Whitlock: "Stat geeks are ruining sports"

[topper09er]

The problem I have with advanced stats is that some are not true math. Uzr and WAR are examples. Sure some amount of these stats involve math to make them look and feel objective, but they're not.

What is subjective about WAR? Read through this and show me where any subjectivity comes in:

http://www.insidethebook....le/how_to_calculate_war/

What is NOT subjective about it? Arbitrary "value" given to different positions, Uzr is part of the formula, estimated factors for games played, etc. But the biggest problem is the junk science that is the leap between above/below avg performance and win shares. It is impossible to conclude player x is worth 2.7 wins a year. There's just now way to prove it.

The values are not arbitrary:

 

"Since players move around (albeit infrequently) from position to

position, and we have decent defensive measurements, we can compare

positions more or less directly. Well, most positions: we have the

problems of dealing with infield to outfield and the issue of left

handed players not being able to play third base, shortstop, catcher,

and second while being favoured at first. But these things can be taken

into account by tweaks to the calculations, and after some work you (by

you I mean Tom Tango) end up with the following values for our

positional adjustment (in runs per 600 PA):

http://assets.sbnation.com/assets/299646/posadjust.png

Please, please note that these values are not set in stone. You could

tweak a lot of the numbers up and down by a run or two and nobody would

bat an eyelid. This is a guideline, and while it’s close it’s not

perfect by any means. The second thing to pay close attention to as that

those are run values per 600 plate appearances. In reality, while plate

appearances will obviously dovetail pretty closely with defensive

chances, it’s not an exact match. We’re also exposing ourselves to a

significant selection bias by only looking at players considered capable

of moving around the diamond. Furthermore, these adjustments look at

the sum of raw defensive ability – they don’t take into account how an

individual position stresses different tools (arm is more important in

right field than left, for example). None of these points are enough to

recommend not using the above values, but as with everything we look at,

the holes in the model are instructive." (http://www.fangraphs.com/.../positional-adjustment/)

 

 

 

Even if you think those values are off they do not make a huge difference in the calculations, so you can look at WAR and round up or down depending on which way you think the positional adjustment should go. And if you dont like the defense part just ignore it, look at offensive WAR or wOBA.

 

[FVBrewerFan]

Of course the position weighting is subjective. Why else would each position by exactly .5 different? What a coincidence that each postion would be exactly .5 more/less valuable. UZR may not be accurate but that has no bearing on subjectivity? Ok, let's call it a guess then. The replacement part is arbitrary as well.

 

But that's the problem. Each step of the process could be a little off, and by the time you get done some players could be quite a bit off in total. It's how you end up with lunacy like Ellsbury having a much higher WAR than Fielder acouple years back. I don't deny WAR is a good ATTEMPT at conclusive evidence of a player's value. But that's all it is, an attempt.

[ramssuperbowl99]

Recap of the discussion with FVBrewerfan:

FVBF: "Positional weighting is subjective."

Rams/Topper: *A whole bunch of mathematically backed information demonstrating that while they are not exactly precise, positional weights are not subjective*

FVBF: "Of course the position weighting is subjective."

[topper09er]

Of course the position weighting is subjective. Why else would each position by exactly .5 different? What a coincidence that each postion would be exactly .5 more/less valuable. UZR may not be accurate but that has no bearing on subjectivity? Ok, let's call it a guess then. The replacement part is arbitrary as well.

 

But that's the problem. Each step of the process could be a little off, and by the time you get done some players could be quite a bit off in total. It's how you end up with lunacy like Ellsbury having a much higher WAR than Fielder acouple years back. I don't deny WAR is a good ATTEMPT at conclusive evidence of a player's value. But that's all it is, an attempt.

Uncertainty is different from being subjective. The numbers are obviously rounded (to the nearest half run over an entire season!) because the calculations cannot possibly achieve that level of accuracy due to the sample size.

 

Your original statement that anyone who uses WAR as the center of their argument should immediately be ignored is way off base, you have not shown any good reason to do so other than in your opinion WAR tells you things that your intuition does not agree with. To say WAR is useless because the positional adjustments may be off by a few runs over the entire season.

