Kalupso's avatar


2687 points

I also use simple preflop and it's a very useful tool. I don't disagree that it's possible to build trees that require 120gb+ RAM to run and takes a week to run but I have not found any practical advantages in application over full 6max sims that require around 48gb RAM to run and takes 12 hours to run to decent accuracy. The biggest I have run was around 200gb on a 48 core server that ran for a little more than 12 hours. We compared that one to one with heavier abstractions and there was no practical difference, and we always need to translate the solver outputs into something practical for humans. More accurate and less noisy sims do make the translation easier but a deep understanding of how the spot works can overcome the need for accuracy. With all the knowledge I have now I could easily just use PokerSnowie as reference for thresholds and come up with close to as good strategies as if I had the best preflop solver sims available.

Adding calls is very expensive and I suspect the software doesn't handle MW postlfop close to perfectly. Sims can become very large if we add limps or tons of overcalls so there are awkward scenarios to sim. I don't deny that.

P.S. I mostly jumped in because I thought the information overall here is misleading. There are cheap preflop solver options when comparing to Rangeconverter and it doesn't require a huge PC to solve most problems. Most of my pracitcal sims takes less than 20min to run and uses between 1gb and 10gb RAM (2, 3 or 4 active players). Simple preflop is 350 before tax so not exactly cheap for a recreational reg but very high value for a pro.

Dec. 2, 2019 | 1:07 a.m.

I am very hard on myself for my mistakes

You can start with that one. Be more forgiving. Maybe see yourself in 3rd person as if you were a good friend. Treat yourself the way you would treat a good friend.

Dec. 1, 2019 | 11:28 a.m.

Rangeconverter is very overpriced. Solving time with a modern preflop solver is much quicker than with the first generation of preflop solvers and has much lower CPU and RAM requirements. The issue with buying a single solution is that you only see the correct strategy under a single set of conditions. In my opinion, it's much more useful to run many lower quality simulations and see how it adjusts and learn concepts about how the equilibrium works. Of course, getting a baseline from a single solution is very useful but Snowie already has a very good baseline available for free but isn't accurate enough to give information about how to structure ranges.

This solution I can build and solve in 20 minutes to a practical accuracy level but they sell it for 500$: Overpriced!

I know they pushed accuracy to the max but there is no practical benefit of increasing it so much. You can buy a preflop solver for less than 400$ and solve it on a PC with less than 6 Gb ram in less than an hour from purchase to completion.

Dec. 1, 2019 | 11:09 a.m.

Honestly, the biggest issue is that the books are too abstract and removed from what is important to think about at the tables and when trying to improve. You won't necessarily derive strong strategies if you follow the approaches outlined in Applications because it's only directly applicable to river play. The "straight-up wrong applications of theory" stems from that fundamental problem with the book (theory only works for river or perfectly polar flop/turn). Will Tipton focuses on equilibrium ideas and maximizing EV of your strategy at equilibrium and with exploits. Maximizing EV and equilibrium ideas are always relevant and are what a solver is doing.

Dec. 1, 2019 | 10:54 a.m.

I think these can get too abstract or too far removed from application to be practical, and there are straight-up wrong applications of theory in "Applications of No Limit Holdem" (proven wrong by solvers):

  • "Mathematics of Poker", Billy Chen and Jerrod Ankenman
  • "Applications of No Limit Holdem", Matthew Janda

Dec. 1, 2019 | 10:31 a.m.

+1 more for Tipton. I stopped reading poker books after studying those. You can't just skim them and expect to get much out of them though.

+1 to that Tipton should probably not be first theory book to read. I think Sklansky's books are great for basics.

Dec. 1, 2019 | 10:27 a.m.

Difficult to say and the longevity of a format is dependent both on the action and how beatable it is. Action is declining and regs do get better but it has weaker regs and more action than HU cash and even there you can make it in 2019. GTO assistance is both more difficult to make and less powerful than in HU formats so the fear of that is also less relevant. I'd say you don't have many reasons to not go all-in on the format if it's your favorite. The best online regs make huge amounts of money and the live scene is booming in certain places.

Nov. 20, 2019 | 10:50 a.m.

Just 4 years too late and the problem has been widely known for over 3 years already :-(

At this point, sites might as well stop offering HU Hypers when close to GTO RTA is used as far down as $60 games. I'm much more optimistic about the longevity of certain forms of ring games online than a few years ago for formats that involve fish dynamics and more MW postflop play. Those are areas bots will most likely never surpass good human players.

Nov. 8, 2019 | 9:51 a.m.

so we end up w a lot of give ups after x line

Yeah, and you can't get bluffed off 0% equity so there is no need to prevent getting bluffed off your equity with majority of range. XR is there to keep IP from betting too thin for value but it has to be as high or higher EV than betting.

