Any statisticians around? Warning: Math!

[Guest]

If 5x MVP awards and 10x All-Star appearances means something to you, then these are the players who could have done such a feat.

Interpret it as you will.

Accurately interpreted as the 5 best seasons of WAR, nothing more. It is valuable but not in of itself.
Modify it with career numbers and it could potentially have better usage, especially if WAR's deficiency in comparing catchers to other positions is updated.

It's not meant to measure a career. If you want to know how someone did over their career, you can simply check their career WAR.

pWAR is designed to give an average WAR over X years.

I set X to 5, because of how the members of the HOF were distributed at 5 years.

X can be set to 7, 10 as I've done in examples for you. X can be any number of years.

I think there's a misconception that it can only be five years. It can be any amount of years, but the results of all the data supported five, as stated above.

 

[tribefan26]

Yes, I have fully read and understand, however that is the theoretical problem with such a formula. What I am saying is, no one would ever include this in a text book or scholarly discussion; the formula and selective data sets are lacking statistical confidence, that's all. Your ideas are fine, noteworthy, and great for discussion just too arbitrary.

Can you expand on lacking statisical confidence. They are what they are. WAR is descriptive, not predictive. You are correct that a 9.0 doesn't show clear value over an 8.8 - there is a margin of error and there are factors not in WAR.

But I defy you to find two players in the same season that are 4 or 5 apart and say the lower score was bettwer for that season.

The question of what value WAR has in HOF admission as well as what the value of Career WR is separate. But in a season it clearly has value renaking the performances of players.

On situational hitting, SABRmetricians have spent a lot of time trying to measure players that over a career display such a special ability. Under rigourous statistical analysis I have not seen any measurement that adds value.

 

[hofautos]

[quote="Chris Levy":1tjd7adu]

If 5x MVP awards and 10x All-Star appearances means something to you, then these are the players who could have done such a feat.

Interpret it as you will.

Accurately interpreted as the 5 best seasons of WAR, nothing more. It is valuable but not in of itself.
Modify it with career numbers and it could potentially have better usage, especially if WAR's deficiency in comparing catchers to other positions is updated.

It's not meant to measure a career. If you want to know how someone did over their career, you can simply check their career WAR.

pWAR is designed to give an average WAR over X years.

I set X to 5, because of how the members of the HOF were distributed at 5 years.

X can be set to 7, 10 as I've done in examples for you. X can be any number of years.

I think there's a misconception that it can only be five years. It can be any amount of years, but the results of all the data supported five, as stated above.[/quote:1tjd7adu]

It's not meant to measure a career, but you can suggest it can pick the 52 greatest players.
Funny i have it where you said pWAR only means 5 years.
Either way, it is valuable, but can be made better in defining greatness, in manners I have already suggested.
Too bad you refuse to share the data with me, or I would show you how.

 

[George_Calfas]

The math and statistics behind this is solid.

The Statistics are weak at best due to the lack of statistical confidence. I have brought this up previously and it has been side stepped. The fact is when you select a portion of the data set and not the entire range you lower your confidence levels to to flaws in arbitrary selection. You can try to strong arm your discussions and explain your "idea" but WAR is just not replicable in a theoretical manner. Please write the Proof/Axiom if you think I am wrong.

 

[hofautos]

Yes, I have fully read and understand, however that is the theoretical problem with such a formula. What I am saying is, no one would ever include this in a text book or scholarly discussion; the formula and selective data sets are lacking statistical confidence, that's all. Your ideas are fine, noteworthy, and great for discussion just too arbitrary.

Can you expand on lacking statisical confidence. They are what they are. WAR is descriptive, not predictive. You are correct that a 9.0 doesn't show clear value over an 8.8 - there is a margin of error and there are factors not in WAR.

But I defy you to find two players in the same season that are 4 or 5 apart and say the lower score was bettwer for that season.

The question of what value WAR has in HOF admission as well as what the value of Career WR is separate. But in a season it clearly has value renaking the performances of players.

On situational hitting, SABRmetricians have spent a lot of time trying to measure players that over a career display such a special ability. Under rigourous statistical analysis I have not seen any measurement that adds value.

