fishcheckmate's avatar

fishcheckmate

75 points

The only postflop solver that can account for the bunching effect is Jesolver, through scripting, and it's very clunky. Other postflop solvers simply ignore the bunching effect.

Yes, this effect can be significant, especially in late position battles. If you compare BvB 3max vs 9max the opening and defending ranges can change upwards of 5% tighter or wider, and postflop strategy would change too.

When you copy preflop strategies into a postflop solver, you lose two key pieces of information:
1) The ranges have changed. The strategy can be identical, but the distribution of cards within your opening range are not the same. Your postflop solver thinks every hand is in range with equal probability, but in reality, your range is more top-heavy.

Here's an example from that visualizing the bunching effect article posted earlier. Let's say you play 9max, folds to SB who (for some reason) opens any two cards. You copy that into a postflop solver and it thinks you have everything at an equal frequency. In reality, your range looks more like this:

2) The distribution of postflop cards changes, which skews your equity/EV and the frequency of runouts. Sometimes it's shifted towards high cards or lower cards, and in some cases it mostly events out depending on preflop actions.

You should check out these 2+2 posts for more info:
The Bunching Effect
The absurdity of high resolution solutions - bunching effect

May 13, 2022 | 12:08 a.m.

This is really common, don't feel bad. Rivers do involve a lot of guesswork, however the root source of your problem is most likely poor range construction on earlier streets. Here's an example:

You rangebet flop, because screw it that strategy is easy and it works.

Then you barrel the turn with all of your half-decent value and draws and probably some pure air (not realizing that you need to check back like 75%+ of your turn range because you rangebet flop and villain doesn't XR enough).

You brick the river and end up with a range of like 20% value 80% garbage/overplayed medium hands. You might feel like you constantly have to give up on the river or overbluff. But the underlying problem isn't the river, it's your bad flop/turn construction. If you could see your actual river range it would be immediately obvious what the problem was.

I'm not saying this is your strategy, it's just an example of how mixing mistakes compound.

March 13, 2022 | 12:44 a.m.

Interesting line. I checked this spot in GTO Wizard and it likes simply shoving with JJ facing this large flop cbet. This puts a lot of their range in a tough spot, like AK, flushdraws, A7s, 88, etc. You could also just call and take a turn. But the minclick isn't very effective against the hands you're targeting imo.


Feb. 25, 2022 | 1:24 a.m.

This is what you call "solver noise". The only way to eliminate that noise is to solve to perfect accuracy (0 dEV).

The size of the mistake is relative to the size of the pot. In this case, it's making a -0.02bb mistake out of a 50bb pot, which amounts to a 0.04% mistake. That's very close to 0 and almost certainly below your accuracy threshold.

Anytime you see a hand mix between two actions, those actions should (theoretically) have exactly the same expected value - this is a hard law of Nash Equilibrium.

Feb. 4, 2022 | 3:23 a.m.

I agree with Kalupso; the main exploit here is to size up.

Against these wide blinds your steal% decreases, so ~0EV hands which rely on fold equity to be profitable become losing opens.

The main thing preventing you from opening larger is the threat of getting raised. If they're too passive that means you can implement a larger size to overcharge your value against the stations, without worrying about getting exploited by 3bets.

This video explores preflop exploits in more detail: https://youtu.be/MPae2gqkeRw

Jan. 25, 2022 | 3:09 a.m.

Keep in mind a lot of them are just using solver libraries to train. For example, GTO Wizard has MTT and HU SNG solutions for every spot. No need to solve everything yourself.

Jan. 19, 2022 | 9:15 p.m.

Keep in mind that some of your calls are supposed to be losing. You call with hands that lose less than folding.

Here's an example. BTN opens 2.5, SB 3bets to 12. The EV of 77 is 0.34. That means 0.34bb better than folding. Since folding is -2.5bb (relative to your stack), that means calling is -2.16bb. That's still better than folding, but in your HUD it will show up as a loss.

In other words, you expect to lose 216 bb/100 calling this hand, but you should still call since folding loses 250 bb/100.

Jan. 19, 2022 | 9:05 p.m.

If you only have two actions, e.g. raise/fold, then you are linear by default. You simply raise the best hands and fold the rest. If your calling range is very tight (e.g. SB facing IP open, or BTN vs HJ), then you're still extremely close to linear.

The only time you get to polarize is when you have a calling range. So BB vs SB is typically extremely polar. BB vs IP is mixed but can become more/less polar depending on your sizing and rake structure.

Jan. 19, 2022 | 7:12 p.m.

Bet sizing is very nuanced, you might need more specific heuristics.

