|
|
 |
card removal all-in equity range calculator |
 |
1
 |
jchysk United States. Dec 31 2009 02:02. Posts 435 | | |
Recently, after talking to woodbrave and firah, who both wanted something pretty similar I started scripting what I thought would be a simple enough calculator.
Here's the php version for just card removal: http://www.jchysk.com/testing/flophands.php
About an hour into scripting I realized the big problem I was facing was computational time and switched over to C++. Another hour in I realized this was going to be extremely slow.
The basic idea is take any preflop hand and use card removal to figure out opponent hands. based on the range you select them having it goes through all their hands in that range (not nPr 48,5 but nCr 48,5 or nCr 45,2 if there's a flop)
Use a 7-card evaluator on every board to figure out win, loss, or tie.
Loop through the steps to accumulate all wins, losses, and ties until each opponent's hand in range has been accounted for.
The numbers start getting big, but still doable, until an additional player and range is included.
After I realized this wasn't going to be practical for any use at all simply because computers are too slow and it would take me forever to optimize and tweak it full of shortcuts to speed the process up I decided to look online and see what has already been developed. Too bad I didn't make that step 1, because I definitely wasted a lot of time on something that wasn't going to happen.
I suppose I'm just so used to thinking hardware is cheap and incredibly fast these days that you can write inefficient brute force code to get what you want without much consequence. It's yet another big punch in the face wake up call to be reminded how voluminous the permutations in poker really are.
|
| |
|
| 0
 |
devon06 Canada. Dec 31 2009 02:24. Posts 116 | | |
holy fuck man... just play solid poker instead. and tell your friends the same
or don't |
|
| 1
 |
pluzich . Dec 31 2009 08:32. Posts 828 | | |
Maybe use sampling (Monte Carlo) instead of iterating through every possibility. You can derive confidence intervals for equities.
Maybe share the code and we can have a look together. I always wanted to do something similar but was extremely lazy 
And then we could maybe profile that stuff and see which part takes most time to execute; usually if you manage to speed up something in the innermost loops you'll get very good overall speedup. |
|
| | Last edit: 31/12/2009 08:33 |
|
| 1
 |
lachlan Australia. Dec 31 2009 10:11. Posts 6991 | | |
amen to that sir
confidence intervals... haha i remember that shit from STAT2201... UQ ftw
i have 95% confidence that you will succeed in life |
|
| full ring | Last edit: 31/12/2009 10:12 |
|
| |
|
|
 Poker Streams | |
|
Contact Users: 475 Active, 0 Logged in - Time: 16:30
