minor request: could you move the icon above the "raise" label ? pretty easy to figure out if you raise/call/fold on that node but always a bit annoying to actually click on the node to see what the first condition is.

I think I could move it behind the name of the raise action ...
That would probably be an improvement.

Thank for the feedback!

Yes.

The easiest way to approach this would be to not get into detail too much, but just use a checkdown at the flop.

Given that BB will be out of position you may want to apply a percentage to the checkdown.
For more on checkdowns, please go here:
http://www.cardrunnersev.com/manual/checkdown.html
Specifically the section on custom checkdowns.

After you have set up the tree as you've wanted (so with fixed ranges for Button), apply the Max Exploit tool for BB to see how to maximally exploit Button.

The tree would look something like this:



PS: In older versions, once there was play on an unknown flop, the monte carlo engine would be used. More recent versions however do stíll allow the math engine to work provided that all play is for "all hands". So in the tree above, the math engine will still function.

Regarding the checkdown function: has anyone figured out realistic numbers to approximate the positional advantage?
Would something like
110% if no river betting
120% if no turn betting
130% if no flop betting
sound like a reasonable starting point? I.e. if some preflop betting is modeled, and players (somehow) end up seeing the flop with 50/50 ranges and there's no betting from flop onward, BTN would get 1.3*50% = 65% of the pot?

I have some ideas as for how to grab some empirical data from a PT database... anyone interested in this kind of stuff, feel free to get in contact.

I'm also wondering whether those numbers would be actually totally thwarted by the equity that the better range can grab just by betting optimally.

I'd like to put in a feature request for savable strategies for both players, and a drop down menu to quickly alternate between them. By strategy I don't mean any single condition such as a post-flop action condition, or a pre-flop range, but the players whole action tree. This would come handy when comparing how different adjustments and counter-adjustments play out vs. each other.

Another cool feature I can think of is being able to pin point what percentile a single hand falls into from a set range and a board, but maybe this would be more suitable for Flopzilla?

Hi!

I just bought the program and I am trying to look at a basic cold 4bet situation. I want CO to open 25% BTN to 3bet 12% and I want to know how much I can cold 4bet from the BB if they play optimal against my range. This seems like a situation a program like this should be able to solve pretty well if we assume that all players are 100bb deep and that BTN can only fold or shove.

Hi,

I'll need to get back to you tomorrow on this.
About to go to bed.

Sorry.

Cheers,

Scylla

thanks scylla, will do

@pasita
What did u mean "grabbin empirical data from database"?

There's a typo in the Beta's script editor window One too many t's in the "Close edittor".

In a script I can't set a variable to a negative number. Is it possible at all?

Ok, well given that you're asking for a situation where:
- You play optimally
- Cutoff plays optimally
- Button plays optimally

we're looking for a spot where you're looking for the equilibrium for 3 players.
This is rather hard to accomplish.

It becomes a lot easier though if you fill in Cutoff's and Button's pushing ranges as well and júst make BB be the only optimal player.

For example, please see the tree below.
I've set it up so that if Cutoff pushes, Button will automatically fold.
It's entirely possible to also included a line where Button would continue as well, but I thought I'd keep the tree simple.



After computing this tree (I'd recommend roughly 1M runs), mouse over locations 1 and 2 to delete the -EV hands for BB (by pressing Alt+D). Don't worry too much about the somewhat scattered results for the hands. The scattering only occurs for hands that are close to 0 EV anyhow.

Then recompute with F7 for the changes to take effect, mouse over location 3 and again press Alt+D to remove the -EV hands.

That should give you the optimal 4bet range for BB versus Cutoff's and Button's play.

Click here for the savefile.


The challenge here is that you're playing versus two players, rather than this being a heads-up tree. If it were a heads-up tree, the math engine would work (results are now based on Monte Carlo) and you could simply use the Max Exploit tool to calculate BB's optimal play.

&*^*&^%*&^&
I make that typo a lot.
Another one I get wrong a lot as well is "address".
I'll fix it for the next release.

No, that's not possible.
It's a security feature, since there's no value that variables can be set to that can ever be negative.
You cán sort of "hack" the system though by using the "Decrease variable" command.
The software won't notice that you're going below 0 if you use that.

