thanks a lot!May seem dumb question but it needs something different
to work for plo?it starts for holdem but it seems stuck again if i try a plo sim..

I use the same settings, I have no issue starting either HE or PLO.

Nvm they already replied. Very fast.

I downloaded the software and created an account(Got the activation code on email). But when I log in I get the "log in failed" error. I have tried to tweak my windows defender, but it does not help. OP please PM me.

Have you tried disabling all firewalls? I believe Windows Firewall for example is separate from Windows defender. When last I've had this issue, it was due to firewall issues I believe.

Hi. I have question about calculation strategy on flop only for PLO. Is it possible or we need to calc all 3 streets?

For example: I put sizings only for flop and exclude preflop, turn and river, and don't add sizings for turn and river. the strategy that he calcs is right or not?

Your flop strategy depends on what happened pre flop, as well as how you will play the various turn/river runouts. So it doesn't make sense to calculate flop play in isolation. You need pre flop ranges (that you can guesstimate or get from a pre flop solution) to know what the players start out with, and you need to give the players bet sizes for turn/river. Every time you solve the game on some street, you are necessarily solving for play on that street and all future streets. And to get started with a flop/turn/river solution, you need ranges carried over from the previous street (guesstimated or solved).

Quick'n dirty logic to show why this is so:

Let's say you eliminate all turn/river betting. Now calling a flop bet is the same as calling all-in (as in, you can call a pot bet with any hand that has 33% equity and you'll get to showdown with it and realize all equity). In the real game there's future bets to think about, and we often have to fold 33% equity hands on the flop.

Ok. what you can recomend to find the most optimal betsize or two betsizings very fast?

Because if I create with all sizings tree can be more than 100-200gb ram

Run cheap sims (reduced accuracy) to screen for preferred sizings. Choose a few reasonable options and see what solver prefers. Then run a more accurate sim with the size(s) you end up using.

For all kinds of modeling, don't chase perfection. You can't execute perfect strategy anyway, so it easy to waste a lot of study/solver time that way.

What differences are between MonkerSolver and other GTO solvers in the market like Pio? Or even GTO+ from CREV?

MonkerSolver handles both HU and multiway scenarios, for Hold'em/Omaha/Omaha8.

I have done some testing and have two grids for those who wonder how Monker compares to Pio (keep in mind, we can only compare HU strategies, since that's what Pio can solve). Below you see BB's shove/call/fold strategy vs BTN min raise with 25bb stacks. The BTN open range was computed in a high accuracy MonkerSolver 3-way simulation (abstractions: 120 buckets perfect/perfect/large with convergence to 75 iterations/node) and then I ran a high accuracy Pio pre flop simulation (198 flops, converged to 2 bb/100) fixing BTN to that openrange to see how the Pio BB strategy compared to the MonkerSolver grid.

The strategies are identical for all practical purposes with shove/call/fold frequencies 14/62/24 for Monker and 14/61/25 for Pio. Since Monker and Pio agree very well on HU strategy, it's seems fair to assume that the Monker algorithm performs well. For the multiway solutions we have nothing to compare MonkerSolver to (yet), but other solvers might give us something to compare with in the future.

MonkerSolver BB vs BTN minraise (25bb)


Pio BB vs BTN minraise (25bb) - BTN on MonkerSolver 3-way open range



People with questions about performance, what can be calculated, how to calculate it, where to rent servers, etc, are welcome to join the MonkerSolver Hangout Skype group (unofficial group where we figure things out and answer each others questions).

When you know when to stop a sim?

Experience helps. If relevant ranges are still changing, keep it running. For pre flop work, In pick a couple ranges a little deep in the tree (for example, a BB overcall range vs opener + caller) and monitor how many % change you see per completed iterations/node).

Like so (I/N to the left, range % to the right):

10.0 13.6%
11.3 15.4% (change: 1.8/1.3 = 1.4 % per I/N)
.
.
.

Then you stop when the change crawls to a halt. But you'd like to see it slowing down and approach zero. For example, if you're looking at a range changing 0.3% per I/N consistently for 10 consecutive measurements, it's not converged yet, it's just converging slow.

Hi,
I made an ICM monker simulation SBvsBB (40 iterations). Seems to be not accurate, because I took the SB openjamingrange of the monkersolution, put this into hrc and compared the nashcallingrange of the BB with the monkercallingrange. HRC tells me to call 44+, A9s/ATo and Monker wants to call 66+, ATs/AJo. Doublechecked stacksizes/payouts - how is that possible?

Hi. Can you explain what means small and large method on the river with some example?
Is it possible to study with only small method?

Recommended number of nodes/iterations for solving Hu Plo preflop spots?

I haven't done HU PLO sims, but in NLHE, the default settings are pretty decent. To reproduce a very accurate Pio simulation, you'd want to up it to something like 120 buckets/perfect/large/large, or 30 buckets/perfect/perfect/large.

If perfect/perfect/large is too big for HU PLO, go with perfect/large/large and as many buckets as you can fit in. Go with 100+ iterations unless it's very slow (can settle for 50 then). Assuming you're after quality solutions that you only do once.

Ended up doing 30 buckets perfect/large/large. Going for 100 iterations. Will this be sufficient or am I better off restarting with 120 perfect/large/large? (Couldn't fit perfect/perfect/large)

If you can manage 120/perfect/large/large, go for it, imo. You'll redo it later anyway, and you know it. ;-)
Don't skimp on the buckets for solutions that you will use a lot for study work.

Well, I can't fit 120's across the board. Which street would be most important to allow max buckets? I can do 70,70,70 or 120,70,65 F,T,R

Also have the option to just buy max CPU power and go all out. Should probably just go w that.

Not sure how buckets are best distributed, and I just pick the same number for all streets. Agreed on maxing computer specs (or going to cloud) if you plan to run a lot of sims. $200/month gets a dedicated server with 20 cores and 256 GB RAM at Contabo.

I just purchased the full version but now i cannot login.It says log in fail.It even shows log in failed even if i restart my pc without trying again to login

hi, we would like to ask for custom service, do you support?

I've never used Contabo before. Any recommendations on specs outside of 256gb Ram? It looks substantially cheaper than AWS 128's.

wow nice to know. i've been paying ~270(incl. windows license) for 16c 128gb @ ovh. how much is a windows license on contabo or is it included?