Threads13 gets into the winrate and math behind it and demonstrates how variance can affect it.
Variance is huge in poker and it can drastically slow down the learning process. This series is split into two parts: 1) Identify variance and explaining the fundamental mathematics of variance. 2) Shifting the focus to learning (instead of results) and maximizing our learning.
Premium Subscribers can download high-quality, DRM-free videos in multiple formats.
Cool first episode.
I am nowhere close to a math expert but been interested in statistics since I started poker lol.
Are you using the central limit theorem to approximate the distribution to a normal-distribution? Does the theorem apply because we have a sample greater than 50 or sth?
Looking forward to calculating our theoretical winrate using a F-distribution or whatever it is in future episodes.
When we do an estimate of our winrate, is it okay to put in rakeback/bonusses?
For example a player has a winrate of 5bb/100. He earns 2bb/100 from rakeback and/or bonusses. Can we say his winrate is 7bb/100 and use this number in our calculations?
When we do an estimate of our winrate, is it okay to put in rakeback/bonusses?
For example a player has a winrate of 5bb/100. He earns 2bb/100 from rakeback and/or bonusses. Can we say his winrate is 7bb/100 and use this number in our calculations?
No, because rakeback isn't influenced by variance, it's a steady income.
Just make the calculation with your 5bb winrate and add the 2bb/100 of rakeback at the end result.
Wowow - I can´t tell you how thankful I am for this video. It clears things up in an unbelievably manner, things I were "superficially" aware of have gotten crystal clear now. THANKS!
Aside from that - real professional video, the screen play is just awesome, never saw that before. Hats off! I can imagine that it took a lot of work (no matter how helpful the tool was you used for it).
Yep,
one of the best vids lately,
really looking forward to next episode!
tx for a lot of good work!
Great video. I think since the majority of us use Holdem Manager, an option to hide the results throughout HEM can make us far less distracted by variance. You would of course be able to turn this option back on if you wanted. Or with the ability to disable it for a set period of time.
Anyways, the Holdem Manager guys are open to suggestions if there is enough interest.
http://forums.holdemmanager.com/user-interface-ui-suggestions/31491-hide-results-globally.html.
This looks like the series have have been waiting years for. Really exciting.
Fantastic video, production quality etc. is very good and the way the material is presented is clear and concise.
The Dude was very cool before this video, but now he is just a freakin baller!
Great video, I can't wait to watch the upcoming episodes.
Very nice work sir. The only thing I found a bit strange was the guy referred to as "The Dude," was in a pic with viscous, white, gunk dripping from his mouth. Everything else was awesome.
Whoever told you to use that new presentation style is an absolute genius!
Depressing, yet comforting. LOL! Nice work sir!
@threads13:
Hi!
I am concerned that this will be a great series, but I wanna be sure that it is valuebale enough for me to watch it (you know, the day has only 24hours and hence time is always scarce -> you need to select more...)
So:
1.) I know how the variance works from the mathematic perspective - my knowledge comes here from the known book "mathematics of poker"
So will I find here anyways something new for me (in this part1)?
2.) Will you tell in further vids also something about the psychological side of "acceptance of variance"?
3.) why is this a series regards to the section "No Limit" instead of "others".
4.) Interesting that you are doing such a series and not one of the "mental" or "Mindset" Coaches - any special reasons for it?
Thank you for your answers in advance
Very nice work sir. The only thing I found a bit strange was the guy referred to as "The Dude," was in a pic with viscous, white, gunk dripping from his mouth. Everything else was awesome.
You don´t know the movie?
Watch it. Soon. No, better even sooner ... 
Are you using the central limit theorem to approximate the distribution to a normal-distribution? Does the theorem apply because we have a sample greater than 50 or sth?
Yes. Not sure what "sth" stands for, but we will mostly be dealing with large sample sizes so yes, the theorem applies.
Very nice work sir. The only thing I found a bit strange was the guy referred to as "The Dude," was in a pic with viscous, white, gunk dripping from his mouth. Everything else was awesome.
It's a character from this movie.
http://www.imdb.com/title/tt0118715/
@threads13:
Hi!
I am concerned that this will be a great series, but I wanna be sure that it is valuebale enough for me to watch it (you know, the day has only 24hours and hence time is always scarce -> you need to select more...)
So:
1.) I know how the variance works from the mathematic perspective - my knowledge comes here from the known book "mathematics of poker"
So will I find here anyways something new for me (in this part1)?
Yes. I've read it and I from what I remember they do very few (if any) examples to show the extreme variance present in poker. I don't think there is very much overlap at all.
2.) Will you tell in further vids also something about the psychological side of "acceptance of variance"?
Yes, that's the plan for episodes 3 and 4.
3.) why is this a series regards to the section "No Limit" instead of "others".
I use exclusively NLHE examples because that's my game. Of course, variance applies to all games so it could easily be categorized as "other". I'll speak with the guys about it.
4.) Interesting that you are doing such a series and not one of the "mental" or "Mindset" Coaches - any special reasons for it?
Not particularly. It hasn't been something I've put much thought into, tbh. I've never done any mindset exclusive coaching, but I definitely discuss it with my students during out strategy coaching.
Yes. I've read it and I from what I remember they do very few (if any) examples to show the extreme variance present in poker. I don't think there is very much overlap at all.
Yes, that's the plan for episodes 3 and 4.
