Meph Has a New Article Up

[CaliforniaRaisin]

It's his third, you can find it on the front page.

[Derwood]

Interesting read, but it reminds me of why I wasn't a math major.

[CubinNY]

No offense to Meph, but it's one of the more ridiculous things he's written. There is no normal distribution in baseball. Talent is not equally dispersed across the league.

 

He's still has never defined what "true talent" means.

 

In 162 game season the best team finishes first.

[Transmogrified Tiger]

No offense to Meph, but it's one of the more ridiculous things he's written. There is no normal distribution in baseball. Talent is not equally dispersed across the league.

 

He's still has never defined what "true talent" means.

 

In 162 game season the best team finishes first.

 

Changing the likelihood that you win the games doesn't really change the standard deviation. A team with a 105 win talent level will still have a standard deviation of 6 games.

[CubinNY]

No offense to Meph, but it's one of the more ridiculous things he's written. There is no normal distribution in baseball. Talent is not equally dispersed across the league.

 

He's still has never defined what "true talent" means.

 

In 162 game season the best team finishes first.

 

Changing the likelihood that you win the games doesn't really change the standard deviation. A team with a 105 win talent level will still have a standard deviation of 6 games.

The problem isn't the math, it is the assumption that math is based on.

 

The odds of winning or losing a game are not 50/50 unless "talent" is equally distributed across the two teams. That almost never is the case.

 

It's kind of like in vegas. The roulette wheel is set up so that no matter what happens, in the long run the house wins.

[Transmogrified Tiger]

No offense to Meph, but it's one of the more ridiculous things he's written. There is no normal distribution in baseball. Talent is not equally dispersed across the league.

 

He's still has never defined what "true talent" means.

 

In 162 game season the best team finishes first.

 

Changing the likelihood that you win the games doesn't really change the standard deviation. A team with a 105 win talent level will still have a standard deviation of 6 games.

The problem isn't the math, it is the assumption that math is based on.

 

The odds of winning or losing a game are not 50/50 unless "talent" is equally distributed across the two teams. That almost never is the case.

 

It's kind of like in vegas. The roulette wheel is set up so that no matter what happens, in the long run the house wins.

 

Yes, that's what I was just saying. Every team still has that same standard deviation regardless of their talent level. I don't see how that invalidates his conclusion.

[CubinNY]

No offense to Meph, but it's one of the more ridiculous things he's written. There is no normal distribution in baseball. Talent is not equally dispersed across the league.

 

He's still has never defined what "true talent" means.

 

In 162 game season the best team finishes first.

 

Changing the likelihood that you win the games doesn't really change the standard deviation. A team with a 105 win talent level will still have a standard deviation of 6 games.

The problem isn't the math, it is the assumption that math is based on.

 

The odds of winning or losing a game are not 50/50 unless "talent" is equally distributed across the two teams. That almost never is the case.

 

It's kind of like in vegas. The roulette wheel is set up so that no matter what happens, in the long run the house wins.

 

Yes, that's what I was just saying. Every team still has that same standard deviation regardless of their talent level. I don't see how that invalidates his conclusion.

A binomial distribution assumes that the there is an equal likelihood of a yes/no outcome. In most cases in sports there is not. Therefore, the SD is correct only in theory.

 

It's a nice thought experiment though and shows how much chance there when two teams are equal.

[Mephistopheles]

No offense to Meph, but it's one of the more ridiculous things he's written. There is no normal distribution in baseball. Talent is not equally dispersed across the league.

 

He's still has never defined what "true talent" means.

 

In 162 game season the best team finishes first.

 

Changing the likelihood that you win the games doesn't really change the standard deviation. A team with a 105 win talent level will still have a standard deviation of 6 games.

The problem isn't the math, it is the assumption that math is based on.

 

The odds of winning or losing a game are not 50/50 unless "talent" is equally distributed across the two teams. That almost never is the case.

 

i already addressed this in the article. even if the split is .650/.350 the standard deviation is still over six.

[Mephistopheles]

No offense to Meph, but it's one of the more ridiculous things he's written. There is no normal distribution in baseball. Talent is not equally dispersed across the league.

 

He's still has never defined what "true talent" means.

 

In 162 game season the best team finishes first.

 

Changing the likelihood that you win the games doesn't really change the standard deviation. A team with a 105 win talent level will still have a standard deviation of 6 games.

The problem isn't the math, it is the assumption that math is based on.

 

The odds of winning or losing a game are not 50/50 unless "talent" is equally distributed across the two teams. That almost never is the case.

 

It's kind of like in vegas. The roulette wheel is set up so that no matter what happens, in the long run the house wins.

 

Yes, that's what I was just saying. Every team still has that same standard deviation regardless of their talent level. I don't see how that invalidates his conclusion.

A binomial distribution assumes that the there is an equal likelihood of a yes/no outcome. In most cases in sports there is not. Therefore, the SD is correct only in theory.

 

It's a nice thought experiment though and shows how much chance there when two teams are equal.

 

actually you can mathematically derive all of this if you go game by game if you want to, but in the end it's going to average right around this, but the accuracy gained is inconsequential compared tot the amount of work added. since we're dealing with an extremely large sample, we can do this. a binomial assumes there is an equal chance of yes/no each time. it doesnt assume that it's 50/50.

[jersey cubs fan]

I think it reads like an extended intro into a bigger article.

[CubinNY]

Meph,

 

I think you give far, far too much value to chance. Competitive sports like baseball is not the same thing as rolling a six sided dice and looking at variance and probability. There are far too many variables in play to chalk victory up to luck. It's my one big gripe with the saber community. They have a weird fetish to want to chalk up any unexplained variance to luck. It's really a piss poor way to do behavioral analysis.

 

A manger making a poor decision that cost his team 1 game in April can have a big effect in September if his team loses the WC by one game. That's not bad luck. We could go round and round on this one but I think luck is a default position, one only used when all other possibilities have been exhausted.