Fossana
45 points
Yes cbetting range on many flops would be exploitable in that way, but the point is that piosolver says my polarized cbetting strategy is nearly as exploitable as cbetting range on those same boards.
Sept. 13, 2018 | 5:27 p.m.
Assume that we are IPvsBB in a single-raised pot. My theory is that we can cbet range on every flop. Take J97tt BTNvsBB for example. This is one of the flops where cbetting range is most exploitable, so naturally we would not want to cbet range and instead approximate a GTO polarized cbetting strategy on this flop. I input my best conception of a GTO polarized cbetting strategy on this flop using the nodelocking feature in piosolver, and while the EV of my strategy was somewhat better than cbetting range (2.2% of pot lost vs 2.66% of pot lost), I'm not sure that this difference is significant enough to overcome the fact that cbetting polarized complicates turn strategy (have to worry about betting and checking ranges on turn, have to worry about lines like delayed cbet and facing turn probe).
Maybe this means that I need to get better at approximating GTO on flop, or perhaps cbetting range on flop doesn't actually simplify turn strategy. Either way, I think there can be some merit to cbetting range on every flop, since trying to approximate a GTO strategy that cbets less frequently may be unsuccessful. As humans, our approximations of GTO may be too far off of actual GTO, so we'd be better off just betting range on every flop and making up the EV on turn.
Sept. 13, 2018 | 4:44 p.m.
I've been trying to figure out specifically why being in position is a good thing.
First, the in IP player gets to act with one additional street of information at all times.
Second, the IP player's option to check is much more powerful than the OOP player's option to check, because it either goes to showdown immediately or the next card pops out. When the OOP player checks, the IP player can still bet on the current street.
Third, if the OOP player checks his entire range to avoid giving away information, then it's basically like he had his turn skipped, which can't be a good thing.
Do you think this is accurate?
I think the reason that higher SPR favors the IP player is because it gives the IP player more leverage to use when he has the informational advantage, and there will be on average more streets played with higher SPRs, so the IP player gets to use his information advantage on more streets and the OOP player will split his range more, and thus he'll have more checking ranges overall, which are inferior to IP's checking ranges.
Perhaps information is more valuable when ranges are wide, which would favor the IP player, or perhaps the OOP player has to check more with a wider range, and we know that the IP has an easier time pot controlling and applying pressure.
I'm not sure about the board textures, but people say that dynamic boards favor the in position player. Perhaps wet isn't a good measure of volatility, and what you could do is find a program that shows you how much equities of hands change on turn cards for a certain range and a certain flop, and then order the boards on how likely equities are going to change on the turn. I suspect that the IP player's information advantage is more useful if equities are going to get mixed up a lot so that information sort of resets for both players, and the OOP player will have to check on more turns if the board is volatile, and as already stated, the IP player can take advantage of this better than the OOP player can.
Sept. 8, 2016 | 5:42 a.m.
The assumption is that Villain has slow played hands that beat Hero's "nut" hands.
Sept. 4, 2016 | 4:24 a.m.
Let's say Hero has either "nuts" or air, and Villain has a bluff-catcher, but also some slow-plays. If Villain is OOP, I have a theory that he only leads his traps if Hero prefers to check back his entire range if Villain checks his traps.
By leading, Villain never gets a bet out of Hero's air. Some of Hero's "nut" hands will also fold to keep Villain's bluff-catchers indifferent between betting and checking, so Villain will also lose a bet from some of Hero's "nut" hands. I also believe that Hero will call a jam more often after betting because his pot odds will be better.
So leading always seems worse than checking, unless Hero is not betting in the first place.
Aug. 4, 2016 | 7:12 p.m.
I'm trying to figure out how much you should bet with hand on the river if both players have the same range of hands (symmetric situation). I thought that perhaps you would use Will Tipton's optimal bet size formula for traps, but instead of the trapping frequency, you put in 1 - EQ(hand). I would expect the optimal bet size to be a bit larger than this since you can bet smaller with your nutted hands to balance your ranges and discourage your opponent from raising and turning your hand into a bluff catcher, but I gave both players AA-44 on 33222 in piosolver, and the bet size used by every value hand is significantly smaller than the optimal bet size for a PvBC w/ traps situation. I had the OOP player check 100%. I suppose the IP player has to bluff using hands that have a tiny bit of equity, so I guess this means the opponent continues less and therefore we get less value. So perhaps it's betting smaller to expand calling range? I don't think the OOP player is jamming more frequently than in a PvBC w/ traps scenario because the bet size piosolver was using for KK was being used quite a bit by AA for example. I tested for card removal effects, but that didn't seem to do the trick. Compared to a PvBC w/ traps situation, the opponent must be either raising more in comparison, which is doubtful since we'll have our own traps in each of our ranges, or our opponent continues less than the minimum defense frequency, but I can't imagine that the fact our bluffs (44) has a tiny bit of equity is the major factor here. Overlapping ranges could have some singificance, but in my scenario only KK and AA was betting, so surely the inclusion of AA wouldn't make the optimal bet size smaller then if we were only betting KK and some bluffs?
