PDA

View Full Version : Winners and Losers

Monty Cantsin
12-16-2004, 11:28 AM
In the theory forum someone started this thread (http://forumserver.twoplustwo.com/showflat.php?Cat=&amp;Board=genpok&amp;Number=1390393&amp;page =0&amp;view=expanded&amp;sb=5&amp;o=14&amp;fpart=all) about what percentage of the player population are long term winners.

Specifically, people are wondering why everyone's database breaks down around 60/40 losers/winners and what the significance of this number is.

I don't know enough to say but I have a feeling that a lot posts in this thread demonstrate a fundamental confusion about some basic principles of statistics.

1. Is there a definitive discussion of these issues somewhere that could be linked to?

2. Do any of you statistics types want to come by and drop a little science on us?

Thanks,

/mc

gaming_mouse
12-16-2004, 03:04 PM
Monty,

That is fascinating -- I would have thought that the % of winners was much lower too.

Unfortunately, I don't see any clever statisitcal reason why this should be true. Maybe it's the way it is, though I'm curious to hear other theories.

gm

Monty Cantsin
12-16-2004, 05:42 PM
It seems that everyone's db showing approx 60/40 is consistent with a true long term percentage of approx. 90/10 because of bell curves and, um, stuff.

JTR did some models along these lines in that thread.

As I said, I'm totally unqualified to address this issue but I find it really interesting.

/mc

Lost Wages
12-16-2004, 05:45 PM
I think it has to do with the way the sample is taken.

You only observe a table for a short time. During that time the losing players have almost as good a chance of winning as the winning players. Then the session is over and the next time you play you join a new table with mostly unknown players and repeat the process.

The result is that your database contains a large number of players but a small number of hands per player on average. I suspect that if you could track 1000 random players for 50,000 hands each that the winner/loser ratio would be much different.

Maybe someone smarter than me can cast this in a more mathematical light.

Lost Wages

gaming_mouse
12-16-2004, 05:54 PM
[ QUOTE ]
It seems that everyone's db showing approx 60/40 is consistent with a true long term percentage of approx. 90/10 because of bell curves and, um, stuff

[/ QUOTE ]

Where is this post?

This doesn't seem right to me. Given the large sample size of the cumulative combined databases, the 60/40 ratio should be accurate.

LostWages, I don't think your reasoning is correct here, either. I'm not positive, though.

I'll PM BruzeZ -- maybe he'll chime in.

gm

Monty Cantsin
12-16-2004, 06:01 PM
[ QUOTE ]
[ QUOTE ]
It seems that everyone's db showing approx 60/40 is consistent with a true long term percentage of approx. 90/10 because of bell curves and, um, stuff

[/ QUOTE ]

Where is this post?

[/ QUOTE ]

here (http://forumserver.twoplustwo.com/showthreaded.php?Cat=&amp;Number=1398648&amp;page=0&amp;view=e xpanded&amp;sb=5&amp;o=14&amp;vc=1) and the ones afterwards.

/mc

gaming_mouse
12-16-2004, 06:37 PM
Thanks for that.

I am now convinced my previous comments were wrong.

gm

randomchamp
12-16-2004, 09:19 PM
In my PT if I sort by player I've played with more than 1,000 hands I've got 72% winners. Bad game selection huh? Nope it's the other % of players that come an go and donate.