Hi Kevin Rabichow, thank you for the video. I have a few questions I hope you can address:
1. It looks like you've set your exploitability to roughly 2.5bb/100 which you said is not highly accurate but suitable for the analysis you are doing. I typically see people recommend accuracy of .5% or .25% which usually results in an exploitability of more than 2.5bb/100. My question is, in your opinion, what is acceptable accuracy for broad analysis and precise analysis?
2. It looks like the difference between the 33% and 75% bet-size is about 1bb/100. Is the difference of 1bb/100 really worth using a larger bet-size given the smaller bet-size is easier to implement and balance?
3. Do you have any suggestions or heuristics on how to implement a mixed strategy such as the 75% bet-size with some level of accuracy? This seems to be one of the most difficult things to do.
Yes, in this case I'm simply looking to compare EVs of a bunch of sims to each other, so as long as the numbers are converging to the same point, i'll get the idea. If I wanted to get precise EVs of individual hands within the strategy, or do a full preflop tree and analyze all the strategies within it, I'd usually just let the sim run overnight and let it get really accurate. Since my plan is to run lots of sims here I'm trading off the time it takes to be more precise, and I don't think there's any need to be that precise when making this kind of comparison. The numbers just need to be relatively close to each other, not precisely close to the true value. I probably could've stopped it way earlier and done just fine, but I usually go to 1.5-2bb just because I'd like the differences to be somewhat accurate in quantity when deciding if it's "worth it" to implement (decent segue to 2).
Probably not. But there's also the question worth considering in today's environment - do your opponents play substantially worse vs. a strategy that utilizes large bets more often (in theoretically accurate ways)?
I like to build heuristics out of what I find in the range explorer. This might be something I work in later in the series, but basically I think the range explorer is great for categorizing your frequencies, rather than trying to remember what every hand is doing and how often.
Good video, and having 2 different betsizes on the flop is also something I've been working on. I think people are playing better and better vs the 1/3rd so I definitely think having a larger sizing is useful.
Good vid. Gonna go preform some of the same analysis myself on a selection of boards I a 3bet pot BBvBN (I'm not a HU player, but I enjoyed seeing the method of attack with PIO here.).
I appreciate your effort in this video. I got a lot out of it.
I wanted to ask if there are any practical benefits of adding 'unnecessary complexity' to our strategy. I'm sort of thinking of a guy like LLinus, who plays so many bet sizes that it is hard to pinpoint what he is doing. I imagine it makes him really intimidating to play against.
Is there some value in intentionally over-complicating our strategy, for the sake of making our opponents more uncertain? I figure, as long as we are making a reasonable effort to balance -- we'll always have better ideas about our strategy than our opponents will, even if we do sometimes play a less accurate/refined strategy.
Thank you! I would say yes, there's definitely value in the approach you described. It's quite common among people who consider themselves naturally strong poker players. I know Sauce advocates funneling opponents into a part of the game tree they're less familiar with, and similarly I've said that when playing 6max I'll design a strategy that gets me into more frequent situations where I believe I have an edge.
That said, we are humans who will inevitably fail to execute our strategy perfectly. There are many strong players who choose an over-simplified, almost robotic approach, with the aim of rarely making mistakes and playing well against the population on average. This is probably more common in settings where multitabling and long hours are involved. To each their own!
If I could be honest for a moment, I think a little bit of voice inflection would go a long way. Your content is likely some of the most valuable content available to players, but I often find myself zoning out and having to re-watch because of the monotone voice.
therapist yeah I do agree but the thing is they are poker coaches and not actors, so I guess so long as they are providing quality content then it is up to us to be able to absorb that content effectively.
when you speed up the video I think it is pretty easy to listen along to and take in imo.
I do hold the opinion that coaches/teachers should be engaging with their delivery, but mean no disrespect to Kevin; there's no debating the quality of the content itself. Just a minor suggestion!
Thanks for the video Kevin, Definitely love your solver approach. I would be interested to see how you make your heuristics or a summary of your heuristics later in the series.
Doesn't giving only 120% turn sizing for IP significantly hurt the bigger flop sizing? I would suspect that on the 962 suited board atleast.
Nevertheless very useful video!
I haven't looked at the effect recently, but I recall having the same thought and being surprised that it wasn't too relevant in the overall numbers. If you want to investigate how the EVs change on specific board textures, I'd be interested in the result!
About your preflop ranges on BB, do you think trying to implement freqs with high mixes like 25%, 75% ,50% rather than just 50% 100% will make a significant ev+ change to your gameplay?
And what do you think of nodelocking flop bb and make assumptions about his frequencies ?
I would regularly rearrange to a simpler mix strat and expect very little effect on EV as long as you're going about it in a deliberate way. Nodelocking is a great exercise for figuring out exploits in the population, but I think it should be considered alongside the unlocked solution. It's easy to get carried away with extreme strats within nodelocking and not realize how vulnerable you are to getting counter exploited.
Great video Kevin ty! I noticed @14:52 with the EV difference you divided by 10 (the bb size in the 5-10 game). So it would be correct to ignore this step if we did a calc for stake 50c-$1 since 1 chip would be 1bb? (for example the pot on the flop is 5 chips (5bb's))?
Loading 25 Comments...
first!!!
Hi Kevin Rabichow, thank you for the video. I have a few questions I hope you can address:
1. It looks like you've set your exploitability to roughly 2.5bb/100 which you said is not highly accurate but suitable for the analysis you are doing. I typically see people recommend accuracy of .5% or .25% which usually results in an exploitability of more than 2.5bb/100. My question is, in your opinion, what is acceptable accuracy for broad analysis and precise analysis?