Nov. 4, 2019 | 8:20 a.m.

"Real MDF" comes from indifference between bluffing and checking to SD for IP. Checking to SD has EV so bluffing needs to be +EV. In this spot, it needs to be very +EV to bluff as the highest equity bluff (the strongest hand that bluffs) has over 25% equity. Thus the EV of bluffing needs to be more than 25% of the pot.

MDF turns out to do more harm than good in a lot of spots, and understanding indifference ideas are much more useful. The naive MDF is just a special case of the indifference principle in poker.

Nov. 3, 2019 | 11:35 p.m.

overall range is weak and wide

That is part of the it but you do miss some parts of the reason we fold so much here.

Nov. 3, 2019 | 10:58 p.m.

Do you make much money from bluff catching in poker?

Nov. 3, 2019 | 12:40 a.m.

Yeah, I also think it's more of a thing in HU. Specifically boards like AJx to AKx and KQx because you 3bet close to all 88+, KQ and AJ+ in HU. I haven't looked at 200% pot for 6max so might still be some nice spots.

The spots where OOP player can't XR much because of capped too low it can make sense to split out some hands like 2nd set (blocking mid calls) into a larger size like 200% pot while betting most other hands for more like 120% pot. This is just an idea and not something I've worked on and examined.

Oct. 26, 2019 | 2:41 p.m.

200% as only bet size is usually a bit large so probably need like a 100% pot size too in most spots and use 2 sizes at the same time. 120% to 150% can be used as only bet size in a lot of spots. The spots it makes sense to go 200% OTT after 33% OTF it often makes sense to start putting in more money already from the flop. Flops like AKx, AQx etc in HUNL for example.

Oct. 25, 2019 | 12:42 p.m.

It's definitely not allowed on Stars and not even in gray area.

Oct. 25, 2019 | 10:29 a.m.

Comment | Kalupso commented on Short Stacking

Yeah, Diego Ramirez aka ramifish. He's not a shortstacker like in 20bb but a 40bb+ one.

There is one Pio video by Uri P on postflop adjustments.

P. S. Might have botched spelling his name but you should be able to find him.

Oct. 25, 2019 | 6:40 a.m.

Sim is just a model and not perfect. KQo is very close to 0EV OOP to two strong ranges and EV of call and 3bet is the same at equilibrium. Situation at the tables at 10NL is not equilibrium tho.

Oct. 24, 2019 | 5:35 p.m.

Seems alright to both squeeze and call depending on assumptions. The solver folds a ton to squeezes so might be best to not 3bet the weakest squeezes bellow:

Oct. 24, 2019 | 3:46 p.m.

Preflop solver ranges are good to use in my experience but charts is not the way to go. I'd put those pictures into a flash card software and drill for thresholds (weakest 3bet, call etc). After you have thresholds down then focus on understanding how ranges are being split between passive and aggressive VPIPs. A combination of finding patterns (understand why and how) and raw memorization has been most effective for me. Easier to memorize when understanding and easier to understand and see patterns when I know many different ranges.

NLHE 6max preflop is pretty easy to learn in my experience compared to more complex games like HUSNGs, MTTs or PLO so shouldn't take too much time to understand all the patterns and memorize the thresholds. Maybe like 10 to 14 hours spread over 3-4 weeks?

Oct. 24, 2019 | 10:51 a.m.

A full $15 for 700 hours of play XD

Oct. 17, 2019 | 8:14 p.m.

Do you play that size as range bet?

It's around the size Pio likes if you cbet entire range on those boards. I used to play that way 3 years ago in HU and a few other regs did too so the idea isn't exactly new. 10% pot is on the lower end of reasonable sizes. I used more like 16% pot on those boards. Strategy is quite close to range check but makes the pot a bit larger when your range has an overall EV advantage but doesn't want to weaken checking range because you don't have nut advantage or even have nut disadvantage but higher EV overall.

Oct. 16, 2019 | 10:49 a.m.

Probably a fold preflop with that big of a raise and a caller. Offsuit hands doesn't play great OOP against two somewhat tight ranges unless they're ahead of similar hands. Neither the raiser nor the caller should have QTo so that should make it a fold. With better odds you don't have to be ahead of stuff but here I think you do.

Postflop you have a weak trips hand given what ranges looks like. I think you have a great hand but not one which wants to narrow ranges too much so I prefer to bet myself or XC rather than XR. I can't say it's a bad play but I'd rather XR KQ+ than QT or QJ.