July 29, 2016 | 11:34 a.m.
Look up polarized range vs bluff catchers + traps. Figure out how many traps villain needs to have before you can't jam river.
July 29, 2016 | 7:01 a.m.
Generally I c-bet with a polarized range. This means I'll c-bet the following:
TPGK+
Some draws, usually the ones that don't have any showdown value
Air with backdoor draws or some sort of ability to beat villain's continuing range
And then I'll check the following:
The occasional slowplay
TPWK
2nd/3rd pair
Some draws, usually the ones with some sort of showdown value
Other air
My question is: When should I cbet 2nd/3rd pair? I imagine that 2nd and 3rd pair would play a mixed strategy at equilibrium, since being able to have trips when the middle or low card on the board pairs up is useful. My guess is that 2nd and 3rd pair do more checking than betting, but that they would cbet at like a frequency of 20%. Since I'll arrive to the flop with a variety of 2nd and 3rd pair holdings, if I were to not to play any of them with a mixed strategy, that is, I'd play each one using a pure strategy but use a mixed strategy for my entire range of 2nd and 3rd pairs, which ones should I choose to cbet with rather than check? Also are there any situations where I would want to bet most of them at a high frequency? Here's what I'm thinking so far:
2nd pair + good kicker is better to bet than 2nd pair + weak kicker.
The more overcards there are to the middle card on the board, the more important it is to cbet 2nd pair for the sake of protection. I think this could be extended to draw heavy boards, but avoiding a raise may be prudent.
The higher the middle card is the more value we can expect from cbetting 2nd pair with a good kicker. This is because people will play more hands preflop containing higher cards, so we can expect to get called by more worse middle pairs. Of course we also need less protection when the middle card is higher, so these two factors work against each other.
If our range is very air heavy, then betting 2nd/3rd pair lets us bluff more. Of course if we're cbetting with a high frequency then we usually want to use a smaller sizing so that our opponent is forced to continue with worse, but if we use a smaller sizing we can't bluff as much...
July 5, 2016 | 3:59 p.m.
I'm blindly cbetting polarized on every board. So here are some situations that I know come up and why I think they come up, but let me know if I'm wrong:
We check our entire range on the flop. We do this on boards where our opponent has an equity advantage and a higher density of nutted combos.
We cbet our entire range on the flop with a small sizing. We do this on boards where we have both an equity advantage and more nutted combos.
We cbet polarized. We do this on boards where we don't necessarily have an equity advantage but our range contains more strong hands.
Then I believe there are situations where we want to cbet a more merged range with a somewhat smaller sizing, but I'm not sure when this comes up.
Usually I cbet TPGK+ for value and I cbet weak air and some draws. In which situations would I cbet 2nd/3rd pair? It's probably unbalanced to never have trips in my betting range when the middle card on the board pairs up. Same with the low card pairing up.
When I cbet air with a polarized range, what sort of air is best to cbet versus check? I don't think I should cbet all of my air because then my checking range is never folding to a turn bet, which can be exploited.
July 4, 2016 | 11:13 a.m.
What do you think of using PioSOLVER to find patterns? See what kind of variables cause you to cbet with a polarized range and see what kind of variables cause you to cbet with a merged range with a high frequency. How does position affect it? How does equity and nutted combos affect it?
Personally I've been struggling to learn GTO and apply it to my game. I can't tell if I'm getting exploited all the time when I'm underbluffing and overfolding at the lower stakes. Often times I feel like I'm just guessing when I bluffcatch. I'm stuck using old fashioned thought processes like: "What worse hands will call me?" "What better hands will fold?" "How many streets of value do I want?" "What's my plan if I get raised?" "Does my bluff represent a strong hand" "What sort of hands does he have that can play this way?" It's hard to come up with a new poker schema that can be used in practice, so the idea of simplifying models resonates with me.
July 4, 2016 | 7:35 a.m.
How did you learn this from your own studies? I'm currently reading Will Tipton's books.
June 27, 2016 | 6:25 p.m.
@8:15 in the hand where Hero had 96ss on 8c7c4c you said that you would triple barrel in that spot, and that you shouldn't only bluff with hands that contain a club. How do you go about learning these things? After all of the books I've read and all of the videos I've watched I can't just look at that board and know that I should triple barrel 96ss and that I should bluff hands without a club in them. Same for when you said that you shouldn't raise all of your Ax from the BB when the SB limps because you'll struggle on Ace high boards when you check your BB.
