Two Plus Two Publishing LLC Two Plus Two Publishing LLC
 

Go Back   Two Plus Two Poker Forums > >

Notices

Commercial Software Discussion of commercial gambling-related / poker software & commercial graphics modifications

Reply
 
Thread Tools Display Modes
Old 06-07-2017, 06:05 AM   #5826
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by mrfunnywobbl View Post
How many people are on the devt team?
We don't really disclose any such information, so unfortunately I won't be able to answer that question.

Last edited by scylla; 06-07-2017 at 06:18 AM.
scylla is offline   Reply With Quote
Old 06-07-2017, 06:18 AM   #5827
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by MU MUF MUFC OK View Post
Hi Scylla, When OOP bets the river how do I check the equity of a specific hand in OOP's betting range vs IP's calling range?
I can't say that we've ever considered such a feature, given that it's not directly relevant to decision making.
That's not to say that it would not be interesting though, so I'll see what I can do for future releases.

Thank you for the feedback,

Scylla
scylla is offline   Reply With Quote
Old 06-07-2017, 08:21 AM   #5828
MU MUF MUFC OK
grinder
 
MU MUF MUFC OK's Avatar
 
Join Date: Jan 2010
Location: at home
Posts: 423
Re: CardRunnersEV

It is very relevant to decision making when deciding whether to value bet the river or not OOP

Can you think of some sort of quick work around to check which hands have enough equity vs the calling range to value bet?
MU MUF MUFC OK is offline   Reply With Quote
Old 06-07-2017, 10:24 AM   #5829
MagicGog
veteran
 
MagicGog's Avatar
 
Join Date: Aug 2007
Posts: 2,611
Re: CardRunnersEV

Quote:
Originally Posted by scylla View Post
It's hard to say without at the very least seeing a savefile, given that I can't tell what sort of trees you have added after each action. If you want me to take a look, then please mail a savefile to support. Should you need to remove calculation data from the tree in order to reduce its size, then:
1..5
Hi

Looks like no errors were here. I removed several bets and left options
0.5p, 1.5p, 5p, check. It counted that 1.5p is preffered sizing. Prob it's actually the optimal desicion.

I have an idea about the speed of calculation process. I think you know that crypto miners use GPU (not CPU). Prob it's possible to load any sub-trees to GPU ?
MagicGog is offline   Reply With Quote
Old 06-07-2017, 01:19 PM   #5830
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by MagicGog View Post
Hi

Looks like no errors were here. I removed several bets and left options
0.5p, 1.5p, 5p, check. It counted that 1.5p is preffered sizing. Prob it's actually the optimal desicion.
It might be, however, once again, if you want me to take a look, then please send a savefile to support. The different lines will only make sense if the continuations after they are taken are similar. Unfortunately I can not tell whether or not this is the case from just the screenshot.


Quote:
Originally Posted by MagicGog View Post
I have an idea about the speed of calculation process. I think you know that crypto miners use GPU (not CPU). Prob it's possible to load any sub-trees to GPU ?
The calculation that needs to be performed on the subtrees are rather complex, whereas, to be best of my knowledge, GPUs can only be used to perform relatively simple calculations. It may still be worth looking into at some point though.
scylla is offline   Reply With Quote
Old 06-07-2017, 01:28 PM   #5831
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by MU MUF MUFC OK View Post
It is very relevant to decision making when deciding whether to value bet the river or not OOP

Can you think of some sort of quick work around to check which hands have enough equity vs the calling range to value bet?
If this is just a regular tree, as opposed to a tree with GTO data in it, then you can use a delete action to remove all actions other than the call action (see bottom pic). After that, compute with F7, mouse over the player's bet action that you're interested in and press Alt+A for the old range analysis tool. This will give you the equities for all of the hands versus only the hands in the call action. The reason why this trick works is because the delete action will retroactively remove all scenarios where villain did not call.

However, please do note that this is actually not relevant to your decision making process.

