Quote:
Originally Posted by TooRareToDie
When you don't know what sort of hands players pat (or snow) with in draw-games, how much sense does a calculator make? How do you think that you can give EV-% there?
First, let me argue that your premise is incorrect, then discuss how much that matters and how useful Galts is or isn't given the limitations of hand range estimates.
To start, you have a tremendous amount of information about an opponents hand in draw games. First you start with how they entered the hand, did they open UTG, call in the BB, 3b the button, or small blind? Then you also know how many cards they draw and whether they bet, called or raised after each draw.
For example assume you raise from SB, BB calls, draws are 2-2, you check, he bets, and 2nd draw goes now 2-pat. BB's range should be much wider and weaker than if he raised UTG in a full game, drew 2, then 1, then patted. In both cases you can estimate hands the opponent would play in those positions, and would draw to and bet in the manner they did to narrow down their range pretty significantly.
Galts can't help you with hand reading to estimate your opponents ranges. But what it can do is make it easy to calculate your equity against ranges you estimate. Calculating by hand is a bunch of work, and Galts also compensates for dealt cards. So in spots where your equity isn't obvious, such as vs. multiple opponents, or vs. an opponent with multiple hands in their range, or across multiple draws, or any combination of the three, Galts makes it very quick and easy to calculate a reasonable equity for the spot.
The caveats are
a) It's hot/cold equity, it's not taking into account implied odds, or reverse implied odds, i.e. hands that have your opponents dominated or drawing dead, or hands that have your you drawing thin or dead, and how many extra bets can go in when you or your opponents make second best hands. You need to consider the potential IO/RIO of your hands separately from the equity, if the hot/cold equity is borderline for a decision, IO/RIO should be the deciding factor.
b) It's a simulator, equities are calculated by running a large number of iterations of the hands (20k is default) so equities are estimated, not calculated. What that means is the equities are still very accurate, but there is random variation (1% or so depending upon the situation and number of iterations) in each run.
c) Galt's hand ranges are currently limited to a list of hands, i.e. 732, 743, 842, 853, etc. A true hand range should be definable as a percentage, such as top 10%, or as hands better or worse than a hand, such as 732+. I haven't tackled that yet because there are issues with how you rank hands in each game and situation.
d) Galts weights each hand in a hand range by the number of combinations possible for each i.e., such that a starting hand such as 732 has more combinations than 8732, and the combinations possible in each hand are also reduced by any of their cards in other hands or dead cards. This is important and very difficult to do by hand.
But given we are running on an iPhone, the number of combinations can be far too large to fully generate and calculate (would take millions of iterations to fully simulate every combo in many cases), so Galts has to reduce individual hand combinations to a manageable subset that are still representative of the range. Using subsets can't help but add some random variance to the results. But even with that extra variance, weighting still makes the equities Galts generates much more accurate than not weighting at all.
Lastly another factor is that Draw Decisions are editable in version 2.0, so you can change how hands are patted/broken/folded to simulate tighter/looser play. This can change equity pretty significantly in some spots, as can relative position (since the players how act in late position get to see how their opponents are patting/drawing before making their own decisions).
Estimating equities is the standard method for determining optimal play in every game of poker, despite the inherent uncertainty we always have about our opponents hand ranges. Galts doesn't have any magic bullet to make estimating opponent ranges more accurate, it just makes it much easier and faster to calculate equities for them for Deuce, Badugi and Ace to Five.
Last edited by DesertCat; 10-06-2014 at 12:37 AM.