anyone old enough to remember the 8 sided stratomatic dice?

Posted by milwood on 2/15/2014 4:18:00 PM (view original):
Maybe he'd prefer it's a roll of the dice?

Hey definitely prefers whining and reporting people...so he might be able to better understand an analogy that uses those as examples???

Posted by gillispie1 on 2/15/2014 4:14:00 PM (view original):

Posted by ettaexpress on 2/15/2014 3:18:00 PM (view original):

Posted by gillispie1 on 2/15/2014 2:26:00 PM (view original):

Posted by ettaexpress on 2/15/2014 1:00:00 PM (view original):
LOL this is help? Wow. I think about two posts on this thread so far would fit that description for me.

50 in a row is easy math with the google. So how does one define "huge preponderance" then? 48 out of 50? 45 out of 50? Surely has to be at least 45 right? At what number out of 50 do you think the probability increases to even 1%? Bottom line is he made a bad analogy. It also wasn't helpful in any way toward furthering knowledge of the game, only making excuses for the sim.

bad analogy? the sim is, in simplistic form, a series of 1000 or so weighted coin flips. most likely, less than 1000. its not even really an analogy, hes just straight up describing something for what it is. not exactly right on some of the parameters, but as far as things go, its pretty darn close to the real thing.

if you have any interest in actual statistical discussions, you don't just pick "50 in a row" - that is a complete joke. its still pretty simple math to pick something reasonable. you will never find anything remotely like 50 in a row in any simulation output here in HD - i challenge you to try - that would effectively be like "team A scores! team A steals the ball. team A scores! team B throws the ball out of bounds. team A scores!" and on and on in the PBP. you talk about bad analogies, and then come out with this 50 in a row nonsense? come on...

The coin flip analogy just doesn't work. This is shown by you making the qualifier "weighted coin flip" as if that's really a thing. All coin flip scenarios assume a fair coin, just like dice throw scenarios assume the dice aren't loaded. Not blaming you, you didn't make the flawed analogy, but just saying.

I agree with you that you'd never find anything remotely like 50 in a row (well technically you could, but the chances are practically 0) ...that was my point. If you prefer another definition of "huge preponderance"...it's still going to be the same story. My guess is that you don't get to a reasonable probability until the definition is weakened to less than 40 of 50. So yes, I'm talking about a bad analogy. It wasn't my analogy to start with. 

as if thats really a thing? seriously? weighted coin flips are discussed all the time, even in the most elementary statistics classes. maybe you assume a fair coin, fair and weighted coins are both used all the time in very basic statistical analysis.

ill take fd's approach... i am reminded of a time when prior (and more interesting) version of a certain person claimed that because there were two outcomes - the sun could come up tomorrow - or it could not - that there must be a 50% chance of either. its sad we are back there :(

let's review

fd - "less relevant to ability to find in a play by play a sequence in which there is a very bad 3 or 5 minute stretch - if one tossed a coin 1000 times one would almost surely be able to identify a shocking 50 toss segment of that sequence"

etta - "So you're implying that my team has had an equal chance to win those games (a coin toss). OK, so why hasn't my team put on such runs to win more games? Just unlucky? "

how you went from that to it must be 50/50 is beyond me. even if they are fair coin flips, the chance you would see something more reasonable, like 40/50, is way less than the stuff you are posting about being outside of the statistical norm. i'm not sure why you are even arguing this point, the bottom line is a 5 minute bad stretch is perfectly reasonable and it happens all the time. as a basketball fan, you must know that. why get into statistical arguments if thats not your thing, AND its not even necessary?

Just because some eggheaded stats folk'll talk about it doesn't mean it's a thing.

Maybe the mob should contract some statisticians to make some of these weighted coins for unfair flipping.  

Posted by gillispie1 on 2/15/2014 4:17:00 PM (view original):
for what its worth, coin flip analogies have been used to represent the sim here hundreds of times, and you are the first to really object to that model. maybe its because you are unfamiliar with weighted coin flips or whatever. the point is its a series of random evens with some probability, and in that sense, its a great analogy. its why so many people go to it to describe the simulation.

Maybe I'm wrong then. Maybe the sim just is that bad.

It would be hilarious to watch someone watch a real game with that kind of outlook. 

Would it be as hilarious as watching you babble ridiculously on and on???