 

Also what alternative do you suggest, if you disagree that Ellsbury could more valuable than Fielder what stats do you recommend to back up your stance? I assume you are talking about 2008, according to fangraphs Fielder put up 1.7 WAR and Ellsbury put up 4.3.

 

WAR Breakdown:

Season	Team	Batting	Base Running	Fielding	Replacement	Positional	RAR	WAR	Dollars	Salary
2008	Brewers	25.1	-9.8	-9.2	23.1	-12.3	17.0	1.7	$7.6	$0.7

Season	Team	Batting	Base Running	Fielding	Replacement	Positional	RAR	WAR	Dollars	Salary
2008	Red Sox	-0.4	4.1	21.2	20.3	-2.5	42.7	4.3	$19.2	$0.4

Traditional stats:

Season	Team	G	PA	H	HR	R	RBI	SB	AVG	OBP	SLG
2008	Brewers	159	694	162	34	86	102	3	.276	.372	.507

Season	Team	G	PA	H	HR	R	RBI	SB	AVG	OBP	SLG
2008	Red Sox	145	609	155	9	98	47	50	.280	.336	.394

 

So the big difference comes from defense, the biggest problem with WAR. Single year defensive values are not very reliable and more like a 3 year average is better. Fielder's average from '07-'09 = -4.9 runs, For Ellbury Ill just use his career average since its about 3 full seasons, = 9.3 R/150G.

 

If you use those values you get Fielder = 2.1 WAR, Ellsbury = 3.1 WAR.

 

Whats wrong with that? Fielder is way above Ellsbury hitting and way below in everything else? Do you not agree with that? How do you justify Fielder being more valuable than Ellsbury? The important thing to note is that WAR says who is more valuable, not necessarily better. It is more valuable to a team to have a the same hitter play OF than 1B, take Jeter as an example. The Yankees are better because he can play SS instead of 1B. The way it is now the Yankees have Jeter and Texiera, but if they have Jeter at 1B and Marco Sucaro at SS they are worse off, even though Jeter is providing the same offense from either position.

[homer]

topper09er wrote:

 

So when Christian Laettner goes a perfect 10-10 from the floor and 10-10 from the line in the 1992 Elite 8 it ALL random luck, no room whatsoever for an increase in performance from being "clutch" or just focusing more when it really matters? And you know this for a fact as much as you know 1 + 1 = 2? You will not concede that it was possible for Laettner to control his performance at all instead of him just simply getting lucky?

I don't have Laettner's numbers overall but isn't it possible he also had a crappy game at some clutch time? How do you explain that? Did he choke one game and come up clutch in the other?

[Ennder]

"So when Christian Laettner goes a perfect 10-10 from the floor and 10-10 from the line in the 1992 Elite 8 it ALL random luck, no room whatsoever for an increase in performance from being "clutch" or just focusing more when it really matters? And you know this for a fact as much as you know 1 + 1 = 2? You will not concede that it was possible for Laettner to control his performance at all instead of him just simply getting lucky? "

 

No but the reality is it could just be random luck so there is no way to know if it is clutch or luck. These kind of streaks happen all the time in simulated leagues which proves that even if you take the human element out of the equation they are going to happen just randomly. There is no good way to know if it is random or some sort of skill shown by the player. It makes it extremely hard to judge what causes streaks like this when it makes completely sense that it could just be random noise.

[TheCrew]

Whitlock is garbage and has been for a long time. I stopped reading his drivel after his "theory" that NFL teams with more white players, won more games.

 

All that said, he has a point to an extent - a number of posters around here are guilty of it as well - they point to things like UZR and WAR to try and summarize a players worth and act as if it is the end of the discussion. Yet they ignore things like UZR saying Casey McGehee has good range this season, or Nyjer Morgan had equal value to Bruan a few seasons back to name a few.

