Quote:
Originally Posted by shetu
Ok, take this range as example : All Broadways, All Pocket Pairs.
On a particular board with the range above, the solution could said that :
- the JdTd hand must be raised 31% of the time, called 54% and folded 15%.
- the JhTd hand must be raised 11% of the time, called 89% and folded 0%.
- The global EV of this flop node is 3.6 bb.
Is it possible to alter the algorithm to round up these numbers like below without sacrifice too much EV: (That solution will be the easiest to remember)
- the JdTd hand must be raised 100% of the time,
- the JdTd hand must be called 100% of the time,
- The global EV of this flop node is 3.2 bb.
As an option, I would like a hybrid solution where percentages are only 0%, 50% or 100%. (That solution will be the harder to remember, but less than the GTO+ solution)
- the JdTd hand must be raised 50% of the time, called 50% of the time,
- the JdTd hand must be called 100% of the time, folded 0%.
- The global EV of this flop node is 3.4 bb.
A problem here is that if, for example, a range has 500 combos, and you have 20 different trees, this would come down to having to memorize 500*20=10.000 combos. At the very least I personally would not be able to do this. Also, there's the challenge of figuring out how to adapt if the ranges are even slightly different, or if the board is a bit different. Even if simplifying to 0% or 100%, I still don't see how memorizing combos is an attainable goal. I am happy to offer rounding if our users really want it, but for the moment it's not there for this reason. The approach that we offer is to provide a breakdown according to stats, as well as many graphing features. The goal here is to try and find patterns in the data that occur generally across different boards or ranges. Such patterns, like for example that OOP should always bet with a polarized range on the river (only bet the strongest and weakest hands) are far easier to remember and apply in practical play.
Last edited by scylla; 09-14-2018 at 04:27 AM.