June 27, 2016 | 3:40 p.m.
Javier is opening 2.25x, so I think that means he can open with a wider range. If villains 3bet to 8x, Javier's minimum defense frequency is around 32%.
June 15, 2016 | 8:14 a.m.
@4:00 min, what do you think of villain's cbet with Th9h? Is it good or bad? Usually I'd like to see a cheap showdown with bottom pair, but I suppose with his 1/3 pot sizing he can get some value from floats as well as some protection. He could easily value own himself though.
In that same hand, did you decide to just call with your combo draw OTF because of the great pot odds you were being given? Why does raising become a better option when villain bets larger? Value betting the river seems scary because villain could have KK, JJ, 99, K9, J9, and better flushes.
June 14, 2016 | 12:08 a.m.
What's your general cbetting strategy? As far as I can tell you cbet with a polarized range. You size up versus fish with value and you size small exploitatively on dry board textures versus opponents who are folding too much. It seems that you size larger on boards that favor your opponent and otherwise you use a sizing that's around 1/2 pot. This is of course probably specific to 50z. Is the rest of your cbetting strategy made up of exceptions or advanced topics that intermediate players shouldn't bother with until they master the general cbet strategy? What makes a hand bad enough that you want to give up with it rather than cbet? How do you decide if you want to cbet or check a hand that has overs and backdoor equity? It appears that you sometimes cbet and sometimes cbet such a hand.
June 13, 2016 | 3:11 p.m.
If you play on Bodog a lot, you can get table tamer for hotkeys and general multi-tabling purposes. What HUD stats would you use if you think you can get 100 hands on someone? Just VPIP/PFR/3bet/f3bet? Or would you mix in AF/steal/fold to steal/cbet/fcbet?
Nov. 19, 2015 | 1:39 a.m.
Wow, this video really hits home...
When I get deep in tournaments, I play with scared money. In the middle stages if it's folded to me on the small blind, I have no problem exploiting my opponents by pushing any two cards when the effective stack size is ten big blinds, but if I'm anywhere near a final table, I pass up on +EV all ins because I have the irrational fear that my opponents will wake up with a hand or that they will call me light. Anywhere near final tables, I'm reluctant to get a playable hand. I'm not bold enough to re-steal. I'm unable to follow through with my bluffs. I pass up on value by over betting just so I don't have to make a decision on a later street. If someone wanted to maximally exploit me deep in tournaments, they should double barrel their entire range, re-steal every time I open in late position with a vulnerable stack, and check-raise every time I c-bet.
I guess the core problem is I love the idea of scoring big. I want to tell everyone about my final table cashes. I want random people on the internet to think I'm good at poker. I want to stop taking out loans to pay for my college expenses. I want to feel adequate. Perhaps that's why I get depressed when I'm on a big downswing. Maybe that's why I follow bad bank roll management when it comes to MTTs, but I'm a nit with sit and go buy ins. I ask myself: Why is it I put in so many hours into this game at the cost of my health and social life, only to have my bankroll revert to where it was a month ago?
Competition has a been a driving factor for me. My dad was a chess master, and he took me to chess tournaments when I was in elementary school, which is why I was able to relate to your section involving chess. Quick chess has always been a weakness of mine by the way. In high school, I devoted most of my free time to reaching masters league in Starcraft 2, and more recently, I tried to get enough championship points to qualify for the Pokemon World Championships. Poker is no different. I love the thrill of winning and moving up. Cash games don't do the trick. A steady income is boring. Winning is fun. But the cost of this mindset is an unstable mental game and a lack of self-improvement outside of poker. Of course if one recognizes that the bowl is dirty, one ought to clean it.
Oct. 13, 2015 | 8:16 p.m.
Do you have any favorite run it once pros? I play MTTs, but I imagine the NLHE content here would be helpful too. So far I like Ben Sulsky, Jason Koon, Sam Grafton, and James Obst. I'm Foosa on two plus two by the way.
Also, I took 6th on Sunday in the $33,333 Gtd on Bovada for $1300, my biggest score so far.
Sept. 18, 2015 | 7:47 p.m.
hey aren't you the person on two plus two who is crushing Bovada MTTs?
Sept. 18, 2015 | 7:27 p.m.
I think there are too many factors at a table for push charts to be reliable. Resteals also depend on other people's opening ranges. I think it's better to review hands later with ICMizer and use programs like ICMtrainer, and then to extrapolate at the table.
Sept. 18, 2015 | 9:40 a.m.
