I think the most important nodes to study are the ones that can generate the highest immediate EV gain for you. In other words, the weakest parts of your game. This can obviously be different for each player, but NLHE is particularly hard OOP, specially in high SPR scenarios. So I try to prioritize those situations. I would say turn is where I spend the most time. On the flop I mostly build simplified heuristics for approaching range construction. On the river I mostly study macro range tendencies. There are too many nodes and variables for river play so approaching with a micro perspective is kind of inefficient in my opinion. And then its on the turn where I try to get very accurate with my strategy, given that it will heavily impact my river play.
Different algorithms and most importantly different abstractions. Solvers can't compute the whole game tree from preflop so they use abstractions to reach a solution. For example, they group certain hands into categories by similarity and assume they will be played the same way, which reduces a lot the computation (monkersolver). Or they pick a subset of flops instead of all flops to reduce game tree size (pio solver). At the end of the day its very hard to know which one is the most accurate in specific situations.
Really surprised Qd10c checks at 100% at 37:00. I get that it does a really bad job at generating immediate folds on turn but it seems like it would be a good river bluff as double barrel. Unblocking spades, folding out Ace highs that peeled turn, and our exact suits block the Q6s and 106s combos that are high frequency checks on flop. You think it checks purely cause it does so poorly on turn and we have so many bluffs available? Like do you think this hand might bluff more against a 3x open where our range is a bit more narrow and we dont have as many hands to bluff with?
Blocks turn folds, blocks river folds from floats and can't valuebet made pairs profitably because outpipped by the highcard hands that do float. Basically no river that is great for our hands ev
Good, detailed analysis though I'm not crazy about the format. In particular these two hands ended up being pretty low frequency spots. The JJx hand because villain should have cbet so frequently on flop as you touched on. The 865 because we should have a substantial leading range which I don't recall you touching on.
So I think I'm of the opinion that I would have liked to see a bit more common spots and possibly a little less depth of analysis in favor of a more spots.
Usually quite enjoy your videos, but this one was painfully slow. Far too much time spent delving into the minutiae of each range, street, and action frequency. By your own admission the spot in the first example is encountered extremely infrequently, yet it took more than a third of the video (15 minutes) to complete.
It seems to me what differentiates these types of software from PIO, and makes them so great, is the ability to quickly and effectively practice a multitude of spots without having to invest exorbitant amounts of time on each individual spot or getting overly bogged down in data.
I enjoy this video format in general, but would rather see you get through many more spots in a 45 minute video as that's really what this type of software is for in my opinion.
Hi Saulo, very interesting video.
Is there a Program or a tips to build all aggregated report for all game trees actions as you showed (1:30), or everything was done manually?
Thanks for your answer!
Loading 14 Comments...
what nodes do you think are the most important to study? how do you split your time between flop/turn/river OOP/IP when working with simplegto.
I think the most important nodes to study are the ones that can generate the highest immediate EV gain for you. In other words, the weakest parts of your game. This can obviously be different for each player, but NLHE is particularly hard OOP, specially in high SPR scenarios. So I try to prioritize those situations. I would say turn is where I spend the most time. On the flop I mostly build simplified heuristics for approaching range construction. On the river I mostly study macro range tendencies. There are too many nodes and variables for river play so approaching with a micro perspective is kind of inefficient in my opinion. And then its on the turn where I try to get very accurate with my strategy, given that it will heavily impact my river play.
5:15 Do you have an idea why different solvers produce different results?
Different algorithms and most importantly different abstractions. Solvers can't compute the whole game tree from preflop so they use abstractions to reach a solution. For example, they group certain hands into categories by similarity and assume they will be played the same way, which reduces a lot the computation (monkersolver). Or they pick a subset of flops instead of all flops to reduce game tree size (pio solver). At the end of the day its very hard to know which one is the most accurate in specific situations.
Makes a lot of sense. Thanks
Really surprised Qd10c checks at 100% at 37:00. I get that it does a really bad job at generating immediate folds on turn but it seems like it would be a good river bluff as double barrel. Unblocking spades, folding out Ace highs that peeled turn, and our exact suits block the Q6s and 106s combos that are high frequency checks on flop. You think it checks purely cause it does so poorly on turn and we have so many bluffs available? Like do you think this hand might bluff more against a 3x open where our range is a bit more narrow and we dont have as many hands to bluff with?
Blocks turn folds, blocks river folds from floats and can't valuebet made pairs profitably because outpipped by the highcard hands that do float. Basically no river that is great for our hands ev
Good, detailed analysis though I'm not crazy about the format. In particular these two hands ended up being pretty low frequency spots. The JJx hand because villain should have cbet so frequently on flop as you touched on. The 865 because we should have a substantial leading range which I don't recall you touching on.
So I think I'm of the opinion that I would have liked to see a bit more common spots and possibly a little less depth of analysis in favor of a more spots.
Great video format, but rather see the high freq spots than the first one you showed. Skip those extra rare spots in the future :) Thank you!
26:20 sauloCosta10 can you explain the micro and macro you are referring to? I am not used to these terms for poker.
I think when he says macro he means overall range frequencies and micro is referring to specific holding frequencies
Thank you
Usually quite enjoy your videos, but this one was painfully slow. Far too much time spent delving into the minutiae of each range, street, and action frequency. By your own admission the spot in the first example is encountered extremely infrequently, yet it took more than a third of the video (15 minutes) to complete.
It seems to me what differentiates these types of software from PIO, and makes them so great, is the ability to quickly and effectively practice a multitude of spots without having to invest exorbitant amounts of time on each individual spot or getting overly bogged down in data.
I enjoy this video format in general, but would rather see you get through many more spots in a 45 minute video as that's really what this type of software is for in my opinion.
Hi Saulo, very interesting video.
Is there a Program or a tips to build all aggregated report for all game trees actions as you showed (1:30), or everything was done manually?
Thanks for your answer!
Be the first to add a comment
You must upgrade your account to leave a comment.