Posted by ettaexpress on 2/15/2014 4:53:00 PM (view original):

Posted by gillispie1 on 2/15/2014 4:17:00 PM (view original):
for what its worth, coin flip analogies have been used to represent the sim here hundreds of times, and you are the first to really object to that model. maybe its because you are unfamiliar with weighted coin flips or whatever. the point is its a series of random evens with some probability, and in that sense, its a great analogy. its why so many people go to it to describe the simulation.

Maybe I'm wrong then. Maybe the sim just is that bad.

It would be hilarious to watch someone watch a real game with that kind of outlook. 

well then, why don't you suggest a simulation model not predicated on a series of random number generations to decide outcomes? if its so bad, should be no sweat for one as intelligent as yourself to suggest a better model, shouldn't it?

its not just "eggheaded statistics folks" who are aware of weighted coin models in probability. even an idiot who barely passed statistics 1 should be aware. there are high school drop outs to whom weighted coins are not some nebulous unknown concept... sounds like you consider those guys too smart to be relevant in discussions by the "normal people", huh?

http://www.stat.columbia.edu/~gelman/research/published/diceRev2.pdf

Is Andrew Gelman a good enough source for you?

Not. A. Thing.

I believe that's what I said that you had such a problem with, is it not? 

Why start an argument like that if you don't know the fact, especially when you don't HAVE to? 

pwn3d

Just stop talking. You've been exposed.

Posted by ettaexpress on 2/15/2014 7:32:00 PM (view original):
Just stop talking. You've been exposed.

I never thought I'd see the day were even the word "irony" just simply fell short.

... You do realize that we aren't talking actual physical coins, Etta, right? I mean, metaphors and models may be complicated things, but I do think you are smart enough to handle them ...


This is a virtual coin Coin is used as a standby for a random event, a fifty-fifty event. A model of a coin that is "biased" as a simulated coin which no longer exhibits fifty fifty probability. Whether said coin actually exists in the physical world is utterly irrelevant. Talk about completely missing the point.


The article itself even says at the end that "this study is certainly not important for probability theory".

So ...

And I said it's not actually a thing, which it isn't...

So...

Posted by ettaexpress on 2/15/2014 7:31:00 PM (view original):
http://www.stat.columbia.edu/~gelman/research/published/diceRev2.pdf

Is Andrew Gelman a good enough source for you?

Not. A. Thing.

I believe that's what I said that you had such a problem with, is it not? 

Why start an argument like that if you don't know the fact, especially when you don't HAVE to? 

pwn3d

WOW, i didn't think you could surprise me with a post, with all the prior comments you've made. but you managed it, this one takes the cake - your finest work. really proved me wrong, lol, nicely done!

colonels, we were wrong about you - apparently, you are a fricken genius. i guess its true, you don't know what you have until you've lost it :(

So the cited source says that weighted coin doesnt effect toss results in real life but remains an excellent way to learn probability lessons as a hypo

This would seem to be what we have seen in this thread.  One lesson learned is that with a sufficient number of die rolls or coin tosses, one expects to see one or more segments in which the results do not conform to the expected overall probability distribution.  And, we learn that attributing causation to those coincidental events is a mistake.  Surely, we now recognize, the SIM no more "decides" that a team should have an awful five minutes than does the coin decide that amongst 1000 tosses it should over some set of say 50 happen to be heads 40 times - no more than a die decides or is caused to roll a six 8 of 10 times at some point in 100 rolls.

now there may be real phenomena going on - bad mix of players - tiredness - adjustment - other -  but the mere fact that one can find a bad period tells one very little. 

Posted by fd343ny on 2/15/2014 9:13:00 PM (view original):
So the cited source says that weighted coin doesnt effect toss results in real life but remains an excellent way to learn probability lessons as a hypo

This would seem to be what we have seen in this thread.  One lesson learned is that with a sufficient number of die rolls or coin tosses, one expects to see one or more segments in which the results do not conform to the expected overall probability distribution.  And, we learn that attributing causation to those coincidental events is a mistake.  Surely, we now recognize, the SIM no more "decides" that a team should have an awful five minutes than does the coin decide that amongst 1000 tosses it should over some set of say 50 happen to be heads 40 times - no more than a die decides or is caused to roll a six 8 of 10 times at some point in 100 rolls.

now there may be real phenomena going on - bad mix of players - tiredness - adjustment - other -  but the mere fact that one can find a bad period tells one very little. 

No, no. The statistical test isn't valid unless you physically flip every single coin yourself. Roll every single dice by hand. All by himself Etta has overturned 215 years worth of statistics.