Poker-AI.org Poker AI and Botting Discussion Forum 2013-11-05T07:57:10+00:00 http://poker-ai.org/phpbb/feed.php?f=26&t=2617 2013-11-05T07:57:10+00:00 2013-11-05T07:57:10+00:00 http://poker-ai.org/phpbb/viewtopic.php?t=2617&p=5176#p5176 <![CDATA[Re: HUNL SNG Antes]]> Nasher wrote:

I was wondering what the typical ante levels were for HUNL Super/Hyper Turbo SNGs across different casinos? I was also wondering how some of you handle them in the context of your strategy use/development, as they're sometimes there and sometimes not, given any particular effective stack size.


I haven't really though about it in depth but when looking at the two naïve approaches 1) and 2) [...]

// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

1) Lower bound (Ignoring ANTE): In this case your bot will play too tight since your bot is assuming the pot being too small
2) Upper bound (Distributing ANTE):

Let A be ante, S_old be the small blind and B_old be the big blind, so at the beginning of each hand let the pot be: (LaTeX-Plugin for the forum pleaseeee)

Code:
P = 2*A + S_old + B_old,

A := Ante
S_old := Small blind
B_old := Big blind


You can now let

Code:
P' = S_new + B_new,

S_new := S_old + 2 * (1/3) * A
B_new := B_old + 2 * (2/3) * A


resulting in the new effective stack size E_new (in BB)

Code:
E_new := E_old * (B_old / B_new)

for each player


which would lead to your bot being too loose since it is assuming the pot being too large

// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

[...] it turns out that the truth lies somewhere between these two approaches. Two quick fixes for approximating the truth could be:

1) Modifying effective BB-Size: B_new = f(A) * B_old (with a somewhat smart function 1 <= f(A) <= 2 * (2/3) * A)
2) Playing preflop according to your \sigma_i and manipulate the size of the modeled pot on later streets to match the actual pot size (i.e. bet slighly larger portions of the pot than your model would suggest such that the pot size of your model P_mod == P_real (the actual pot size) when being all-in)

Ya ... something like that ... :)

Statistics: Posted by Nose — Tue Nov 05, 2013 7:57 am

]]> 2013-10-19T17:32:22+00:00 2013-10-19T17:32:22+00:00 http://poker-ai.org/phpbb/viewtopic.php?t=2617&p=5120#p5120 <![CDATA[HUNL SNG Antes]]> Statistics: Posted by cantina — Sat Oct 19, 2013 5:32 pm

]]>