4 or 5 apart...that is a huge % when 12 is a measure the likes of bonds?
I buy it has value in comparing players of the same position, but just out of curiosity how does Brian Mccann's value compare to Left Fielders?

I like WAR, for comparing like players, but the arbitrary numbers used in comparing positions may be the best we have, but is far from ideal.
Even worse is the proposition that pWAR can create a "greatest" 52 players list, disregarding career numbers...where 10 pitchers are present and not one catcher, that is just plain ridiculous. WAR clearly does not appreciate the value of the catcher, or how easy it is to replace one.

If you can get 20 pitchers before one catcher, you better stock up on catchers while there is a wholesale, after all pitchers only pitch one in 5 games (excuse me one in 4.3 games on AVG).

CHEVY>> Historically catchers have just been a bunch of fat guys that squat there and throw the ball to the pitcher.
That's what a catcher is.

 

[Guest]

[quote="Chris Levy":14qtyewz]

If 5x MVP awards and 10x All-Star appearances means something to you, then these are the players who could have done such a feat.

Interpret it as you will.

Accurately interpreted as the 5 best seasons of WAR, nothing more. It is valuable but not in of itself.
Modify it with career numbers and it could potentially have better usage, especially if WAR's deficiency in comparing catchers to other positions is updated.

It's not meant to measure a career. If you want to know how someone did over their career, you can simply check their career WAR.

pWAR is designed to give an average WAR over X years.

I set X to 5, because of how the members of the HOF were distributed at 5 years.

X can be set to 7, 10 as I've done in examples for you. X can be any number of years.

I think there's a misconception that it can only be five years. It can be any amount of years, but the results of all the data supported five, as stated above.

It's not meant to measure a career, but you can suggest it can pick the 52 greatest players.
Funny i have it where you said pWAR only means 5 years.
Either way, it is valuable, but can be made better in defining greatness, in manners I have already suggested.
Too bad you refuse to share the data with me, or I would show you how.[/quote:14qtyewz]

No. I don't suggest that at all.

What I suggest is this. I feel that WAR is accurate in rating a season. I think you would be hard pressed to find a season that received a WAR of 8.0+ and was not 'great.'

Therefoe, I continue to say that it can find X number of average great seasons a player had.

I feel that a player who averaged at least 5 "great"/MVP seasons (8.0+) and 10 All-Star seasons (5.0+) is a great player.

If you believe in measuring a player by a portion of the overall data (X # of years instead of the entire career) then pWAR is a valuable tool that can be used in research of this type.

Yes. I have said that pWAR represents 5 years, because in this thread, each time I have posted pWAR it uses the five year data. I don't want anyone to be confused when I post something like "pWAR 8.2". I want people to be clear that it is a five year average in this thread.

Your idea of 'greatness' as described to me involves character (which can't be numerically measured) and adding a 'magic number' to catchers so they rank higher than they already have. As a result of Rpos, Catchers have already been adjusted with a tested, scientific number, not a 'magic number.'

 

[Guest]

[quote="Chris Levy":1hoqqul0]The math and statistics behind this is solid.

The Statistics are weak at best due to the lack of statistical confidence. I have brought this up previously and it has been side stepped. The fact is when you select a portion of the data set and not the entire range you lower your confidence levels to to flaws in arbitrary selection. You can try to strong arm your discussions and explain your "idea" but WAR is just not replicable in a theoretical manner. Please write the Proof/Axiom if you think I am wrong.[/quote:1hoqqul0]

George.

If a player has a WAR of 8.8, 8.2, 7.2, 4.2, 6.1 over five seasons then he has a total WAR of 34.5 duing that span. It can be correctly stated that over that span he averaged a WAR of 6.9.

The question is, does it mean anything?

To me it does. To you it doesn't.

I find it interesting to be able to compare players using a small sample size, because some careers ended prematurely due to death, injury, war, lack of interest, etc. These careers don't measure up favorably in career rankings because of this, but I think that's it's interesting to take a smaller sample size when all available players were healthy and playing at their full capabilities.

You obviously don't find it interesting, and that's fine.

