It's interesting how such subtle changes in solver input creates such drastic changes in strategy. Despite the somewhat "solved" nature of the game, it still boils down to reading your opponent. The old, pre solver soul reading days aren't dead yet it seems.
I'm probably dinosaur, but I can't figure out why the solve isn't an approximation of the game, because we still have to input things like betsizings and raise sizings and preflop ranges which are judgements. Also, the output isn't the highest value line against all strategies, but the highest value line against the best possible opponent (which we never should play).
Hello Tyler, enjoyed the analysis. These turn raise spots can pivot quite drastically on a few combos given that they are such narrow raising ranges at equilibrium. I think I saw that you recently started using Hand2Note - it would be good to look at how humans raise the turn data-wise in these spots vs how solvers play, so that we have a better idea how to react.
In the 2nd hand, when your opponent raised to larger that optimal sizing on the turn, I am never sure how to read this. For some players they are panicked with vulnerable value hands like you suggested in the video. Some players though will raise this massive sizing because they don't want to raise-fold a combo draw. If they raise smaller and get shoved on, they hate the idea of folding so much equity. Perhaps this is a question for Hand2Note too - comparing small raise size vs big raise size, and referencing each vs solver strategies.
Thanks Matt! I always look forward to your contributions.
It's hard to get a good read on this situation in the data, because it's reasonably rare and there's lot of different board textures represented in the turn-raise data along with several player types -- an RIO pro is going to raise differently than amateur who raises differently from somebody from say Nick Howard's stable.
The small and big raises both seem to be less bluff-heavy (though it's a small difference in data). The sweet spot for bluffs is around 3.2x though that doesn't necessarily mean you should call the sizing.
Thanks Matt! I always look forward to your contributions.
Thanks, very kind of you to say!
It's hard to get a good read on this situation in the data, because it's reasonably rare and there's lot of different board textures represented in the turn-raise data along with several player types
That's a good point. The board in question was extremely draw heavy and will of course be the reason behind the extra large raise size the player took. You can filter in H2N by board texture, but of course this will diminish your sample size significantly.
You can also filter by player type, in particular for this example you could use a Rec filter, a Reg filter, and a Reg filter with WWSF > X, where X is your boundary for what you would consider an aggressive reg. Again though, this would of course bring about sample size issues, especially if filtering for board texture too and would make it impossible without a massive hand sample.
The small and big raises both seem to be less bluff-heavy (though it's a small difference in data). The sweet spot for bluffs is around 3.2x though that doesn't necessarily mean you should call the sizing.
From the last sentence here I'm guessing none of the raise sizes are bluffing enough hands compared to a solver to warrant an over bluff? Are all sizes significantly under bluffed?
matlittle I don't think the bluff frequencies here over the last couple of years justify calling very lightly. The typical opponent c/r folds about 30% of the time and the solver is closer to 45% so that would seem to indicate an under bluff.
Tyler,
Enjoyed this video. Really like the way you approach the analysis of these hands and just how sensitive strategies are. You mentioned geometric sizing at one point during the 96s hand. Would enjoy seeing a video on geometric sizing in the future.
Loading 14 Comments...
It's interesting how such subtle changes in solver input creates such drastic changes in strategy. Despite the somewhat "solved" nature of the game, it still boils down to reading your opponent. The old, pre solver soul reading days aren't dead yet it seems.
I'm probably dinosaur, but I can't figure out why the solve isn't an approximation of the game, because we still have to input things like betsizings and raise sizings and preflop ranges which are judgements. Also, the output isn't the highest value line against all strategies, but the highest value line against the best possible opponent (which we never should play).
Nice video and concepts.
I think this concept probably contains something that can refute the phrase "poker is dead."
Poker is dead. Long live Poker!
Hello Tyler, enjoyed the analysis. These turn raise spots can pivot quite drastically on a few combos given that they are such narrow raising ranges at equilibrium. I think I saw that you recently started using Hand2Note - it would be good to look at how humans raise the turn data-wise in these spots vs how solvers play, so that we have a better idea how to react.
In the 2nd hand, when your opponent raised to larger that optimal sizing on the turn, I am never sure how to read this. For some players they are panicked with vulnerable value hands like you suggested in the video. Some players though will raise this massive sizing because they don't want to raise-fold a combo draw. If they raise smaller and get shoved on, they hate the idea of folding so much equity. Perhaps this is a question for Hand2Note too - comparing small raise size vs big raise size, and referencing each vs solver strategies.
Thanks Matt! I always look forward to your contributions.
It's hard to get a good read on this situation in the data, because it's reasonably rare and there's lot of different board textures represented in the turn-raise data along with several player types -- an RIO pro is going to raise differently than amateur who raises differently from somebody from say Nick Howard's stable.
The small and big raises both seem to be less bluff-heavy (though it's a small difference in data). The sweet spot for bluffs is around 3.2x though that doesn't necessarily mean you should call the sizing.
Thanks, very kind of you to say!
That's a good point. The board in question was extremely draw heavy and will of course be the reason behind the extra large raise size the player took. You can filter in H2N by board texture, but of course this will diminish your sample size significantly.
You can also filter by player type, in particular for this example you could use a Rec filter, a Reg filter, and a Reg filter with WWSF > X, where X is your boundary for what you would consider an aggressive reg. Again though, this would of course bring about sample size issues, especially if filtering for board texture too and would make it impossible without a massive hand sample.
From the last sentence here I'm guessing none of the raise sizes are bluffing enough hands compared to a solver to warrant an over bluff? Are all sizes significantly under bluffed?
matlittle I don't think the bluff frequencies here over the last couple of years justify calling very lightly. The typical opponent c/r folds about 30% of the time and the solver is closer to 45% so that would seem to indicate an under bluff.
The best part was the dining room table negotiation tactics. You should do a series on that, :)
Thanks Bob! I'm really happy, you enjoyed it. I'll consider your suggestion :).
Tyler,
Enjoyed this video. Really like the way you approach the analysis of these hands and just how sensitive strategies are. You mentioned geometric sizing at one point during the 96s hand. Would enjoy seeing a video on geometric sizing in the future.
Thanks Tyler
Thanks TripSevens! Glad you enjoyed it!
Hey Tyler once again great video man really appreciate it
Great to see written out your thought process when we’re facing this jam with 85cc
1 we want to have the nuts
Nuts are important hehe
So I guess we block set of 55 and 88
Villain can still have some jj or 66 97 74 i guess,
But still some overpairs maybe?
Interesting that j5 mixed fold given this jam
Oh wow villain hammed 87dd!
What a hero lol
Be the first to add a comment
You must upgrade your account to leave a comment.