Just fold "all hands".
No need to exactly specify which hands to fold, this way you just fold all hands that make it to this part of the tree.

You might want to watch the instructional videos to learn how to use the program. I'll make a third one in a while.
Go to "help->video manual".

Hehe, we've been around since March

I think This Stox calculator with tournament equities is better than the Renaissance, ten moon landings, and a Brittany Spears sex tape all put together.

A few comments relating to tournament calculations:
1) Comparing all $EV numbers to pre-post equities is confusing. I would like to be able to define a 0$EV condition, default to Hero folds, so the best in-game decision is most clearly illustrated. If it was as simple as a manual offset, so I could make the graphs comparable to SNG Wizard and everything else SNGs players are used to seeing, that would be fine.

2) Putting the option for more precision on the ICM equities is necessary. 0.01% of the prize pool is the absolute minimum in the final results and I don't really trust it for inputs into further simulation.

3) The ICM-calculated equities shouldn't be truncated. Why not just round?

4) Export the graph results?

Already there. After you perform an EV calc, hover over the action/condition from which you want to remove all -EV starting hands and press Alt+D.
In case it's preflop these starting hands will be removed from the starting hand matrix. Postflop a skip condition will be added.

By the way, Alt+D is not the only thing you can do while displaying a popup. Here's the list of options:
- Remove all -EV hands with Alt+D
- Toggle to displaying equities with Alt+E
- Toggle to comobs with Alt+C
- Press Alt+O for a dialog with Excel output of the displayed matrix. First line displays a string that can be exported to PokerStove.

The entire list of options is shown at the bottom of the popup.
Is this a confirmation of my decision to include the ICM warning or a request for me to change something? Anyhow, just FYI, the % is not a % of the prizepool. It's the standard deviation that is accepted in the outcome. For instance if the ICM value of your stack is $100 and the precision would be 1% than it would basically mean that the program's acceptable output range would mostly be within a $99-$101 range.
Furthermore, if either the number of players or the number of prizes is 8 or less then the results are the mathematically correct results. The MC approach (with the % error) will only occur for larger fields.

At some point during writing the program I had to make a decision where to round the EV numbers. Since we're talking about money, I felt that displaying down to the cents would make most sense.

Already there. Push the yellow E button on the graph field. You'll get a dialog from which you can copy paste to Excel. It's not pretty, but it works.

For reference, I have started working with this program per Hood's instruction here, analyzing the SNG hand I presented here.

I think I should try to explain again. Here is the excerpt from Hood's post that think best explains my problem.

In other words, I see that the $EV value is calculated for the case where we raise AQo. I want to compare that value to the value that would result from our other alternative in the hand, which is to fold. It is not relevant to cash games but it is to tournaments, since your equity can change a lot based on what happens after you fold. Right now it seems like I have to construct a totally new tree in order to get the $EV value for folding. Is this correct?

There are two ways this has been done in the past. The SNGPT way was to assume all action folded around to the BB if Hero folds. This was OK but not great, especially as the players get better. The SNG Wizard way is to allow the user to construct that branch, and that was one of the major improvements in Wiz over SNGPT.

I'm pretty sure this is a request to fix the tolerance on the ICM equities to a specific % of the prize pool, not an individual player's equity, and make that tolerance 0.01%. It seems as is, it makes the ICM calculation error larger or smaller depending on the stack size.

There is a way. I believe I describe it in the second part of the video manual.
You simply create both branches and then assign a 50% weight to the top branch. Below a screenshot of what this would look like (without the subbranches):


To get the EV of both lines, perform an EV run and then
- hover over the "raise" action to get the EV of raising
- hover over the "fold" action to get the EV of folding

Obviously, the EV of the decision itself is now a mix of raising and folding and makes no sense whatsoever . So please ignore that number in this setup.

I'm already using the second option of letting the user create the subbranch, so that should work in the way you're asking for here. The EV of this line is then compared to a scenario where the remaining players split the current pot.