 

[tribefan26]

Accurately interpreted as the 5 best seasons of WAR, nothing more. It is valuable but not in of itself.
Modify it with career numbers and it could potentially have better usage, especially if WAR's deficiency in comparing catchers to other positions is updated.

How many of the 50 should be catchers - do we figure out a priori how many of each position need to be in the top 50? If we measure only statistical factors it almost a certainity that some positions will be better represented in any list of 50.

I have a problem with a system that that says no catchers belong in the HoF. I don't have a problem with a statistical measure that says no catcher is in a top 50. I have a real problem if someone decides that a statistic solely decides who gets in.

 

[Guest]

Accurately interpreted as the 5 best seasons of WAR, nothing more. It is valuable but not in of itself.
Modify it with career numbers and it could potentially have better usage, especially if WAR's deficiency in comparing catchers to other positions is updated.

How many of the 50 should be catchers - do we figure out a priori how many of each position need to be in the top 50? If we measure only statistical factors it almost a certainity that some positions will be better represented in any list of 50.

I have a problem with a system that that says no catchers belong in the HoF. I don't have a problem with a statistical measure that says no catcher is in a top 50. I have a real problem if someone decides that a statistic solely decides who gets in.

Thank you.

This is not an attempt to state that only the Top 50 I posted should be in the HOF.

It simply says that these players had the best 5 year runs in terms of WAR.

I feel that it's significant if a player could average a WAR of 8.0+ for 5 years.

If you don't feel that's significant, that's fine.

 

[hofautos]

^^^ depends on the objective.
If you want to create a list of the "greatest" 52 players, pWAR is not your answer as you have suggested many times. You will find no support in accepting your list of the 52 greatest players...not even the author of WAR would grant you that.

If you want to use WAR to compare players of the same season, it does a "fair" job of doing that, provided you don't try to compare a catcher to a pitcher, or any other position player for that fact.

I see real value in both WAR and pWAR..but it is not ideal. I am certain that with enough analysis and intelligent critics, a newer version of WAR will be provided that will better address what can currently be easily seen as a flaw.

A 5 year pWAR will never be accepted in its current state to suggest greatness...it's a good step in the right direction, but without additional data, it is nonsense.

 

[tribefan26]

The Statistics are weak at best due to the lack of statistical confidence. I have brought this up previously and it has been side stepped. The fact is when you select a portion of the data set and not the entire range you lower your confidence levels to to flaws in arbitrary selection. You can try to strong arm your discussions and explain your "idea" but WAR is just not replicable in a theoretical manner. Please write the Proof/Axiom if you think I am wrong.

The lack of statistical confidence is the same as it is for any baseball statistic. Every year we award a batting title in each league - rarely has the person awarded this crown been better than no. 2 by a statistically significant margin.

We play whole seasons to determine slots in playoffs were random variation often produces winners from lesser teams. And we celebrate the championships as if they are statisitcally significance.

Very little in sports in statistically significant. Thats why we enjoy them so much.

 

[hofautos]

^^^ which defends the thought that "greatness is subjective" and cannot be measured by flawed statistics....but i think we are a whole lot closer.

 

[Guest]

^^^ depends on the objective.
If you want to create a list of the "greatest" 52 players, pWAR is not your answer as you have suggested many times. You will find no support in accepting your list of the 52 greatest players...not even the author of WAR would grant you that.

If you want to use WAR to compare players of the same season, it does a "fair" job of doing that, provided you don't try to compare a catcher to a pitcher, or any other position player for that fact.

I see real value in both WAR and pWAR..but it is not ideal. I am certain that with enough analysis and intelligent critics, a newer version of WAR will be provided that will better address what can currently be easily seen as a flaw.

A 5 year pWAR will never be accepted in its current state to suggest greatness...it's a good step in the right direction, but without additional data, it is nonsense.

I think it is more likely that your pre-conceived notions of Catchers is flawed, rather than WAR is flawed in regard to Catchers.

Here is the leap you seem unwilling to make.

Take away every MVP award a player has ever won.

Now award an MVP to every player for each season that they had a WAR of 8.0+.

If a player won 5 MVP awards, would you conside them great?

I do.