 

Sometimes common sense has to come into play - but anyone who visits this board knows there are plenty of stat-heads that like to point to UZR or WAR to summarize an entire players worth. To me, using these flawed stats (at times) is as lazy as Whitlock's writing.

[TheCrew]

But that's the problem. Each step of the process could be a little off, and by the time you get done some players could be quite a bit off in total. It's how you end up with lunacy like Ellsbury having a much higher WAR than Fielder acouple years back. I don't deny WAR is a good ATTEMPT at conclusive evidence of a player's value. But that's all it is, an attempt.

Quoted for Truth - the big problem is that some people present these stats (like WAR and UZR) on the same level as something that is strictly based off the math, like OBP.

[RockCoCougars]

I'm pretty dubious of WAR, because I believe that it overrates some positions by subjectively assigning them a value. Most specifically, center field. Though it may be more difficult to play center at a high level, I don't really think that it's that difficult to play adequately. More specifically, if you are bad in center, you are likely to be bad in the corner positions as well. On the other hand, if you are solid at the corner positions, you should be decent in center. Why should a guy be penalized if he is blocked and forced to play corner outfield because the team has an established CF? It seems to me that almost every guy that appears vastly overrated by WAR is a centerfielder (Mike Cameron, I'm looking at you) who puts up decent offensive stats relative to his CF 'peers'. Going further, shortstop is more difficult to play than center, but is it really that much more difficult than third?

[logan82]

Going further, shortstop is more difficult to play than center, but is it really that much more difficult than third?

 

Yes. How many 3B do you think could play a decent SS? They would have Yuni range in many cases.

[RockCoCougars]

Going further, shortstop is more difficult to play than center, but is it really that much more difficult than third?

 

Yes. How many 3B do you think could play a decent SS? They would have Yuni range in many cases.

At a passable level, many. The throws are about the same, obviously the differences are in turning the DP and the fact that you need almost 50% more range. I will grant you that it's harder to play shortstop at a high level than any other position, but most SS don't play at a high defensive level. My argument is that short isn't that much more difficult to play adequately than third, nor is center that much more difficult relative to right or left.

[sveumrules]

Anyone that quotes UZR or WAR as absolute is not a "stat geek" as a true geek would understand the limitations of those metrics and not treat them as infallible. It seems to me the real problem is not the stats themselves but rather the lazy people who misuse them without understanding what they really mean.

[topper09er]

because I believe that it overrates some positions by subjectively assigning them a value.

 

Well I believe that the Earth is flat, but as it turns out that is not correct. The values are not subjective, they may not be based on a nice empirical formula derived from tons of data like linear weights, but they are based on some data.

 

On the other hand, if you are solid at the corner positions, you should be decent in center.

 

These numbers are based on guys who switch positions, so guys who are average in LF/RF are 2.5 runs below average in CF over the season, and guys who are average in CF are 7.5 runs above average in the corners over the season. Not perfect but based on some data. And it is still possible to have a positive WAR on defense from anywhere, good defenders will show a positive value even at the penalized positions (Pujols) and bad defenders will show a negative value at a rewarded position (Jeter)

http://assets.sbnation.com/assets/299646/posadjust.png

It seems to me that almost every guy that appears vastly overrated by WAR is a centerfielder.

 

The CF boost is 2-3 runs for an entire season in CF, so at most 0.2 - 0.3 WAR which is basically negligible.

[LouisEly]

I know less about zen than I do about stats, but here's my effort to be zen: In my humble opinion, John Briggs and Louis Ely know a heck of a lot about baseball. I learn many interesting and worthwhile things from their analyses of players and situations, quite frequently. Therefore, at least from my limited perspective, their prominence in discussions on this board must be judged a good thing.

 

Now I'm going to stay away for a while.

Thank you Greg, and I enjoy reading your posts as well. I do want to emphasize that I am a stats person, and I reference stats frequently in my posts. I just believe strongly that stats aren't the end-all of the discussion. Stats definitely are a part of the picture, however they aren't the whole of the picture.

 

Don't stray for too long.

