#1
|
|||
|
|||
Data mining Round 2
With TGoldmans help I just broke 100K hands of data mining. Here is the analyis.
Sorted by BB/100: http://www.csc.calpoly.edu/~sjaspan/...nds_BB100.html Variance on BB/100 sorting: http://www.csc.calpoly.edu/~sjaspan/..._Variance.html Sorted by VP$IP: http://www.csc.calpoly.edu/~sjaspan/...ands_VPIP.html Sorted by WtSD: http://www.csc.calpoly.edu/~sjaspan/...ands_WtSD.html Sorted by W$SD: http://www.csc.calpoly.edu/~sjaspan/...Hands_WSD.html Sorted by W$SF: http://www.csc.calpoly.edu/~sjaspan/...Hands_WSF.html Enjoy. |
#2
|
|||
|
|||
Re: Data mining Round 2
Can you explain what all these numbers mean to the lay people? I'm really interested, but don't have any idea where to start with that chart.
|
#3
|
|||
|
|||
Re: Data mining Round 2
Once again, Thanks for your efforts!
I need some help reading the varience chart. Looking at first lines: Looking at the first chart, "Sorted by BB/100," first line (players w/ > 1000 hands, BB/100 < -3), There are 6 players, 1199.00 hands ( I assume this is the average), and BB/100 = -6.88. Then on the second chart, "Variance on BB/100 Sorting," I'd expect the numbers on the first line to be the same. There are still 6 players, but the hands = 266656.20, and the BB/100 = (+?)7.96. So I guess I'm not sure what's being measured here. Thanks for any help. Mack |
#4
|
|||
|
|||
Re: Data mining Round 2
The variance tables are the variances for the BB/100 grouping (I know that doesn't help but bear with me).
Hypothetical example: in the players with more 1000 hands and a BB/100 in [-3, 0) there were 6 players. Lets say their VP$IPs were 20, 22, 30, 28, 40, 60. The BB/100 grouping would show the average (20 + 22 + 30 + 40 + 60 + 32) / 6 and the Variance table would show the variance of those players VP$IP. In this case the variance would be: ((20˛ + 22˛ + 30˛ + 40˛ + 60˛ + 32˛) – (20 + 22 + 30 + 40 + 60 + 32) ˛ / 5) / 6 Hope that helps. |
#5
|
|||
|
|||
Re: Data mining Round 2
Well all of these number are gathered from Poker Tracker Omaha so if use PTO then you should have a good start. If not at the bottom of the pages there is a table that gives an abrupt explanation of each column.
Beyond that is it just like a normal table. Here is an example: If you want to know the pre flop raising of player who played more then 1000 hands and who has a BB/100 between 0 and 3 do the following. 1) Goto http://www.csc.calpoly.edu/~sjaspan/...nds_BB100.html 2) Find the table with more then 1000 hands (the first table) 3) Find the row for players with a BB/100 in [0, 3) (the 3rd row) 4) Find the column for PFR column (the 6th column) 5) Read the number If you have any specific questions, or if this doesn’t make sense just ask. |
#6
|
|||
|
|||
Re: Data mining Round 2
Good data.
What conclusions do you draw from it? -g |
#7
|
|||
|
|||
Re: Data mining Round 2
[ QUOTE ]
Hope that helps. [/ QUOTE ] It does, thanks. |
#8
|
|||
|
|||
Re: Data mining Round 2
Just looking at the data sorted by BB/100, it looks to me like winning players (+ BB/100) fold slightly more often preflop and slightly less often on all other streets. Which is pretty obvious, of course, but what is surprising to me is that the differences are not as substantial as I would have thought.
|
#9
|
|||
|
|||
Re: Data mining Round 2
Let me know when you have a billion hands at the 10 20 and higher level.
Thanks, Ezcheeze |
#10
|
|||
|
|||
Re: Data mining Round 2
These are the conclusions that I have drawn so far
- My new ideal numbers: VP$IP 20-25, W$SF 25-30, W$SD 62+. - VP$IP doesn't seem to matter much. It seems like you can have a VP$IP between 15 and 35 and still make money. - W$SF seems to be the most important number. Get this in the high 20's with reasonable other numbers and you will make money. - WtSD seems to be bell shaped; the good and bad players both seems to show down more often. - Overall either the sample is too small or there are lots of different ways to make money. There is no one statistic that has a extremely high correlation individually with BB/100. |
|
|