 

[hofautos]

Accurately interpreted as the 5 best seasons of WAR, nothing more. It is valuable but not in of itself.
Modify it with career numbers and it could potentially have better usage, especially if WAR's deficiency in comparing catchers to other positions is updated.

How many of the 50 should be catchers - do we figure out a priori how many of each position need to be in the top 50? If we measure only statistical factors it almost a certainity that some positions will be better represented in any list of 50.

I have a problem with a system that that says no catchers belong in the HoF. I don't have a problem with a statistical measure that says no catcher is in a top 50. I have a real problem if someone decides that a statistic solely decides who gets in.

Forget the HOF, lets just talk "Greatness".
It's not rocket science that there is a flaw in ANY system where 10 pitchers (who pitch 1 in 4/5 games) make top 50, but not one catcher does.
How can anyone say pitcher #10 is greater than catcher #1?
The catcher handles the ball 300% more than pitchers.

Or even compare to #10 Left fielder...is the worst left fielder on the list "GREATER" than the best catcher?

The only one that can state that is the author of WAR who uses fudge to determine his calculations.

Come on this is not hard people.

 

[hofautos]

Forget anything else...forget how many runs, how many hits, how man wins....just forget it all for 10 minutes.

Is the catcher one of the more important positions or is it like Chevy suggests:
CHEVY>> Historically catchers have just been a bunch of fat guys that squat there and throw the ball to the pitcher.
That's what a catcher is


If you believe that, then you are not worth debating with anymore.

The catcher is one of the most important positions, and if you don't agree, perhaps you can agree that it is at least more important than LF.
But do your WAR comparisons between LF and CATCHERS.

Sure Chevy says they subtract x points for LF, and add 99 points for catchers, and that the math they used is scientific.

Come on...don't make me laugh so hard.

Chevy also told me he is a salesman in his real job...so calculate that in the equation.
Sorry, I am not buying.

 

[pigskincardboard]

The Statistics are weak at best due to the lack of statistical confidence. I have brought this up previously and it has been side stepped. The fact is when you select a portion of the data set and not the entire range you lower your confidence levels to to flaws in arbitrary selection. You can try to strong arm your discussions and explain your "idea" but WAR is just not replicable in a theoretical manner. Please write the Proof/Axiom if you think I am wrong.

The lack of statistical confidence is the same as it is for any baseball statistic. Every year we award a batting title in each league - rarely has the person awarded this crown been better than no. 2 by a statistically significant margin.

We play whole seasons to determine slots in playoffs were random variation often produces winners from lesser teams. And we celebrate the championships as if they are statisitcally significance.

Very little in sports in statistically significant. Thats why we enjoy them so much.

I've always found this interesting, especially the batting statistic portion. Even at 550-700AB(iirc), you're still dealing with +-3 points 95% of the time. I think most batting titles are decided by more than a handful of points, but I see your point. It really depends on what you're awarding. If you're saying that a hit is a hit, regardless of luck involved or whether or not it'll be replicated in the future, you're good to go. Quite frankly, I'm okay with that. People win golf championships based on piss luck; and we're not really deciding who *should've* won the tournament.

It really comes down to how you want to evaluate the results.

If one team has a 53% chance of winning and the other 47%, I'm okay saying the 53% is a better team. Yah, it's outside of our margin of error and it probably doesn't make a difference, but who cares?

On the other hand, I get upset when the winning team proclaims that they are the best team. It's funny to look at the statistics and realize that some of the biggest upsets in baseball history really aren't that statistically improbable.

Anyways, I want this thread to die. I cannot believe that "The Best Five Seasons Per Player By WAR" is even a thing.

 

[Guest]

Forget anything else...forget how many runs, how many hits, how man wins....just forget it all for 10 minutes.

Is the catcher one of the more important positions or is it like Chevy suggests:
CHEVY>> Historically catchers have just been a bunch of fat guys that squat there and throw the ball to the pitcher.
That's what a catcher is


If you believe that, then you are not worth debating with anymore.

The catcher is one of the most important positions, and if you don't agree, perhaps you can agree that it is at least more important than LF.
But do your WAR comparisons between LF and CATCHERS.

Sure Chevy says they subtract x points for LF, and add 99 points for catchers, and that the math they used is scientific.