Against a non tilting and not spewy reg you're likely never good on the river. It's closer against fish or regs that can be spewy/tilty. Regardless I lean towards a fold given the strength of both your and his line. In terms of GTO you're also at the bottom of your range and can definitely fold some trips without folding way too much. That shouldn't be much of a concern at 5nl but I'm just adding it in as reference to show that we don't have to make light calls.

Sept. 26, 2019 | 11:03 a.m.

Nice Post!

i truly believe there is unconscious pattern recognition and knowledge that you can access when your playing your best and have played and studied the game a lot.

I don't think this is anything controversial. Top performers in most fields including sports, golf, arts and chess mostly uses their intuition when performing.

Sept. 1, 2019 | 12:53 p.m.

This is really interesting because my first thinking was: " Rake is too high, see more flops doesn't seem like a great idea " but as you said when we are SRP IP is "common" to have a bigger share than optimal.

My sims indicates that SB win rate isn't impacted much by opening size but BB win rate is so it's BB that mostly takes the hit from rake when sizing down. Going smaller than 2.2bb or 2.3bb seems to be too small though and rake might become too much of an issue.

I ran a quick sim to compare 5x vs 4x 3bet and there was a small increase in flops seen as Caller for BB, and SB was allowed to raise slightly wider because we can't 3bet as much so the point about 3bet frequency seems valid.

Aug. 30, 2019 | 5:06 p.m.

Really nice approach to making the video and content is very solid.

Do you have any thoughts on picking preflop sizes? I ran a bunch of sims with around 100nl rake recently and my conclusions overall was that size really doesn't matter much from GTO perspective as long as you stay within a certain range. For opening anything between 2.2bb and 3.Xbb seems fine and for 3bet anything between like 3.6x and 5.5x seems fine. Do you think it's viable to size down opens to see more flops assuming we win more than our fair share as SB in SRPs? For 3bet pots when called I'm not sure if I'd be able to reach higher win rate than GTO against regs of similar skills but surely would against fish or significantly weaker regs.

My overall conclusion is that it seems to make sense for SB opens but not necessarily BB 3bets and could even be good to try to see less flops there. Against fish we often have other considerations than just seeing more flops or not.

Aug. 30, 2019 | 3:32 p.m.

Very good advice on solver. I like to check a ton of my hands with sims but it's the accuracy of my thought process I like to check, not what I should have done with exact hand.

Second best way I think is to send someone who is interested to help you (likely coach, not sure study group work in sufficient way)

I think coach or forum here is much better than a study group if you don't have access to midstakes+ players (probably at least 5/10 for live). I even think a lot of the micro study groups can do more harm than good because the members don't really know how to improve yet and probably don't see the big picture of a lot of stuff.

the accuracy of my thought process

This is only possible if you know the assumptions you made and remember how you viewed the spots and can convey that information to the party helping you.

Aug. 26, 2019 | 7:42 a.m.

Interesting stuff and it makes sense that it's exactly low flops people play the worst flop strategies as they are by far the least common flops along monotone flops.

The 100% cbet strategy makes very little sense to me without a significant advantage against anyone reasonably competent. What you basically do is give up your positional advantage on the flop as BTN so BB can be aggressive with good hands and other good raising hands without worrying about splitting range and capping his calling range because he can immediately see the turn with all his medium hands and draws.

Aug. 25, 2019 | 8:08 a.m.

From a theory perspective you have the perfect bluff catcher because you block villains highest frequency check turn and shove river hand, AsAx.

Aug. 23, 2019 | 3:10 p.m.

Comment | Kalupso commented on 16z River decision

Yeah, I often just range bet those for 1/3rd pot in 6max. In looser formats like MTTs or HU it's bit much to range cbet as advantage isn't as big.

Aug. 22, 2019 | 4:40 p.m.

Preflop is very loose and close to bottom 25% of hands.

Flop seems fine.

I think turn should just be a clear fold and even something like J-high FD isn't great to call either. The extra overcard should increase EV a bit as he can have some AT. Reasons are low equity vs everything including his bluffs and domination problems when hitting flush.

River play isn't bad per se but if you call all FDs OTT and then shove OTR you put in too much money against a really strong range. The mistake is turn play and not river here.

BB river range: A GTO model will mostly check bluffs that give up, AT and rivered 9x like you said together with a small amount of traps selected because of specific blocker effects. A bluff is supposed to be higher EV than check for you because you didn't get direct odds to call turn and need to "make back" that EV by having +EV river bluff (so if OOP you wouldn't be able to call even Jh8h as it wouldn't make back EV by bluffing river). In practice some weaker players "trap" too much and give up too many bluffs which can make the small stab better than all in (all in is what makes sense vs GTO model).

Aug. 21, 2019 | 1:25 p.m.

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