[topper09er]

I don't have Laettner's numbers overall but isn't it possible he also had a crappy game at some clutch time? How do you explain that? Did he choke one game and come up clutch in the other?

YES!

 

He is human being, not a fair die.

[rluzinski]

"If you find a player that is less streaky over the course of "x" number of games, it is only because of random chance."

 

So when Christian Laettner goes a perfect 10-10 from the floor and 10-10 from the line in the 1992 Elite 8 it ALL random luck, no room whatsoever for an increase in performance from being "clutch" or just focusing more when it really matters? And you know this for a fact as much as you know 1 + 1 = 2? You will not concede that it was possible for Laettner to control his performance at all instead of him just simply getting lucky?

Again, that is not what I'm saying at all. Was it possible that Laettner somehow willed himself to momentarily being a better shooter than he otherwise was? Sure, why not. As always, I can't rule it out. Did he will himself to having a 100% chance of making all those shots? Of course not. That's common sense, isn't it? His performance is still governed by the laws of probability. Even if he had somehow willed himself to having an 80% chance of making each of those 10 shots from the floor, the probability of him making 100% of them was still only 10%. As with just about everything in life, luck played a significant role.

I can see how that might take some of the magic away from exceptional sports performances but it's reality. If you don't want to tarnish the mystic of it all, don't debate me or any other stat geek about probability theory or Bernoulli trials.

[topper09er]

Did he will himself to having a 100% chance of making all those shots? Of course not. That's common sense, isn't it?

I disagree here, why was it not possible? I don't think it is an obvious common sense thing. You don't think it was possible for him to just take it 1 shot at a time, focus on perfect form/follow through?

 

If I told you that you had to make 3 straight free throws or you would die don't you think you would approach them differently than if they were 3 random free throws in a pick up game?

[topper09er]

Even if he had somehow willed himself to having an 80% chance of making each of those 10 shots from the floor,

This is my main point, that the player has control and not just an insignificant amount. If he went from a .575 shooter (his actual number that year) to a 75% shooter, that increases his odds of going 10-10 by 1400%, pretty big. I would say that is a lot of control, obviously luck and randomness is in there, but the player can control it more than you give them credit for, in my opinion.

[kramnoj]

Why should a guy be penalized if he is blocked and forced to play corner outfield because the team has an established CF?

 

If you have a CF quality defender playing in a corner, he is going to have more range, and won't be penalized, because he is making more plays than the peers at his position.

Going further, shortstop is more difficult to play than center, but is it really that much more difficult than third?

 

Pujols started off his career at 3B, imagine how much better his offensive numbers would have compared against shortstops.

 

I think there's a certain level of defense that teams will tolerate. It might have been worth it for the Brewers to try Gamel at SS in the minors this year, so he could have replaced Yuni at some point. He may be a -25 or -30 defender, but if managed to put up an .800 OPS, it would have been a decent tradeoff, maybe even a good one.

 

But teams don't seem to try to do that. Look at catchers. Catchers offense is just terrible for the most part. If you were Brendan Katin and you put up a .845 OPS in the minors over your career, wouldn't it be worth a try to learn how to play catcher in the offseason? But catcher defense is really, really important, so teams don't try to do stuff like that.

[brettac1]

I certainly lean toward statistics as something I enjoy when it comes to the game. However, I definitely understand the people who don't care for them. To me, when it gets to the uber advanced metrics that I have zero idea on how they come to exist, it loses me. I don't want to see some stupid heat chart that makes zero sense. Look at something like SIERA, that basically was a junk stat. There are loads and loads of them, and more and more noise every week in the statistical "community", and they don't make the game clearer or easier to understand, or events easier to value. It just muddies the waters more and more and creates more skepticism.

 

I mean, I love statistics, etc. I love when something comes along that makes me go "yes, that makes sense". Lately, it seems like that hasn't happened as much and you have a lot of elitism and jackassery. Fan Graphs, for instance. I've begun to love the people that make fun of it (the "Fan Graphs as Tweets" page, etc.) , because most of their stuff is just silly. I like when people are actually talented writers, like most of the BP roster was.