Come on...don't make me laugh so hard.

Chevy also told me he is a salesman in his real job...so calculate that in the equation.
Sorry, I am not buying.

Rpos is a component of RAR that I had no part in creating.

It is designed to adjust the positions in terms of RAR, which plays a role in the formation of WAR.

99 'points' are not added for Catchers. Johnny Bench' career Rpos is 99, meaning over the course of his career he received 99 bonus runs because he was a catcher.

This number was not chosen by me. It was not 'chosen' by anyone. It was determined by statistics. Statistics I played no role in developing.

 

[Guest]

The Statistics are weak at best due to the lack of statistical confidence. I have brought this up previously and it has been side stepped. The fact is when you select a portion of the data set and not the entire range you lower your confidence levels to to flaws in arbitrary selection. You can try to strong arm your discussions and explain your "idea" but WAR is just not replicable in a theoretical manner. Please write the Proof/Axiom if you think I am wrong.

The lack of statistical confidence is the same as it is for any baseball statistic. Every year we award a batting title in each league - rarely has the person awarded this crown been better than no. 2 by a statistically significant margin.

We play whole seasons to determine slots in playoffs were random variation often produces winners from lesser teams. And we celebrate the championships as if they are statisitcally significance.

Very little in sports in statistically significant. Thats why we enjoy them so much.

I've always found this interesting, especially the batting statistic portion. Even at 550-700AB(iirc), you're still dealing with +-3 points 95% of the time. I think most batting titles are decided by more than a handful of points, but I see your point. It really depends on what you're awarding. If you're saying that a hit is a hit, regardless of luck involved or whether or not it'll be replicated in the future, you're good to go. Quite frankly, I'm okay with that. People win golf championships based on piss luck; and we're not really deciding who *should've* won the tournament.

It really comes down to how you want to evaluate the results.

If one team has a 53% chance of winning and the other 47%, I'm okay saying the 53% is a better team. Yah, it's outside of our margin of error and it probably doesn't make a difference, but who cares?

On the other hand, I get upset when the winning team proclaims that they are the best team. It's funny to look at the statistics and realize that some of the biggest upsets in baseball history really aren't that statistically improbable.

Anyways, I want this thread to die. I cannot believe that "The Best Five Seasons Per Player By WAR" is even a thing.

If you don't think it's worth investigating how two playes compared to one another using a smaller sample size, then how do you propose I compare two players with a large gap between their career lengths?

 

[hofautos]

I cannot believe that "The Best Five Seasons Per Player By WAR" is even a thing.

I actually give it real value.
It should carry a huge weight in defining greatness....it's just missing a few spices.
I really look forward to getting those numbers.
As much as I disagrree with Chevy in his shallowness, the work he has done, i really like.
Hopefully he will lighten up and put it to better use, or give it to me, and I will.

 

[hofautos]

Forget anything else...forget how many runs, how many hits, how man wins....just forget it all for 10 minutes.

Is the catcher one of the more important positions or is it like Chevy suggests:
CHEVY>> Historically catchers have just been a bunch of fat guys that squat there and throw the ball to the pitcher.
That's what a catcher is


If you believe that, then you are not worth debating with anymore.

The catcher is one of the most important positions, and if you don't agree, perhaps you can agree that it is at least more important than LF.
But do your WAR comparisons between LF and CATCHERS.

Sure Chevy says they subtract x points for LF, and add 99 points for catchers, and that the math they used is scientific.

Come on...don't make me laugh so hard.

Chevy also told me he is a salesman in his real job...so calculate that in the equation.
Sorry, I am not buying.

Rpos is a component of RAR that I had no part in creating.

It is designed to adjust the positions in terms of RAR, which plays a role in the formation of WAR.

99 'points' are not added for Catchers. Johnny Bench' career Rpos is 99, meaning over the course of his career he received 99 bonus runs because he was a catcher.

This number was not chosen by me. It was not 'chosen' by anyone. It was determined by statistics. Statistics I played no role in developing.

99 points, runs, whatever .....
there was some variable awarded because he was a catcher. There was a formula "created" that derived that award, which no one can suggest is "scientific".