jesolver will be 20% off as a special xmas gift for 10 hours

(hopefully it's ok to post here piotr?)

It's ok but make it longer than 10 hours, c'mon

ok 15 hours sealed with a kiss

Hi,

Unfortunately the Christmas sales was a bad idea and I am taking them down. The reason is that it creates resentment among people who are already our customers and I guess for a good reason.
Here is what is going to happen:

1)everyone who I promised a discount and/or who contacted me before about it will get it
2)everyone who was about to purchase it but didn't have time yet will get the discount as well as long as it happens in the next 2-3 days (just send an email or contact me on Skype)
3)everyone who already bought it - of course keeps it

I am yet to learn about business and I haven't expected such a reaction from existing customers. I guess they have a point.

Our policy in the future is going to be the same as it was before:
1)discounts for people who buy the software in groups
2)no affiliates (we will not have any nor are we going to act as one)
3)no random sales periods

Apologies for the confusion. I've learnt my lesson on this one.

Hello,

We've just released an update for a preflop solver featuring:
-faster solving times (20%-35% depending on CPU)
-many fixes for the preflop building interface as well as browsing
-optimizations for small ranges

More here:
http://piosolver.myshopify.com/blogs...solver-updated

There will be more work going into it in Jan/Feb 2016 as the solver could still be faster/more precise. A video describing some more advanced preflop tree building will appear in coming days as well.

Have fun!

Is there a way if I had time to find the ev/eq for every hole card combo across all flops?

I know we can aggregate actions at nodes but I want to know if it is possible to do it for ev/eq.

When you run an multiflop aggregation report there are now 4 files being produced:
-handsEV.csv which contains EVs, Equity, Equity Realization (EV/EQ) and matchups (probability)
-report.csv which contains a traditional aggregation report
-ipPreflopOrder.txt and oopPreflopOrder.txt which contains all the combos from given ranges sorted by EV (in copy-pastable format so you can use that order in the range selector).

I hope the first of those files is what you need.

hey punter,

you've made a bunch of big updates since i've been using piosolver regularly, i'd like to know if the following feature has been implemented:

Say I solve a situation to a high degree of accuracy, and notice that the optimal OOP strategy donks 15% of the time. I'd like to determine how much a player would lose by never donking, but playing perfectly given that constraint.

That is, I want to eliminate donking from OOP strategy, re-solve, and then see the difference between the value of the game to OOP when he's allowed to donk vs when he's not.

Can I do this?

If so, how about this: Eliminate the strategy of "waiting until the turn" with a strong hand, and do the same as the above? This is tougher, because the concept of "waiting until the turn" is not as well-defined as donking. But you might say something crude like this, which would be a good approximation: The strategy is delaying a raise or check raise until the turn if it raises on more than 50% of turn cards. Alternatively, if it raises when it's equity has not changed by much. Is there any way to do calculations like this in piosolver?

To your first question yes you can

Solve with OOP being able to bet then resolve it without OOP being able to bet first. Compare the 2 evs (subtract one from other and x100) to get a bb/100 lost for the spot.

With regards to the second one it's not so easy but should be possible with node locking. I would suggest watching the free video on pio solver you tube channel which shows how to node lock

In the 1.7 version you will be able to combo-lock nodes so you will be able to lock "strong hands" to check-raise the flop (and leave rest of the node to the solver).
It is doable in 1.6 as well but unfortunately requires using solver console directly so it's better to wait for 1.7 and it will be part of the graphical interface.

Cool, thanks for the update.

What will be the options for selecting "strong hands"? Will I have to do that manually, or will I be able to select using a condition such as "hands with at least X% equity"?

Speaking of which, is there a way to view strategy and equity at the same time? Eg, show the equity number in the corner of each square while looking at the strategy view? Or, while looking at a single strategy view such as "call", which currently shows each square in yellow, to use a heatmap to indicate equity? I think you can already do this with the range explorer (correct me if i'm wrong), but it would be nice to just check a box or click a button and have the normal browser mode view change.

It's true that in a game of chicken there are 2 Nash equilibria that have different EVs for each player. The NEs are Player 1 swerve 0% Player 2 swerve 100% and Player 1 swerve 100% Player 2 swerve 0%. But the game of chicken is not a turn based game. Consider the following pokerfied chicken based toy game, the rules are as follows:

1 street
if the hand reaches showdown, the pot is always split 50:50.
the starting pot is 10bb
effective stacks are 100bb
the only legal bet size is 100bb
rake is 0% for any pot under 150bb
rake is 100% for any pot 150bb or greater.