For an example, see the top and second pic. In the first BB's EV is 1.79 with SB calling with >=tp. However, when SB only calls with >=quads (second pic), then BB's EV becomes 39.64, which is a much higher EV, despite the fact that SB's calling range is far stronger. This is because in the second scenario SB folds 99.7% of the time, with BB winning the pot of $40. So, not only is your equity versus the calling range important, but the money you make when villain folds is important as well. It's for this reason that you actually need to look at your EV when playing versus villain's entire strategy, as opposed to looking at equities.

scylla is offline   Reply With Quote
Old 06-07-2017, 06:36 PM   #5832
SoyLate
stranger
 
SoyLate's Avatar
 
Join Date: Oct 2011
Posts: 13
Re: CardRunnersEV

is it possible to make the solver round strategies to certain increments? is this in development? i constantly find myself looking at spots, write down what i think is optimal for each part of my range, only to find the solver producing a very mixed strat. i was wondering if there is a way to translate the results to something more practical in certain spots.
SoyLate is offline   Reply With Quote
Old 06-08-2017, 08:37 AM   #5833
MU MUF MUFC OK
grinder
 
MU MUF MUFC OK's Avatar
 
Join Date: Jan 2010
Location: at home
Posts: 423
Re: CardRunnersEV

Quote:
Originally Posted by scylla View Post
For an example, see the top and second pic. In the first BB's EV is 1.79 with SB calling with >=tp. However, when SB only calls with >=quads (second pic), then BB's EV becomes 39.64, which is a much higher EV, despite the fact that SB's calling range is far stronger. This is because in the second scenario SB folds 99.7% of the time, with BB winning the pot of $40. So, not only is your equity versus the calling range important, but the money you make when villain folds is important as well. It's for this reason that you actually need to look at your EV when playing versus villain's entire strategy, as opposed to looking at equities.

I don't know what point you're trying to make here. Obviously if villain is folding 99% our bet is going to have a high EV.

I have already set the IP players calling range. I need to know the EQ of OOP's hands vs the calling range, so I can value bet hands that have around 50%+ equity vs the calling range, and decide on a balanced bluff range.
MU MUF MUFC OK is offline   Reply With Quote
Old 06-08-2017, 03:08 PM   #5834
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by MU MUF MUFC OK View Post
I don't know what point you're trying to make here. Obviously if villain is folding 99% our bet is going to have a high EV.

I have already set the IP players calling range. I need to know the EQ of OOP's hands vs the calling range, so I can value bet hands that have around 50%+ equity vs the calling range, and decide on a balanced bluff range.
Ah, ok, in that case I somewhat misunderstood your question.

The direct answer to your question remains the same though. We currently have no dedicated feature for this, and I can look into providing one, however, for the moment, I can offer the following workaround:

1) Add a "Delete all hands" action below villain's "Call" action (see pic below)
2) Compute (with F7)
3) Mouse over hero's "Bet" action
4) Press Alt+A for the old range analysis tool (see pic below)

This tool will now give you the equities versus the calling range.
You can even use it to select a top+bottom% of the range.
For this, adjust the sliders like in the picture below.





Please do note though that you don't need as much as 50% equity to justify a shove, given that the pot offers an overlay. For example, if the pot is $40 and the stacks are $50, then shoving and getting called will create a pot of $140. You will only need to win the showdown 50/140*100%=36% of the time in order to justify shoving and getting called.

Other than that, I'd like to point out that the GTO solver can actually figure out the mathematically correct answer to your question. For example, you can use it in the tree below to figure out a GTO solution:



However, you can also enter your assumptions about villain's play, lock it to the solver and then let the solver figure out the solution within the given restrictions:



From what I can tell, betting ranges on the river tend to be polarized between value bets and bluffs, with middling hands often being checked, and it appear that this example hand is no exception.

To take a look at the ranges generated by the solver, please use its analysis tool (mouse over the relevant decision/action and press SPACE). For more on the GTO solver, the analysis tool, the locking system and other related subjects, please watch the videos here: http://www.cardrunnersev.com/download.html

Here is a link to the savefiles: www.cardrunnersev.com/savefiles/savefiles.zip



PS:
Please do note that the delete trick will not work in conjuction with solver data, given that they are not designed to work together. You can use the Delete trick with regular conditions though, like in file1.stx. Or also in file3.stx, given that there villains' actions only contain regular conditions.
scylla is offline   Reply With Quote
Old 06-08-2017, 03:27 PM   #5835
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by SoyLate View Post
is it possible to make the solver round strategies to certain increments? is this in development? i constantly find myself looking at spots, write down what i think is optimal for each part of my range, only to find the solver producing a very mixed strat. i was wondering if there is a way to translate the results to something more practical in certain spots.
I've run some tests with this to see if the solution converges if increments are used, and this was unfortunately not the case. At a certain point, the solver stops converging and starts oscillating, often relatively far away from the solution.