GTO Wizard recently released this video on cbet sizing:

https://youtu.be/dVZ1CdESSTw

Jan. 15, 2022 | 6:07 a.m.

Jan. 8, 2022 | 9:13 p.m.

This is the problem with using raw equity to calculate EV. The hand doesn't get checked down, you might win/lose more on later streets.

If your pot odds tell you that you need 25% equity to call, what that really means is you need to win 25% of the pot. That means accounting for all the postflop variables, including your implied odds. In order to justify putting 5 more chips into the pot, you need to win at least 5 more chips on average.

Jan. 5, 2022 | 11:56 p.m.

Comment | fishcheckmate commented on HREC to PIO

Stop spamming, nowonasperholdem.

Dec. 22, 2021 | 6:50 a.m.

Why not just use a subscription service like GTO Wizard to gain access to every spot?

The Pio pro version is limited to 16 cores as far as I know. You'll wanna upgrade to Edge and contact the dev directly.

Dec. 22, 2021 | 6:48 a.m.

Comment | fishcheckmate commented on Study on MW-pots

They will evolve linearly if you expand the ranges. Try (JJ+, AQs+ AK).

The reason it doesn't evolve linearly, in this case, is because the combinatorics force really funny scenarios like QQ being up against 4 players with AK 1/3 of the time.

Dec. 20, 2021 | 3:02 p.m.

Comment | fishcheckmate commented on Study on MW-pots

It's a fascinating puzzle because combinatorics limit the possibilities.

There are only 12 cards in the deck containing a Q K or A. You've got 6 players with two cards each holding Q/K/A. Therefore, the deck cannot contain any more queens kings or aces. There are only three possible scenarios, three ways that use up all 12 cards with the range QQ+, AK:

If you hold QQ, 1/3 of the time you end up in Scenario 1, which means you have fantastic chop equity with the other QQ. You can't actually win unless you make a straight flush, but you can chop with the other QQ a lot.

If you hold KK, you're in Scenario 2 or 3. It's literally impossible for you to win. Someone else MUST have AA, so you need to make a straight/flush/trips/boat or something to win. You can't make a straight because all the queens are gone, you cant make trips/boats because all the kings are gone, you can't make a winning flush because another player always has a better ace. The best you can hope for is a 6-way chop on a low straight runout, something like 56789. Kings literally cannot win the pot.

If you hold AK, you're in Scenario 1 or 2. you can win with a flush, or chop with another AK on a wheel straight. But unless you have a suited AK you're mostly drawing dead.

Really interesting scenario! The limited ranges make the combinatorics very direct.

Dec. 20, 2021 | 12:52 p.m.

Comment | fishcheckmate commented on Study on MW-pots

Thanks for this, it's actually a fascinating mathematical puzzle. I don't want to spoil the answer just yet though.

Think about the implications of 6 players holding (QQ+, AK). Is it possible for a 7th player to also have (QQ+, AK)?

Dec. 20, 2021 | 7:42 a.m.

People underestimate just how sensitive the BB calling range is to pricing.

Check out the "General 2.5" solutions for comparison. Here's your defence against a 2.5bb open, where UTG is opening virtually the same range:

The next natural question to ask is "why does UTG get to open the same range, but I have to defend tighter, that doesn't feel fair!"

UTG tends to face a 3bet more often than they face a call when using this size. They took an extra risk opening 2.5, which they pay for by losing more when folding against a 3bet. After it folds to you in the BB, that risk has paid off, and you can't call as wide.

Opening 2bb is a safer option for UTG, risking less against a 3bet. But it also allows BB to call much wider.

To put things into perspective, here's BB's response vs a 43% range 2bb BTN open (rakeless). As you can see, this price lets BB call ridiculously wide lol

Dec. 4, 2021 | 5:06 a.m.

And prf raise from ep/mp kto/qjo but no 76s87s ?

When you open UTG, you're more likely to face a 3bet than you are to face a call (in GTO), so the solver chooses hands that have stronger blockers against 3bets, such as KJo, K6s, QJo etc.

If you get rid of the rake and open to a small size then ppl call more, and the solver starts to prefer hands that do better when called like 76s etc.

What do you think about 3bet ranges from BB 3bet vs every position? Why it prefers 3b f.e vs ep/mp AQo /Kqo then AQs /Kqs

Huh? I think you're asking why EP folds the offsuit hands against a 3bet?

AQ/KQ are both very near breakeven against villain's 3bet range. The suited hands realize more equity - they can more easily continue under pressure as they can draw to a flush. This improves implied odds too. Additionally, if you call all your offsuit hands, then your range becomes more exploitable on many boards.

