre_season/

 

Pitchers are damn hard to project.

[joepepsi]

The upshot of it all is one person forms their opinion of a player based on one figure and another a different one then argue the merits and everyone agrees/disagrees they are roughly worth X. Which in the end is the exact same thing all of us lesser in depth stats calculating guys do. We look at how well they played, their relative age and health and decide if we think they are worthy of a spot on our team or not.

And those using lesser in depth stats usually come to similar conclusions. Which raises my point that all these complicated formulas don't really add much more insight that what is available through reasonable observation and evaluation of rate stats.

[Ennder]

And those us lesser in depth stats usually come to similar conclusions. Which raises my point that all these complicated formulas don't really add much more insight that what is available through reasonable observation and evaluation of rate stats

 

But they do, they point out things like why Dunn was overvalued before this years contract (GMs put a lot more value on defense this year so he didn't get as much as expected), they also point out why some players are way undervalued (like Cameron). The fact that most teams are now using advanced sabermetrics when deciding on player signings has just changed the overall outlook of things. 20 years ago the average fan's opinion of which players were valuable and which weren't was horrible, even most GMs were pretty bad at it. Now most of them have a clue.

 

All a stat like WAR is doing is weighting defense vs offense vs position. That more or less sums up the entire process. Sure you can do the same thing by just discussing all 3 areas separately but I think it is pretty obvious that it is important to include all 3 in every discussion of a positional player. While we might not always agree on how much to value each of the 3 I think WAR is an ok starting point to most discussions.

[rluzinski]

BA, OBP and SLG are subjective measures of a player's worth. They have to just become accepted. If people are looking for absolute truth, you'll have to stick with philosophy. That's not how science works. Create a model, test a model, improve it, repeat. If someone isn't familiar with the scientific method, I can see why the whole thing could make someone uncomfortable.

 

JJ hardy might have looked to be worth 20 million but he would never have got that if he was one the open market after that season because nobody would have been sure enough of his continued production to offer it to him. He simply hadn't proven himself yet. If he couldn't get what fangrpahs said he was then why say he was worth that much?

 

This is exactly what we keep trying to explain. Even if Hardy had been a free agent for the 2009 season, he wouldn't have been projected to repeat his 2008 performance, so we wouldn't have expected him to get anywhere near $20 mil. Hardy's performance in 2008 was estimated to be worth $20 mil; HE wasn't worth $20 mil/year. Like you said, it's when a player consistently performs at a particular level that his performances aligning with his projection.