An alternative here would be to round mixed strategies to certain increments after the solver has completed its task, however, personally I think that this is something to avoid. One reason is that by doing so, we would be substituting one system that's very hard to understand with a different system that's still very hard to understand. A second, more important reason is that by rounding to, for example [0,100%] increments, important information is lost. Namely, if a hand is mixed between two actions in a tree that is in an equilibrium (meaning dEV=0) then the EVs for both actions will actually be exactly the same and the player will be indifferent between the two actions. Rounding to one of the two actions would actually hide this fact.

The most important thing to take away from a hand being mixed is that the player does not really care which action he takes with the hand; the EV will be the same in either action. The frequency of the mix is not because the player is tempted more or less towards a certain action (there is no incentive), but only and solely to set up relative frequencies of hands in all lines that follow the decision. So, when a hand is mixed, the frequency of the mix is not as important as noting the simple fact that there is a mix in the first place.

We do have plans for future releases where we hopefully can make more sense out of the solutions that come out of the solver, however, in my opinion, rounding the solutions afterwards is one that we would like to avoid for the moment (on the other hand, rounding while solving would be great, but, once again, as far as I've been able to tell, this does not seem to converge).
scylla is offline   Reply With Quote
Old 06-08-2017, 07:14 PM   #5836
SoyLate
stranger
 
SoyLate's Avatar
 
Join Date: Oct 2011
Posts: 13
Re: CardRunnersEV

Quote:
Originally Posted by scylla View Post
I've run some tests with this to see if the solution converges if increments are used, and this was unfortunately not the case. At a certain point, the solver stops converging and starts oscillating, often relatively far away from the solution.

An alternative here would be to round mixed strategies to certain increments after the solver has completed its task, however, personally I think that this is something to avoid. One reason is that by doing so, we would be substituting one system that's very hard to understand with a different system that's still very hard to understand. A second, more important reason is that by rounding to, for example [0,100%] increments, important information is lost. Namely, if a hand is mixed between two actions in a tree that is in an equilibrium (meaning dEV=0) then the EVs for both actions will actually be exactly the same and the player will be indifferent between the two actions. Rounding to one of the two actions would actually hide this fact.

The most important thing to take away from a hand being mixed is that the player does not really care which action he takes with the hand; the EV will be the same in either action. The frequency of the mix is not because the player is tempted more or less towards a certain action (there is no incentive), but only and solely to set up relative frequencies of hands in all lines that follow the decision. So, when a hand is mixed, the frequency of the mix is not as important as noting the simple fact that there is a mix in the first place.

We do have plans for future releases where we hopefully can make more sense out of the solutions that come out of the solver, however, in my opinion, rounding the solutions afterwards is one that we would like to avoid for the moment (on the other hand, rounding while solving would be great, but, once again, as far as I've been able to tell, this does not seem to converge).
makes sense. thanks for explaining
SoyLate is offline   Reply With Quote
Old 06-10-2017, 11:24 AM   #5837
MagicGog
veteran
 
MagicGog's Avatar
 
Join Date: Aug 2007
Posts: 2,611
Re: CardRunnersEV

Quote:
Originally Posted by scylla View Post
It might be, however, once again, if you want me to take a look, then please send a savefile to support. The different lines will only make sense if the continuations after they are taken are similar. Unfortunately I can not tell whether or not this is the case from just the screenshot.
I sent you link to stx file last night to support@cardrunners-ev-calculator.com
Is it working email, prob send to any other one (you have so many different emails as I see ))
MagicGog is offline   Reply With Quote
Old 06-10-2017, 01:41 PM   #5838
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by MagicGog View Post
I sent you link to stx file last night to support@cardrunners-ev-calculator.com
Is it working email, prob send to any other one (you have so many different emails as I see ))
It's a working e-mail, however, I can't say that I have received this mail.
It sounds like a spam filter has been involved here.
Can you please try sending to one of the other addresses?
scylla is offline   Reply With Quote
Old 06-10-2017, 04:40 PM   #5839
Siesta
stranger
 
Join Date: Dec 2015
Posts: 5
Re: CardRunnersEV

I'm running into a character limit on the import program settings field. Would it be possible to remove/increase this limit in future versions?
Siesta is offline   Reply With Quote
Old 06-10-2017, 10:24 PM   #5840
MagicGog
veteran
 
MagicGog's Avatar
 
Join Date: Aug 2007
Posts: 2,611
Re: CardRunnersEV

Quote:
Originally Posted by scylla View Post
It's a working e-mail, however, I can't say that I have received this mail.
It sounds like a spam filter has been involved here.
Can you please try sending to one of the other addresses?
sent to
contact@cardrunnersev.com and support@cardrunnersev.com

One thing I wanted to ask you for several monthes but forgot each time. Here big problem with importing files from KO tournaments. It does not work. And I need to edit manually HH text by temoving some strings and transforming HH as it was freezout. It takes time, sometimes very annoying and often impossible cause of big tireness after play.