Oct. 8, 2021 | 6:57 a.m.

The evidence is very statistically significant and reported from multiple sources. I've read that multi-tablers tend to get underdealt CO spots, whereas people who single table tend to be overdealt those spots, likely due to a flaw in their seating script.

If you play on party, I encourage you to post your total hands by position. Even if your sample is small, the smart folks here can run simple statistical analysis to see if it's significant.

Sept. 18, 2021 | 1:47 a.m.

You can calculate it manually with the equities. But just looking it up:

AKhh is indifferent, and AKdd is a -1.71 call, (out of a total 201bb pot). In terms of human-level accuracy, these are all indifferent.

Sept. 12, 2021 | 7:25 p.m.

I'm just wondering why in this solution AdKd is a pure fold where as AhKh is mixed? Is it because AhKh blocks more AQs and KQs combos.

Go into their range, filter for all in, change from "strategy" to "range".

Now filter by suit, then tally how many bluffs and value combos you block with each suit of AK.

AKhh unblocks KThh bluffs and blocks AQhh value. They both block the same amount of KJ. It's extremely marginal though, only about a 1% equity gain for hearts.

--

The above process is pretty much the universal procedure to "why is X blocker preferred" kind of questions.

Sept. 11, 2021 | 12:18 a.m.

Post villain's shoving range

Sept. 9, 2021 | 11:10 p.m.

I don't think it's a good call in practice. 4bets IP are too overpair-heavy in reality.

Facing the flop shove you have a pure bluff-catcher. Do you think half of villain's shoving range is ace-high garbage? Because that's about how often they need to be shoving A-high garbage for your call to make sense. Even at midstakes I seriously doubt it.

Aug. 15, 2021 | 5:05 a.m.

Comment | fishcheckmate commented on 50z JJ squeezing

Seems fine. You raised over an iso so technically not a squeeze, but still a fine hand to raise with. cbetting flop is reasonable. The turn strat depends on the context.

Aug. 11, 2021 | 3:46 a.m.

Aug. 11, 2021 | 3:29 a.m.

Don't reinvent the wheel, there are tons of free, high-quality solved preflop ranges out there. +1 for free GTO Wizard ranges.

There's nothing wrong with trying to invent your own stuff, but that 3b range is horrendous. You're completely overbluffing. You're mixing in a ton of garbage that shouldnt even be played, medium hands that don't belong in a super polarized range, and not enough value in general.

Aug. 10, 2021 | 7:48 a.m.

The main factors to consider are position and linearity.
If you're out of position, anything less than a 75% raise is almost always a poor choice. If you're in position, anything larger than a pot-sized raise is almost always a poor choice.

Use smaller sizes when your range is linear. If you only raise/fold from a certain spot, then you'd use a smaller size as your range is top-down and you want to get called by worse. If you're using something polarized, then you're aiming for fold equity and need to size up.

For example, modern preflop theory shows that in most raked games, 100bb deep, the optimal 3bet size in the BB vs BTN 2.5bb open is about 12-14bb, as we're out of position and are supposed to go polar. However, something like a HJ 3bet vs a 2.5bb UTG open is gonna be closer to 7bb, as we have position and a purely linear range.

3x IP and 4-5x OOP is a good enough rule of thumb if you want something simple.

Aug. 10, 2021 | 3:38 a.m.

This is always going to come down to extremely delicate blocker effects. Observe which value/bluffs you block and unblock in SB's river betting range with the Ad.

Don't get too hung up on this. The exact strategy by the river is gonna be hyper-sensitive to your inputs, your tree, the rake structure, the assigned preflop ranges, the possible raise sizes, etc.

Here's a high-resolution sim from GTO wizard 50NL solutions, which have between 6-19 sizes per node. This sim assumes SB opened for a normal 3bb size and not the funky 2.5bb size.

All you need to take away from this sim is that A7o is indifferent - a 0EV bluff catcher.

Aug. 8, 2021 | 7:31 p.m.

Comment | fishcheckmate commented on playing deep

Flatting JJ is fine 200bb deep, can also 4b.

As played, go for more value in general.

As a rule of thumb, It is almost never correct to quarter-pot the river in position - halfpot minimum and usually higher.

Aug. 1, 2021 | 4:36 p.m.

this kind of bet,2 into 100,how will u guys deal with it?

Just treat these microbets as a check. Call hands you would normally check, and raise hands you would normally bet.

Aug. 1, 2021 | 4:29 p.m.

Load more
Runitonce.com uses cookies to give you the best experience. Learn more about our Cookie Policy