I think both of these are fine suggestions, but they have the disadvantage that it's extremely difficult to add them without them messing up the interface. Every time I add a button or popup the interface becomes more complex and therefore harder to work with. I don't think I'll be able to add them.

I needed to store a negative value to compare the EV against and find the max EV. If I set it to 0 but the EV is always negative it becomes impossible to find the max EV.
Yeah, thought this workaround might work, rhanks!

I meant filtering my db for certain situations where both holecards are known, calculating the exact equities on the flop, then comparing to how well button actually did in this situation. At least with PT4, this would require some external program. Anyone know if HEM stores street by street equities for every hand with known holecards? PT does that for the all-in scenario only.

I can see huge bias problems with this, though. Also, since I have next to none NLHE hands in my db, the results might only be interesting to the 10% of the community who actually prefer a game with multiple streets of betting usually occurring

I have some rough ideas on how to overcome the bias, but really not too much time or inspiration right now to start looking at this in detail. If anyone is interested in this, feel free to PM though, never know if the inspiration pops up.

How would you go about finding all hands that have at least x equity vs a given range? Ex. All hands that are at least 30 percent equity vs top 40 percent of hands?

1) Perform the action for "all hands"
2) Compute with F7
3) Mouse over the "all hands" condition
4) Press Alt+F for the filter menu
5) Select "Only keep hands with equity of at least X"
6) Under "Use this value for X" enter 30
7) Press OK and recompute with F7

This is great help. Thanks you very much!

Hey Scylla, apologies if this question has been answered somewhere in this thread but I didn't want to trawl through 185 pages. I am trying to compute a graph which show's Hero's EV vs Villains calling range where I want to vary the width of their range I.E. top(X%). However, it says I cannot substitute a variable and that I must use an integer. Is it possible to compute the graph to which I am referring?

If you're talking about preflop, then you can use the "Use variable" button to attach a variable to the top %. See C in the pic on this page:
http://www.cardrunnersev.com/manual/preflopmenu.html

If you're looking for postflop, then indeed it's not that simple to just attach a variable to a top %. After all, there's no ranking for the hands available (the same is true preflop, but there the standard hand ranking will go a long way). If you want this graph anyhow, then you'll need to write a script. To learn how to work with scripts, please watch the instructional videos on this subject. They are linked to in the beta's startup screen. Click on the "Videos" button. The videos will walk you through the meaning of all the commands. The scripting system is drag and drop based, so it should (hopefully) be fairly easy to learn. Once you learn how to work with the script commands you should be able to automate all sorts of things.

The script should look something like:

1) Set variable 1 to 5
2) Set "all hands" at location 1
3) EV run
4) Set top #1 % at location 1
5) EV run
6) Add data point (#1,EV(location 2))
7) Increase variable 1 by 5
8) IF(#1<=100)
9) GOTO 2
10) ENDIF
11) Draw graph

Here, location 1 should be set at villain's decision.
And location 2 should be where you want to measure hero's EV.

Hi Scylla. What is the formula used by the program to calculate the variance?

As the square root of S^2 here:
http://davidmlane.com/hyperstat/A16252.html

Technically the number the software gives is actually the "Standard Deviation", however "Variance" as a number has no real world application.

For example, if you bet for $100 with a 50% chance of winning then:
- Your EV is $0
- Your Variance is $10.000
- Your standard deviation is $100

And if you bet for $1.000 with a 50% chance of winning then:
- Your EV is $0
- Your Variance is $1.000.000
- Your standard deviation is $1.000

And if you bet for $10.000 with a 50% chance of winning then:
- Your EV is $0
- Your Variance is $100.000.000
- Your standard deviation is $10.000

So, "Variance" is a meaningless quantity (imho).
You need to take its square root for it to have practical meaning.

I also strongly suspect that whenever someone uses the term "Variance", they actually mean "Standard deviation" (given the fact that variance is meaningless until you take its square root). Just to make communication a bit easier, I've just stuck with the term "Variance".

FYP to underline your point.

thx scylla.