There are 3 Nash equilibria in this game. They are as follows: (X = check, B = bet, C = call, numbers are the %s)



However, only one of these Nash equilibria is stable.

With strategy pair 1, if you change p1's betting % to anything above 0, algorithms designed to find Nash equilibria will not converge back to it. Assume P1 now bets 0.1% say and p2 calls 100%. In the next iteration it may look like 0.05% bet, call 50%. But here the EV of P2's call is more negative than the EV of P1s shove. So it will begin to converge faster to 0. Eventually the shove of p1 will become positive. The same applies to strategy pair 2. The only strategy this doesn't happen for is strategy pair 3, where p1 open shoves always. This is the only stable strategy - any slight change (in fact any change other than assigning 0% or 100% to the values) will always result in it converging back to the strategy.

My guess is that this is the case for poker in general in tournament scenarios. Only 1 will be stable and this will be the one used in real life scenarios and this is the only one that will be found by algorithms searching for NEs.

Games that this is relevant to are tournament games and games with rake. I am hopeful that one day you will allow piosolver to be used for tournament calculations. I think it's worth working on, it would attract many new people to piosolver (it's currently not useful at all for later stages in tournaments). If not then at least I hope you change PioViewer in a way that would allow other engine developers to do this kind of thing.

How will this work?

Will we be able to say stuff like always check raise flush draws, straights, etc, like the type of hand?

I have to admit, I have been missing and waiting for something like that, because so often when we get to the turn we do not know what type of hand we have unless the turn card is specific (so it won't help for instance to solve on the flop when we may know that opponent always barrels flush cards etc)

I would expect the average buyer of this product to be a bit more rational

I don't think there are many 'irl' situations where you:

1) buy a product and are happy
2) the product goes on sale
3) you go complain to the store because you didn't get to take advantage of the sale

No conditions for now but you can always use range explorer to choose the hands you want and copy paste to the node locking interface.

Only Strategy + EV at this time (which you can get by clicking strategy + ev button) which is displayed as a mix of numbers (EV) and colors (strategy).

For now the only way to select hands will be the same as current node-locking (that is per combo basis). Again, you can choose the whole groups using range explorer and copy-paste there. This will be improved some day but not in 1.7 release.

Here we are talking about specific case of locking flop combos to always check-raise. Locking of many nodes using some kind of language to describe general conditions is a very cool feature but sadly won't make it to 1.7 release

I thought the same but I kinda see the difference between physical products and digital ones here.
The thing is we did well and I was willing to offer discounts during limited time for people who weren't able to pay the whole price. I wasn't willing to adjust all the transactions from before though (as that would be equivalent to lowering prices for like the whole month and that I wasn't willing to offer).
Unfortunately it turns out you can't have both without causing a lot of resentment so there is a lesson for me.

When i make a small save and open that later, node locking option tells me "setting and locking strategy doesnt make sense on incomplete tree" , proceed anyway?
I get that i must have a tree built in order to calculate again, so how do i copy one from the save
How do i use node locking for study when dealing with small saves?

https://www.biblegateway.com/passage...20&version=NET

http://forumserver.twoplustwo.com/pr...gnore&u=305529

For what it's worth I think the customers I've mentioned are right and the sales idea doesn't make the slightest sense in software business. I apologize once again for resentment/hurt feelings/anger it caused. The case is now closed.

Do the various GTO program software return the same calculations for the same queries or do they come to different conclusions about ideal play?

For the very small sample of spots where I compared, they seem to arrive at the same results.

It looks like I managed to mess up postflop browsing on preflop trees. This will be fixed (along with PioViewer interface to go to postflop) in the January update.
It's the browsing that doesn't work, the solving process is ok. Thanks to all the people who helped me with useful bug reports.

Same more or less although even at seemingly low accuracies strategies for specific combos may vary a lot. Now when there is Oskari's solver and we have capabilities to run solutions to 0.01% (for now in a dev version) as well I am comparing a lot of trees and the solutions are almost exactly the same (usually the strategies differ at 3rd decimal place a bit).

Is this publicly available, and if so, where?

edit:

how do i modify individual hands only? seems like in some instances when i try to node lock there are multiple runouts attached to a single node, not that that is a big deal or anything.

Is it possible to pay for product in Skrill?

Sent from my GT-I9300 using 2+2 Forums