OK yeah I see now. So I'm guessing that's how StoxEV comes up with unexploitable pushing/calling ranges, but how does it come up with the optimal Step 1? Just tons of trial and error?

Also, you said you wouldn't know how to start for a mixed strategy when considering unexploitable pushing.. I think he means, say you've decided you want to call or raise smaller with AA,KK or something, he's looking to exclude those range of hands from analysis when the software tries to calculate unexploitable pushing, giving those hands a different action than a push, because he thinks it will be more profitable than pushing, even though he wants to push unexploitably with the rest of his range.

Also, the unexploitable push function runs really slowly on my computer (prob greater than a minute or two for it to run for even preflop decisions). I'm running 2.19 GHz w/2GB RAM and Windows XP, fwiw.

I'm afraid that's super super secret

The current algorithm is already very complex. Adding more complexity would be an enormous amount of work with no guarantee of ending up with a satisfactory result.

If the ranges are very wide and you're looking at a push/fold situation at the flop it can indeed take something like 2 minutes.
Preflop and turn/river pushes go pretty fast though.

In a attempt to simulate this hand:
http://forumserver.twoplustwo.com/36...h-flop-353263/

I can't find a way to add a condition like:
CO bets 60% of his preflop range. (his preflop range is variable #1 in my simulation).
But, I'm not sure that's possible :/
What I'd really like to do is (i mean, it will be amazing if it's possible to do that):
CO bets (variable #2)% of his preflop range (which is variable #1).
or CO bets (variable #2)% of his preflop range.

I hope you see what I try to explain (my english is :/)
Thanks.

Working on this now, ignore previous answer.
B right back.

Expressing one variable as a function of another is unfortunately not possible. When I started writing StoxEV I intended to include this functionality but I never got around to it.



I think the best way to look at this is by using a graph with 2 variables.
I've made a tree for this hand with the following assumptions:

- Villain cold calls preflop with his top #1% of hands. However, I've discounted AA-KK and AK. He'll only cold call with these 30% of the time.

- Villain will definitely bet the flop and turn with top pair or better or a flushdraw, no matter what turncard comes off.

- Added to that he will also bluff some unknown portion of the time. I've assigned variable #2 to this. Villain will stack with his bluffs regardless of the turncard.

- If villan has missed and decides not to bluff, he wil give up on the turn unless he can beat top pair.

- If the turn comes an A you will give up the hand.

You can mail me at stoxpoker_ev@hotmail.com for the file if you want it.


The key decisions are at the flop and you make two of them:
Decision 1: Check with the intention of calling off
Decision 2: Villain indeed bets and you decide to stack.


Decision 1:
Given these assumptions I've made a 2 variable graph. Variable 1 is villain's preflop range (with AA-KK,AK discounted), variable 2 is his bluff %. The numbers are your EV for your flop line:

The numbers are slightly jumpy because they were made with the Monte Carlo engine, but this doesn't have much of an effect on the analysis.

The numbers are expressed as a function of the prize pool, asuming a 50,30,20 prize structure.

As you can see, already at a preflop range of 6% (88+,AQo+,AJs+,KQs,QJs) the line is +EV, even if villain never blufs. This is because with this range villain will only hit the flop 29% of the time and otherwise give up (if he doesn't give up with AQ,AJs,99-QQ and blufs all in it's even more +EV).

Decision 2:
Same graph, only now for the decision to stack off after villain bets.

Ok, a slightly worse picture. The decision to stack off is only +EV if villain bluffs his misses 30% or more, almost regardless of his preflop tightness.

Given the fact that villain's range is probably looser than 6% (which is the point where you require him to bluf 30% of the time) the first decision will not be affected by this and checking with the intention of stacking should be ok, provided of course that you expect villain to bluff more than 30% with his misses. Otherwise you need to give up. It's this bluff% that's the key in making the decision.

Comparison with bet-folding
Ok, now let's compare with betting 400. Villain will raise all in with at least top pair, a flushdraw and #2% of the time as a bluff.
Here's the graph for that:


The check-call line had an expectation of about 1-4% of the prize pool depending on villain's bluffiness.
The bet-call line goes from about 2% at 0% bluff to around 0% at 40% bluff.


I would say that the check-call line is the preferable one, given villain's aggresion numbers so far. The check-call line simply seems to have more potential EV-wise. Bet-fold decreases in EV pretty rapidly, even asuming bluf percentages that somewhat nitty players would have.


PS: When I'm referring to check-call I'm refering to decision 1 in the previous post.

Thanks a lot for your answer (and your analysis!)
I'm interested in the file (i'm gonna mail you) b/c I don't understand how to add a variable #2 in postflop decision.

I've added variable #2 as a weight for "all hands". It's pretty straightforward in this example because other than >=top pair or a flushdraw all other hands are just misses. Even gutshots are not that likely.

I just watched the first video while working through the demonstrated example. OMFG, this software is awesome!

I look forward learning to use it well.

Hello! I just solving the problem of EV 4-beting, while you hold AK and face to relatevely high fold to 4-bet. And found an intresting thing. A small error. Could you comment it. So the problem:
Let's say we have AK and our villain will fold to a 3-bet all his holdings except AA,KK and 50% of AK, QQ. Cause we hold AK there are still 9 holdings of AK, 3 holdings of KK, 3 holdings of AA and 6 holdings of QQ in business.Since we need only 50% of QQ and AK we have got AA -3, KK - 3, QQ -3, AK -4,5 times(the sum is 13,5) and the number of all holdings is 52*52/2=1326. So this is a 13,5/1326 =1% percent of whole number of holdnigs. Simulating same problem with Stox EV i have got a 1.1%. it's 20% bigger then real number and std isn't known. I just wonder is it a numerical error or probably I mess something. I attach a picture and can send you an stx file

A deck of cards contains 52 cards.
You have AK, so there's 50 cards remaining.
We draw the first card for villain.
We now have 49 cards remaining.
And now the second.

