An absurd statistical example

[DavidC]

Hey guys.

Over my last 1000 hands, I lost 4000 bb.

That's right, 4000 bb.

This is just a horrible downswing for an otherwise good player, right? Or do I suck badly enough to justify this beating?

--Dave.

(Edit: Obviously I'm not serious about having this swing, but please answer me like I'm some 5 post newb telling you about his legitimate downswing... he did indeed lose 4000 bb in 1000 hands... how likely is it that this is a result of pure variance and that he's an otherwise break-even player or a winner?)

Locked at Dave's request

[baronzeus]

There is a 0 % chance you are a winner.

[hemstock]

No limit?

[AdamL]

Interesting reformatting of YP. /images/graemlins/smile.gif

[GrunchCan]

Even if it were no-limit, a loss this large can only be due to bad play.

[TripleH68]

You could not lose 400bb/100 if you played every hand and bet every hand at every opportunity. Noob.

Now go ask mommy if it is alright if you play poker for real money.

/images/graemlins/wink.gif

[gharp]

Whoa - 4BB per hand? That's kind of a lot. You'd pretty much have to see every single flop and then call to every river, I'm thinking.

[MrWookie47]

Well, capping every street, every hand and losing will have you lose at 12 BB/hand, or -1200 BB/100. This is a mere fraction of how bad it could possibly be. I'd say it's just variance, but if you really don't trust yourself, give the ol' SSH a reread and head back into the fray. At my table.

[DavidC]

Alright,

I'm a super-nit, so I just have to get involved with this thread... uh.. oh yeah, I started it, didn't I? /images/graemlins/smile.gif

Anyways, I'd like to start by saying, yes, it's possible to lose those 400bb/100 hands. I mean, you could theoretically lose a maximum of 1200bb in 100 hands, right?

And I'd like to finish by saying that everyone is wrong. /images/graemlins/smile.gif

Yes, there's a chance that Entity/Grunch/Shill could hit -4000 in 1000 hands, but you guys are right that it's infinitely small, like probably worse than a 1/1,000,000 chance.

So you guys are all wrong, but only by a really small amount. /images/graemlins/smile.gif

(Edit: Wookie got in before this post, and he's right about it being within technical possibility.)

[Redd]

I think the odds are WAY worse than 1:1,000,000. I'd guess its somewhere like 1:10^100 or so; would someone who likes stats be able to give us a confidence interval that this guy is a winning player, assuming a normal SD?

[baronzeus]

1/1000000 is close enough to 0% that you still suck /images/graemlins/tongue.gif

[DavidC]

[ QUOTE ]
I think the odds are WAY worse than 1:1,000,000. I'd guess its somewhere like 1:10^100 or so; would someone who likes stats be able to give us a confidence interval that this guy is a winning player, assuming a normal SD?

[/ QUOTE ]

Yeah, I tried to figure that out, and the method is here (http://forumserver.twoplustwo.com/showflat.php?Cat=&Board=inet&Number=1342415&fpart= &PHPSESSID=)

[aces_dad]

I think you're just at the wrong end of the varience curve, and you're 'due' to win it all back. Soon.

And please come play at my table so I can, err, watch you do it.

[gharp]

[ QUOTE ]
Anyways, I'd like to start by saying, yes, it's possible to lose those 400bb/100 hands. I mean, you could theoretically lose a maximum of 1200bb in 100 hands, right?


[/ QUOTE ]
Theoretically, you could -- but in order to do it with your example you're talking about someone getting dealt a second-best-or-worst hand, every single time, over 1000 hands. PLUS, the other players cooperate by capping every street the whole time.

For the 400BB/100 example you'd need some of this too. I feel like you'd have to play pretty much any 2 cards and call down all the time. You might have to fold some winners in there too.


I think it's far more likely that there's a glitch/math error in the counting of hands or BB's lost. /images/graemlins/smile.gif