PokerStars Hand #171594414523: Tournament #1930916414, ($10+$30+$4) -> ($44) USD Hold'em No Limit - Level I (15/30) - 2017/06/10 18:30:35 MSK [2017/06/10 11:30:35 ET]
Table '1930916414 5' 9-max Seat #2 is the button
Seat 1: MilchoBG (4995 in chips, $30 bounty) -> (4995 in chip)
.........
and other players rows needed to be fixed manually.

Please look is it possible to fix in next release. Now so many KO tourneys
MagicGog is offline   Reply With Quote
Old 06-11-2017, 07:11 AM   #5841
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by Siesta View Post
I'm running into a character limit on the import program settings field. Would it be possible to remove/increase this limit in future versions?
Are you only trying to import preflop predefs?
If so, then these settings are contained in the file newdefs2.txt.
Copying this file from one program to another will exchange the ranges as well.

Should you be trying to exchange one of the other settings, then just leave the predefs out of the export format:

scylla is offline   Reply With Quote
Old 06-11-2017, 07:12 AM   #5842
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by MagicGog View Post
sent to
contact@cardrunnersev.com and support@cardrunnersev.com

One thing I wanted to ask you for several monthes but forgot each time. Here big problem with importing files from KO tournaments. It does not work. And I need to edit manually HH text by temoving some strings and transforming HH as it was freezout. It takes time, sometimes very annoying and often impossible cause of big tireness after play.

PokerStars Hand #171594414523: Tournament #1930916414, ($10+$30+$4) -> ($44) USD Hold'em No Limit - Level I (15/30) - 2017/06/10 18:30:35 MSK [2017/06/10 11:30:35 ET]
Table '1930916414 5' 9-max Seat #2 is the button
Seat 1: MilchoBG (4995 in chips, $30 bounty) -> (4995 in chip)
.........
and other players rows needed to be fixed manually.

Please look is it possible to fix in next release. Now so many KO tourneys
Ok, can you mail a sample HH to support please?
scylla is offline   Reply With Quote
Old 06-11-2017, 08:08 AM   #5843
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by MagicGog View Post
Ok, I have replied to your mail. There were a couple of attachments though, which sometimes trigger spam filters, so please let me know if you haven't received this e-mail.
scylla is offline   Reply With Quote
Old 06-11-2017, 09:35 AM   #5844
Siesta
stranger
 
Join Date: Dec 2015
Posts: 5
Re: CardRunnersEV

Quote:
Originally Posted by scylla View Post
Are you only trying to import preflop predefs?
If so, then these settings are contained in the file newdefs2.txt.
Copying this file from one program to another will exchange the ranges as well.
Thank you.
Siesta is offline   Reply With Quote
Old 06-11-2017, 11:13 AM   #5845
MagicGog
veteran
 
MagicGog's Avatar
 
Join Date: Aug 2007
Posts: 2,611
Re: CardRunnersEV

Received your email about tree, thx.

Quote:
Originally Posted by scylla View Post
Ok, can you mail a sample HH to support please?
Tomorrow will send... Sunday grind.
MagicGog is offline   Reply With Quote
Old 06-11-2017, 11:31 AM   #5846
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by MagicGog View Post
Tomorrow will send... Sunday grind.
No problem, there's no rush.
scylla is offline   Reply With Quote
Old 06-18-2017, 11:41 AM   #5847
x64asm
newbie
 
Join Date: Apr 2014
Posts: 24
Re: CardRunnersEV

Hello, I did following:
Define preflop and Flop ranges. Then I have created a node-lock for my opponent on flop. Because I know how he plays. I know that my opponent is not playing
perfect on turn and river so I canīt use the Nash vs. Nash solution from the solver. I know that my opponent is nerver raising on turn and he calls only with top-pair or better on turn. The subtree-wizard has created a tree with a turn-reraise branch.
So in this case do I need to remove the turn-reraise branch from the turn subtree? And do I need also to define that he calls only with top-pair or better on turn?
(Like in this screenshot: https://www2.pic-upload.de/img/33359079/crev.png ).