That comes down to 50*49/2=1225 holdings.
13.5/1225=1.1%

Thanks, that's true. My mistake.

question:

i have a model where i cbet flop and a guy checkcalls with made hands + draws, then i bet the turn.
http://img262.imageshack.us/my.php?image=nofloator0.jpg

when i add to his checkcall range a 50% overcard float, my equity increases, but the EV of my turnbet decreases. (circled in red) why?
http://img181.imageshack.us/my.php?image=floatag3.jpg

Equity increases:
You have AA, he adds overcard floats to his range. Obviously this will increase your equity against his range.

EV decreases:
Ah, that's a tough one. Look at the percentages behind "Raise to 32", "Call" and "Fold". Your EV is a mixture of the EVs of these lines. Unfortunately I can't get the EVs of these lines from the pic, but you can get them in the program by hovering over them (as I'm sure you're aware). In the float scenario villain folds much more often and calls much less often. This is because of the increased amounts of time he has nothing and folds the turn. From what I can tell the call scenario is heavily -EV for villain, which will lead to the float scenario being less +EV since it's comprised of a smaller amount of this line.

But how is it possible that the EV is smaller in a more profitable line? It's because the float line occurs in 63% of the cases and the nofloat line only occurs in 41.9% of the cases. So while the EV of the float line is smaller, the amount of money you make is larger because it happens much more often.

You can toggle to "absolute money" mode in StoxEV by pressing F9. You'll see that the float line is the most profitable.

you are right.

with nofloat, the EV is 31.08, and he raises turn 8.61%, calls turn 81.3%, folds 10%.

when he floats with overcards, he raises turn 5.72%, calls 54.1%, folds 40.2% and my EV is 26.01

toggling to absolute money shows that it is indeed more profitable ($16 vs $13) just as i would have thought. can you tell me the purpose of the relative EV and perhaps give an example of why to use it?

Relative EV is the most commonly used form of EV. Basically any EV calculation you see anywhere will be with relative EV. It asumes a known range for villain (and possibly yourself also) after which you can start figuring out the best line.

I asume in the case you've given here you're looking at your turn decision. You first looked at the situation where villain did not float and then asked yourself "well what happens to my EV at this point if villain has floats in his range?".
The fact that this decision will come up more often if villain is a floater does not affect the fact that at the turn
-you're dealing with this decision
-villain's range ánd your EV has changed because of it

I asume that what you're trying to figure out is "ok, what if I'm wrong about villain's range, is my decision suddenly bad?".

Relative EV is the only way to approach a decision. What you're looking at here is the influence of errors in your estimation of villain's range. Sometimes the influence of such errors will turn out to be marginal on your best decision, other times it will turn out that for the optimal decision you need a read on villain's tendencies and otherwise can't make a good decision.

newbie question: can you explain how the relative and absolute EV calculations differ and what the different application of these calculations are?

Why would I want to look at relative EV on the flop? Why would I want to look at absolute EV on the flop?

Relative EV is the EV of a decision at a certain point in the hand where the ranges are known.

Absolute EV is relative EV weighed by how often it occurs.

In most cases where you're doing an analysis you'll probably want to use relative EV. Here you make assumptions about the ranges and what lines villain will take with certain hands to figure out your best lines.

However, sometimes you want to know the influence of changing the ranges at some earlier point in the hand. In that case you're probably better off using absolute EV, since, as the previous poster pointed out, it can become a bit unclear what's going on.

A second use of absolute EV is to get an idea of how important certain branches of a tree are. For instance you can have a branch where you're dealing with a flopped full house or better. Your relative EV will obviously be huge, however, because it happens so rarely the use of absolute EV will make it more clear that this is not a branch that you need to be too concerned about in your analysis since it happens so rarely.

Hi guys,

I've recently released StoxEV 2.1.8 but for some reason there's a problem with the automatic update.
The program will give an error.

To solve this, just manually start StoxEVmain.exe.
You only need to do this one time, after this it will work fine again.

Cheers,

Scylla

My computer clock sometimes slows down due to me running Windows under a virtual machine, so today my date was in December and I was still running last months version fine until it updated for the new one, so I had to change my date to January to run it, and now it says "The trial period of this application has been manipulated. Please, register this software." Is there anything that can be done? I hope I don't have to wait a month before I can use this great piece of software.

I'm afraid you're boned for the next 2 weeks or so.
Once you get this error from the encryption software, there's no appeasing it.
I typically post an update every two weeks to help out the people who get this error.

OK, it will work if I install StoxEV on a fresh installation of Windows right?

Yes, that should work.

been going through the program, and this thread, the last 2 days. All I can say is this is awesome. IM still a bit fuzzy on the differences between absolute and relative ev and what they can be used for, but Ill watch the videos. Great Job. Any plans to not offer this as freeware, and start selling it?

Scylla,

What's the Holdem Manager Format?

I assume it's the Hand History import you get if you click on the "text" button in the Replayer Hand History Pop Up in HM:

(sample text here
But that isn't recognized. What's then is the Holdem Manager format?

thx!