Analysis from data mining

[GooperMC]

I don't know if you guys are interested in this but I did the work so I though I would share it.

http://www.csc.calpoly.edu/~sjaspan/poker/O8_Summary_Limit_All_50KHands.

These number look a lot like the numbers at 25K hands however the sample is larger so the variance should be a little lower.

[mshalen]

The link doesn't work.

[TheShootah]

Make the end .html and it should work. /images/graemlins/grin.gif

[GooperMC]

TheShootah is exactly right. Here is a working link:

http://www.csc.calpoly.edu/~sjaspan/poker/O8_Summary_Limit_All_50KHands.html

[Phat Mack]

Many thanks for compiling and publishing this. Extremely interesting.

[Buzz]

Gooper - What do you mean when you write that the big blind range is

[ QUOTE ]
-&infin, -3

[/ QUOTE ]?

Similarly, what do
[ QUOTE ]
-3, 0

[/ QUOTE ]
[ QUOTE ]
0, 3

[/ QUOTE ], and
[ QUOTE ]
3, &infin

[/ QUOTE ] mean?

Buzz

[GooperMC]

BBRng is the BB/100 Range. For example [-3, 0) means these stats are for all players that have a BB/100 >= -3 and < 0.

∞ is infinity and most browsers should display the infinity symbol. What browser are you using?

[Phat Mack]

BBRng is the BB/100 Range. For example [-3, 0) means these stats are for all players that have a BB/100 >= -3 and < 0.


Not sure what Buzz was asking, so I'll add this (please correct me if I'm wrong):

The BB/100 is the amount won in 100 hands. The table at the bottom of the page says "Bets," so I'm not sure if it is Big Blinds or Big Bets per hundred. I assumed Big Bets.

[GooperMC]

[ QUOTE ]
I assumed Big Bets.

[/ QUOTE ]
You are correct (and I just changed it on the web page). Wow, I guess I made this as confusing as possible. Sorry guys.

[handsome]

Nice. Sample size too small but interesting data. I wouldn't count it significant until you have 500+ players with over 1k hands logged.

[Buzz]

Gooper - Thanks. I think I understand. I very much appreciate your sharing your data. I'll look at it and try to digest it. I don't play on-line and I play at slightly higher limits, but I'm very interested in this sort of data.

[ QUOTE ]
∞ is infinity and most browsers should display the infinity symbol.

[/ QUOTE ]

I don't see any infinity symbol in the data you have posted, but "&infin" is good enough for me, now that I know what it means.

[ QUOTE ]
What browser are you using?

[/ QUOTE ]

I have aol. If I'm using a browser, it's one that comes with aol.

[ QUOTE ]
Wow, I guess I made this as confusing as possible. Sorry guys.

[/ QUOTE ]

Not at all. I should have been able to figure it out on my own. Thanks again for sharing your data and explaining it. Very much appreciated.

Buzz

[Ironman]

While I agree that the sample size is still relatively small, there is one statistic that jumped out at me that surprised me...it probably shouldn't have, but it did.

It doesn't matter if you look at 100 hands played or 600 hands played, the percent of players who loose money at this game is roughly 50%.

At 100 hands played it was 54%, but even the people who play alot...enough to register 600 hands played on this data mining experiment...still were bad enough to loose 47% of the time.

And that group of 11 people in the 600 category with a VPIP of 45....man, I want to be at that table. :-)

Very interesting. Thanks.

[GooperMC]

I am planning on doing this on fairly regular intervals (75K hands, 100K hands, 150K hands, ...).

Is there any other pieces of data that people want to see? Anything that you can export I can add (% of time check raise, % of time defend blinds, ...).

[AnyAce]

Looks great, Gooper.
I'd be interested to see the results for 3/6 and 5/10 though I understand if you don't have time to do those as well.

Thanks for your hard work on this.
AA

[GooperMC]

It is less a time and problem and more a data problem. With the help of some excel spreadsheets / html templates I can produce those tables pretty quickly, however I don't have any hands data mined for those levels.

Sorry.

[Beavis68]

I would like to see it filtered by VPIP, and WtSD, and see if that affects the bb/100, basically, everyone here looks like a calling stationg, all the VPIPs are over 30% and all the WtSDs are around 40%.

Also, rake paid would be interesting.

[Phat Mack]

Is there any other pieces of data that people want to see? Anything that you can export I can add (% of time check raise, % of time defend blinds, ...).


If available, I'd like to see standard deviations for the groups of BB/100's.

[TGoldman]

[ QUOTE ]
Is there any other pieces of data that people want to see? Anything that you can export I can add (% of time check raise, % of time defend blinds, ...).

[/ QUOTE ]
I would be very interested in seeing the 'When Folds' stat for each street. I have a hunch that this % is very indicative of a successful O8 player.

[GooperMC]

When I get some time this weekend I will create stats for
- Filtering by VPIP, and WtSD
- standard deviations for BB/100's.
- 'When Folds' stat for each street.

Right now I have this mostly automated and I am going to try and make it fully automated so new creating new analysis is easy.