I think that the current method should not be too far off of what you're asking for here, though I have to admit I'm not entirely sure. Can you give me an example file (mail to stoxpoker_ev@hotmail.com) where you're experiencing problems please? It will be easier for me to focus on identifying the problem.


I could set the warning to 0.01%, however, in larger fields this becomes a bit too slow to calculate. However, if you wish more accuracy, just set the % to 0.01 in the "tournament stacksizes" menu.

In the way the program works, when determining the ICM values with an MC algorithm, there's no way to know who hero is. For that reason I'm measuring the standard deviation of the median stack and check for when it falls below the given % to terminate the MC algorithm.
It's possible that the size of hero's std varies a bit from the desired %, but it shouldn't be a dramatic difference. And if it ís, you should just set the % to a smaller number.

Perfect. That is exactly what I was looking for.

Thanks for that link by the way.

I noticed in the first screenshot that I'm displaying the first help message below all the other bitmaps.

Also, I forgot to add bitmaps for the help popups.

Not pretty.

Just fixed it.

OK, so for situations where I have 8 or fewer stacks, the ICM-calculated equities used are always exactly the ones I would see by hitting F11, where the dialog that pops up is an example for the stacks before any action or blind/ante posting?

I just found out that tournament equity has been added and all I can say is WOOOOOOOOOOOOOOOOOOOOOOOOOOHOOOOOOOOOOOOOOOOO!!!

Juk

That is correct.
Same thing if you have 8 or fewer prizes (even if the field is large).
Either one will allow the mathematically correct solution to be calculated.

And as you will see, if you press F11 over and over again the numbers won't change. But if the field+prizestructure is larger they dó change.

It's been around since april I think .

Did take a few months of tinkering since then to resolve some problems (for instance folding with several players left to act, which makes folding non-0-EV).

This is freakin awesome, many thanks

OK, so I still think it is possible to get precision errors.

This is from Hood's thread, using 50/30/20 for the payouts.

The actual values out to 5 places, with equities as a % of the prize pool, are:
30.985
38.648
14.982
15.385

In the hand, we are trying to compare different options to folding, which leaves Hero with 31.35% the way I've set it up. If I use the truncated numbers, I can't tell the difference between 30.10 and 30.19, which is enough to be important.


It looks like the way you've done the ICM calculation, that everything past the first decimal place just gets truncated. If I enter the payout as 50,000/30,000/20,000, I get this, which looks sufficiently precise.



I guess my point is tournament equities should always be calculated as a percentage of the prize pool to 0.001% before they are used as inputs for the tree, regardless of how the user enters it.

Oh crap, I see what you mean now! I'm indeed incorrectly truncating the numbers in the F11 menu. I never remembered to fix it. I ran across this problem myself a couple of times when manually checking the calcs. No worries though, the program dóes use the correct untruncated numbers in the calculations.

I'll put it on the to do list.

For now you can use the ICM conversion tool for calcs (it's the button with "ICM" on it). It should automatically load the situation (tournament mode must be turned on for this) and ít dóes truncate correctly.

The reason why I forgot about the F11 tool is because since adding the ICM calculator the F11 tool is almost obsolete.
I'll update the ICM calculator a bit and remove the F11 tool altogether I think.

Ah OK. As long as it isn't using the truncated numbers as input, it should be fine.

We'd be much obliged if you could help us out in Hood's SNG thread here.

Ok, I just updated the ICM tool and removed the old F11 tool.

I'll post the update somewhere around half this month.

There are 2 green numbers: Eq and $. The Eq is working for me but nothing shows up for $ when I do an equity run. I have everything turned on in the "extra" menu.

Equity runs (F5) do not calculate EV.
EV is only calculated for EV runs (F7 or the button with "arrow+EV").

Does that solve your problem?

Didn't realise it automatically filled in the info, seriously you have thought of everything.

This looks incredibly useful, going to try it out asap!

I unchecked "use speedup" and that seemed to fix it. For some reason I can't do EV runs with that checked.

Could you mail me the save file please so I can get a better idea of the problem?
www.stoxpoker_ev@hotmail.com.