I wouldn't play a $5.50 MTT with only $100 personally. It's not too bad if you are playing like $1-$3 sit and gos primarily and the $5.50 MTT is on the side since the variance balances out some. If you are playing MTTSGNs I would not play the big $5.50. I'm not sure what your mindset is, but it doesn't take that long to build up a roll if one knows what they're doing, but trying to go for a big score in the beginning is the sort of thing that I think leads to redeposits.
Sept. 18, 2015 | 6:29 a.m.
I would only play $1 MTTs and MTTSNGs. I think 180 mans would be a good choice, but 45 and 90 mans are good too. Personally I would play $2-$3 sit and gos and build a roll, and then play MTTs up to $5 once you have a $500 roll.
Sept. 16, 2015 | 6:58 p.m.
Shoving AJo there is completely standard. I think calling with 66 would be read independent, but without any reads I'd probably fold, though I'm not sure that's correct. You can go on ICMizer and run the spot and see what it says.
Sept. 16, 2015 | 6:56 p.m.
I play on Bovada. I started with Sit and Gos and started mixing in MTTs this Summer. Wait until you can deposit around $100, and play the $3 Sit and Gos and build up form there. If you hit $200, throw in some $2.20 MTTs, and when you get to $500, throw in $5.50 MTTs. If you just play 2-3 tourneys each weekend, you might end up depositing for three months, and then if you cash you might only have enough for another 8 buyins or so.
Sept. 6, 2015 | 5:35 a.m.
What sort of tools do the best poker players use? What sort of study routine do pros follow? What medium is best for studying poker?
As far as I can tell these are the basic things you can do to study poker:
Watch videos/live streams.
Read books/strategy articles.
Find a coach.
Read and post on forums.
Discuss the game with other people (skype, in person)
Review hand histories.
Use software such as flopzilla, ICMizer, PT4, etc.
Theorycraft.
Watch poker on TV.
And of course run it up on the tables.
So here are more concrete questions:
1. What should the ratio of studying to playing look like?
2. From the list above, what would you prioritize, and are there specific sites/books/forums/tools you would recommend?
3. How do you prioritize what you study? It might be cool to figure out the nash push range for 7bb before a 18 person pay jump in an MTT from the HJ, but one would probably be better off learning 10bb push ranges from the SB and BTN first in terms of Chip-EV, as in not near the final table.
Machine learning is capable of finding the heuristics you're looking for like always cbet tpgk+ and any good draws. In the final model, I found that you can cbet 1/3 100% BTNvsBB on these boards without major EV loss:
A9+ means a board with 1 Ace and at least one card 9-K.
A8- means a board with 1 Ace and two cards 8 or lower.
2BW+ means a flop with at least two cards 9 or higher.
1BW means flop with 1 card 9 or higher.
Rags is everything else.
But because this is using the simplest model, it will only predict flops that you can cbet range on correctly 80% of the time.
In the second model I went further than the simple model and identified the variables that were most important in determining whether you could cbet 1/3 100% without a large loss in EV or not. These are the results of that:
Predictor variable | Importance
IP Equity: 0.248376
Connectedness OOP: 0.191935
Nut advantage IP: 0.161676
Top pair advantage IP: 0.116107
unpaired % OOP: 0.088248
pair % OOP: 0.076904
m : 0.022818
2BW+ : 0.018598
Rags: 0.018452
r: 0.017319
tt: 0.014563
1BW: 0.013978
Paired: 0.008381
High card paired: 0.002644
Nut advantage IP is the absolute difference in % between how often IP has an overpair or better versus OOP.
Top pair advantage IP is the same as nut advantage IP, but only applies to top pair.
Unpaired % OOP is how often OOP has an unpaired hand without a draw.
m refers to a monotone flop.
Connectedness OOP is how often OOP has a straight draw (open-enders were given double weight).
pair % OOP is how often OOP has a pair.
m, r, and tt correspond to monotone, rainbow, and two-tone.
Paired refers to whether the flop is paired or not.
High card paired refers to whether the pair on a flop is below or above the other card.
This sort of information is more useful because it can be used for extrapolation. These are variables you can think about when deciding whether you want to cbet range or not in other situations such as COvsBB, or against a nitty opponent. It can also help you make up for the 20% inaccuracy of the first model by being able to analyze the specifics of the flop in terms of these variables. It also gives you insight into why the heuristics work at all.
More importantly, this approach with machine learning can be used for many things: predicting preferred bet size, identifying the most important variables when it comes to bet sizing, predicting spots where you can cbet 1/3 100% in 3bet pots, predicting spots where you should check 100% in single-raised pots OOP, etc.
In general, it's important to find principles and patterns that you can use to extrapolate and generalize, but at the same time, you want a deep understanding of them, or else you won't be able to adjust correctly. If there were no patterns in poker, piosolver wouldn't be a very useful tool.
Sept. 15, 2018 | 1:36 p.m.