2. It looks like the difference between the 33% and 75% bet-size is about 1bb/100. Is the difference of 1bb/100 really worth using a larger bet-size given the smaller bet-size is easier to implement and balance?
3. Do you have any suggestions or heuristics on how to implement a mixed strategy such as the 75% bet-size with some level of accuracy? This seems to be one of the most difficult things to do.
TIA
Yes, in this case I'm simply looking to compare EVs of a bunch of sims to each other, so as long as the numbers are converging to the same point, i'll get the idea. If I wanted to get precise EVs of individual hands within the strategy, or do a full preflop tree and analyze all the strategies within it, I'd usually just let the sim run overnight and let it get really accurate. Since my plan is to run lots of sims here I'm trading off the time it takes to be more precise, and I don't think there's any need to be that precise when making this kind of comparison. The numbers just need to be relatively close to each other, not precisely close to the true value. I probably could've stopped it way earlier and done just fine, but I usually go to 1.5-2bb just because I'd like the differences to be somewhat accurate in quantity when deciding if it's "worth it" to implement (decent segue to 2).
Probably not. But there's also the question worth considering in today's environment - do your opponents play substantially worse vs. a strategy that utilizes large bets more often (in theoretically accurate ways)?
I like to build heuristics out of what I find in the range explorer. This might be something I work in later in the series, but basically I think the range explorer is great for categorizing your frequencies, rather than trying to remember what every hand is doing and how often.
Good video, and having 2 different betsizes on the flop is also something I've been working on. I think people are playing better and better vs the 1/3rd so I definitely think having a larger sizing is useful.
Thank you, I agree!
Good vid. Gonna go preform some of the same analysis myself on a selection of boards I a 3bet pot BBvBN (I'm not a HU player, but I enjoyed seeing the method of attack with PIO here.).
Thanks Kevin, great vid!
I appreciate your effort in this video. I got a lot out of it.
I wanted to ask if there are any practical benefits of adding 'unnecessary complexity' to our strategy. I'm sort of thinking of a guy like LLinus, who plays so many bet sizes that it is hard to pinpoint what he is doing. I imagine it makes him really intimidating to play against.
Is there some value in intentionally over-complicating our strategy, for the sake of making our opponents more uncertain? I figure, as long as we are making a reasonable effort to balance -- we'll always have better ideas about our strategy than our opponents will, even if we do sometimes play a less accurate/refined strategy.
Thank you! I would say yes, there's definitely value in the approach you described. It's quite common among people who consider themselves naturally strong poker players. I know Sauce advocates funneling opponents into a part of the game tree they're less familiar with, and similarly I've said that when playing 6max I'll design a strategy that gets me into more frequent situations where I believe I have an edge.
That said, we are humans who will inevitably fail to execute our strategy perfectly. There are many strong players who choose an over-simplified, almost robotic approach, with the aim of rarely making mistakes and playing well against the population on average. This is probably more common in settings where multitabling and long hours are involved. To each their own!
nice video man, seemed a little dry at the start but was very interesting the longer it went on. looking forward to the next part!
I would recommend watching it at 1.25/1.5x speed, though :)
If I could be honest for a moment, I think a little bit of voice inflection would go a long way. Your content is likely some of the most valuable content available to players, but I often find myself zoning out and having to re-watch because of the monotone voice.
therapist yeah I do agree but the thing is they are poker coaches and not actors, so I guess so long as they are providing quality content then it is up to us to be able to absorb that content effectively.
when you speed up the video I think it is pretty easy to listen along to and take in imo.
Yeah will try 1.25x thanks.
I do hold the opinion that coaches/teachers should be engaging with their delivery, but mean no disrespect to Kevin; there's no debating the quality of the content itself. Just a minor suggestion!
I assumed everyone has my videos at 1.5x by now honestly
Kevin Rabichow lol
Nah; I've gotta say Krab is the only one that talks slow enough for me. :)
radtupperware xD
Big fan of this video, can’t wait for the next video in this series - thanks for continuously reviewing the strategy’s shortcomings.
Thanks for the video Kevin, Definitely love your solver approach. I would be interested to see how you make your heuristics or a summary of your heuristics later in the series.
Hey Kev, just got through this first video here on RIO, excellent content.
As a suggestion, have you considered having cliffnotes in the description as to help the viewers with the main takeaway points?
Cheers!
Doesn't giving only 120% turn sizing for IP significantly hurt the bigger flop sizing? I would suspect that on the 962 suited board atleast.
Nevertheless very useful video!
I haven't looked at the effect recently, but I recall having the same thought and being surprised that it wasn't too relevant in the overall numbers. If you want to investigate how the EVs change on specific board textures, I'd be interested in the result!
Hey Kevin,
About your preflop ranges on BB, do you think trying to implement freqs with high mixes like 25%, 75% ,50% rather than just 50% 100% will make a significant ev+ change to your gameplay?
And what do you think of nodelocking flop bb and make assumptions about his frequencies ?
I would regularly rearrange to a simpler mix strat and expect very little effect on EV as long as you're going about it in a deliberate way. Nodelocking is a great exercise for figuring out exploits in the population, but I think it should be considered alongside the unlocked solution. It's easy to get carried away with extreme strats within nodelocking and not realize how vulnerable you are to getting counter exploited.
Great video Kevin ty! I noticed @14:52 with the EV difference you divided by 10 (the bb size in the 5-10 game). So it would be correct to ignore this step if we did a calc for stake 50c-$1 since 1 chip would be 1bb? (for example the pot on the flop is 5 chips (5bb's))?
Yep exactly, this is just because of the bb size. No need to change this if 1 chip = 1bb
Be the first to add a comment
You must upgrade your account to leave a comment.