 

This article pretty much sums up how I feel lately:

 

http://www.lookoutlanding...roblem-with-sabermetrics

[homer]

I don't have Laettner's numbers overall but isn't it possible he also had a crappy game at some clutch time? How do you explain that? Did he choke one game and come up clutch in the other?

YES!

 

He is human being, not a fair die.

Right. So he's not going to go 10 - 10 every game or 0 - 10 every game. He's going to be somewhere in the middle. Stats just say "No he didn't come up clutch, he just moved closer to what we could normally expect out of him over time." That's the whole sample argument. You can't use one game or one stretch of games to say whether or not a guy is clutch. There's just not enough evidence to say one way or the other.

 

People will probably now call Braun a huge clutch hitter because he's hit important home runs twice and that's what people remember. They don't remember the many times late this season when he wasn't clutch in similar situations.

[topper09er]

You can't use one game or one stretch of games to say whether or not a guy is clutch. There's just not enough evidence to say one way or the other.

Why not? He was clutch for that game? Maybe there is no such thing as a clutch player just clutch plays.

 

I made this point way back on the first page and here it comes again, just because you cant prove something exists doesn't prove it doesn't exist, yet your stance seems to be since I can't prove to you if someone is clutch or randomly lucky they must be randomly lucky.

 

I get that randomness exists as a property of the universe as much as gravity does, but randomness applies to random things, and if you can control the outcome something coin flips do not apply. If there is a fair coin but I have the "skill" of flipping mostly heads, then I am not getting lucky and Bernoulli p(heads) = 0.5 does not apply because the coin really has a p(heads) > 0.5. And if I outperform the expected number of heads it is not because I got lucky, it is because I am really good at flipping heads. The model used to calculate my expected number of heads was flawed, so it cannot be relied upon. If it was redone using the exact p(heads) I really possessed and I outperformed that then I got lucky, but since there is no way to know my p(heads), or Braun's p(playoff clinching 8th inning 3-run HR) then I think you cannot know how much luck is in there.

[rluzinski]

Did he will himself to having a 100% chance of making all those shots? Of course not. That's common sense, isn't it?

I disagree here, why was it not possible? I don't think it is an obvious common sense thing. You don't think it was possible for him to just take it 1 shot at a time, focus on perfect form/follow through?

 

If I told you that you had to make 3 straight free throws or you would die don't you think you would approach them differently than if they were 3 random free throws in a pick up game?

If you think players can will themselves to having a 100% chance to hit a free throw (or have 100% chance of hitting a free throw for any reason), we will have to agree to disagree. I find the very suggestion, silly and all the math is on my side. This conversation has turned into trying to argue against the existence of god.

[craigharmann]

So...how players are valued and analyzed statistically, NOT PEDS or anything else, is ruining the game of baseball, Whitlock? That's insane. Performance is what matters, and how it is valued and evaluated changes. There was a time where there was no statistic called a "save" I'm baseball. There was also a time when a statistic called a "sack" was not counted. The game itself doesn't change, evaluation of performance does. Go back to writing about things you sorta have a clue on....

[topper09er]

Did he will himself to having a 100% chance of making all those shots? Of course not. That's common sense, isn't it?

I disagree here, why was it not possible? I don't think it is an obvious common sense thing. You don't think it was possible for him to just take it 1 shot at a time, focus on perfect form/follow through?

 

If I told you that you had to make 3 straight free throws or you would die don't you think you would approach them differently than if they were 3 random free throws in a pick up game?

If you think players can will themselves to having a 100% chance to hit a free throw (or have 100% chance of hitting a free throw for any reason), we will have to agree to disagree. I find the very suggestion, silly and all the math is on my side. This conversation has turned into trying to argue against the existence of god.

There is no math on your side since you don't know if it possible to have a 100% chance of making 1 free throw, if that is possible the math is on my side.

 

What are the variables involved? How many of them does the player have 100% control of? If the player has 100% control of all variables involved then how do they not have the ability to increase their chances to 100%?