Or should I just define a strategy for the flop for my opponent and keep the trees from the subtree-wizatd and start then the solver? My plan would be then to remove the
reraise turn tree from my opponent after I have execute the solver and then I could check-out which hands are still +EV from my Nash solution.

Iīm dont know which method gives me the correct results in my situation.
x64asm is offline   Reply With Quote
Old 06-18-2017, 12:03 PM   #5848
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by x64asm View Post
Hello, I did following:
Define preflop and Flop ranges. Then I have created a node-lock for my opponent on flop. Because I know how he plays. I know that my opponent is not playing
perfect on turn and river so I canīt use the Nash vs. Nash solution from the solver. I know that my opponent is nerver raising on turn and he calls only with top-pair or better on turn. The subtree-wizard has created a tree with a turn-reraise branch.
So in this case do I need to remove the turn-reraise branch from the turn subtree?
Yes, if you feel that hero/villain will never take a certain action, then you can just remove that branch from the tree. With the branch removed, the solver will no longer consider this action.

Quote:
Originally Posted by x64asm View Post
And do I need also to define that he calls only with top-pair or better on turn?
(Like in this screenshot: https://www2.pic-upload.de/img/33359079/crev.png ).
Yes, this will work perfectly.
By doing this, you have defined play for all possible turn runouts.

Quote:
Originally Posted by x64asm View Post
Or should I just define a strategy for the flop for my opponent and keep the trees from the subtree-wizatd and start then the solver? My plan would be then to remove the reraise turn tree from my opponent after I have execute the solver and then I could check-out which hands are still +EV from my Nash solution.

Iīm dont know which method gives me the correct results in my situation.
You need to go with the first method here.
Just enter any play for villain that you see fit, lock it, and then run the solver.

Last edited by scylla; 06-18-2017 at 12:10 PM.
scylla is offline   Reply With Quote
Old 06-18-2017, 01:02 PM   #5849
x64asm
newbie
 
Join Date: Apr 2014
Posts: 24
Re: CardRunnersEV

Quote:
Originally Posted by scylla View Post
Yes, if you feel that hero/villain will never take a certain action, then you can just remove that branch from the tree. With the branch removed, the solver will no longer consider this action.



Yes, this will work perfectly.
By doing this, you have defined play for all possible turn runouts.



You need to go with the first method here.
Just enter any play for villain that you see fit, lock it, and then run the solver.
Thanks, that helps me already alot. But in the case where I dont know how my opponent would play on turn and river would you recommend to use just the Nash vs. Nash solution for turn and river with all trees?

Because in some spots I have not a real clue how my opponent would play so I can guess only in the blue but of course it would be good to know what the best thing is what I can do with the information that I have at this point.

I think in these situations I should just input that what I know and let the solver do the rest.
x64asm is offline   Reply With Quote
Old 06-19-2017, 04:51 AM   #5850
scylla
veteran
 
scylla's Avatar
 
Join Date: Mar 2008
Posts: 3,329
Re: CardRunnersEV

Quote:
Originally Posted by x64asm View Post
Thanks, that helps me already alot. But in the case where I dont know how my opponent would play on turn and river would you recommend to use just the Nash vs. Nash solution for turn and river with all trees?

Because in some spots I have not a real clue how my opponent would play so I can guess only in the blue but of course it would be good to know what the best thing is what I can do with the information that I have at this point.

I think in these situations I should just input that what I know and let the solver do the rest.
Yes, whenever you're not certain, just let the solver figure out GTO play instead. Particularly on unknown boards (49 possible cards for the turn and ~2500 turn+river runouts) I would generally recommend to just use the solver.

The problem is that, particuarly for river play, even if you feel strongly about villain's play, if you enter instructions that do not work for all possible ~2500 runouts, then the damage from this incorrect input will be much greater than assigning villain a GTO strategy instead.

An exception might be turn spots like you described in your previous post, where you're certain that villain would never raise in a certain spot, or only call bets with very strong hands. With "only" 49 possible turns, it's often possible to enter general instructions that work for all turns without risking incorrect input, provided that the instructions are not too complex or board-dependent.

Last edited by scylla; 06-19-2017 at 04:59 AM.
scylla is offline   Reply With Quote

Reply
      

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


Forum Jump


All times are GMT -4. The time now is 09:48 PM.


Powered by vBulletin®
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Search Engine Optimisation provided by DragonByte SEO v2.0.33 (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright Đ 2008-2010, Two Plus Two Interactive
 
Poker Players - Streaming Live Online