I use exclusively NLHE examples because that's my game. Of course, variance applies to all games so it could easily be categorized as "other". I'll speak with the guys about it.
Not particularly. It hasn't been something I've put much thought into, tbh. I've never done any mindset exclusive coaching, but I definitely discuss it with my students during out strategy coaching.
Thank you for responding, I will watch it, seems to become rly valuebale series
Thank you everyone for the kind words. I have put a lot of work into this series and it's very rewarding to see that people are enjoying it.
Very nice work sir. The only thing I found a bit strange was the guy referred to as "The Dude," was in a pic with viscous, white, gunk dripping from his mouth. Everything else was awesome.
I think his point applies even though I love the character & the movie not the most pleasant image of him. Minor point though, very helpful video as I try to tough through a brutal stretch in my return after a year away.
As everyone has already said, Nice video, looking forward to the rest of the series
Beautiful presentation. Also great content. Thank you.
thank you for this very interesting video. Nice work!! Clears something up ;-) Looking forward to the following videos.....
Wow! Can't wait for next episodes! Is 'a must watch' definitely!
Thank you very much!
Also I really liked the whole presentation (its structure and style!). What software did you use? As a guy spending a lot of time drawing different slides for numerous presentations I just wonder if I can use it in my work ... =)
Wow! Can't wait for next episodes! Is 'a must watch' definitely!
Thank you very much!
Also I really liked the whole presentation (its structure and style!). What software did you use?
Prezi.com There's a small learning curve, but as you can see, it's pretty slick once you figure it out.
Prezi.com There's a small learning curve, but as you can see, it's pretty slick once you figure it out.
ty!
Very nice work sir. The only thing I found a bit strange was the guy referred to as "The Dude," was in a pic with viscous, white, gunk dripping from his mouth. Everything else was awesome.
Yeah? Well, you know that's just like, uh, your opinion man!
Time Link to 00:20:12
First of all, I am very glad that I have decided to watch this series - I think this will be an epic series
small poll@all:
I am more or less a 4bb/100-winner over 350k hands - but my Standarddeviation/100 is about 77bb/100hands.
What are your values here?
And what can a larger deviation mean regards to similar WRs?
Just that you choose high Variance-spots?
Absolutly great series. Nice first video.
Very very good, just what Doctor ordered. Shud be mandatory watch in DC, lol...
I didn't even know how much I wanted to see a series about this until I watched this ep. Thx.
Time Link to 00:19:40
Wait, how does the variance per hand end up being 36? If the St Dev per hand is .6 (60/100 hands), shouldn't the variance be .36 per hand?
Wait, how does the variance per hand end up being 36? If the St Dev per hand is .6 (60/100 hands), shouldn't the variance be .36 per hand?
You can't take the standard deviation per 100 hands and simply divide by 100 to get the standard deviation per 100 hands. You can do that with the variance, which is why we start by squaring the standard deviation to get 3600(60^=3600). So, the variance per 100 is 3600. Then divide by 100 to get the variance per hand of 36. Then we can get back to the standard deviation by taking the square root. This is how the standard deviation is 6.
Oh ok, makes more sense. Related question just to clarify, in Hold'em Manager is the St Dev in bb per 100 hands? It seems like it is, but as opposed to bb/100, they don't actually mark it as per 100 so I wanted to make sure.
Oh ok, makes more sense. Related question just to clarify, in Hold'em Manager is the St Dev in bb per 100 hands? It seems like it is, but as opposed to bb/100, they don't actually mark it as per 100 so I wanted to make sure.
Yeah, it's per 100. They have it as big blinds (bb) and big bets (BB).
Just finished this (finally) and I can see why it got such good reviews. Very solid content and professionally done as well. Far more preparation went into this than the average video and it shows.
I have a (maybe) dumb question but it keeps me busy already quite some time: my S.D. is around 90-95bb / 100 (over like 200k hands). From all I heard this seems to be quite high.
1) Is this right?
2) Can we derive any assumptions (or at least indicators) for my play from that (i.e. too fancy, to bluffy, too bluffcatchy, valuebetting too little, going to SD too often, not protecting enought)?
Thx!
I have a (maybe) dumb question but it keeps me busy already quite some time: my S.D. is around 90-95bb / 100 (over like 200k hands). From all I heard this seems to be quite high.
1) Is this right?
2) Can we derive any assumptions (or at least indicators) for my play from that (i.e. too fancy, to bluffy, too bluffcatchy, valuebetting too little, going to SD too often, not protecting enought)?
Thx!
Yes, that seems a bit high to me. I haven't personally seen any 6-max or FR stats that high before. I don't think I can say why it is so high other than you are making some high variance plays.
Excellent video... gonna crack on with episode two... but before I do, I can safely say I ran good tonight with some positive variance lol
I just wanted to join the amen corner and say that this is an outstanding video. The information is vital and the presentation was first rate.
Great job sir!
Yes, that seems a bit high to me. I haven't personally seen any 6-max or FR stats that high before. I don't think I can say why it is so high other than you are making some high variance plays.
I don´t want to beat a dead horse, but I want to get it clear for me. 
1) As far as I understand it, "high variance" play does not correlate with bad play, correct? I mean, I could have a huge negative bb/100 and still have a low variance, right?
2) What plays do in general support high variance? Aggressive plays or passive plays? I always thought the more aggressive you are the higher variance gets - but I´m more on the passive side, so I´m amazed about my high S.D. Or is it in general playing for too big pots? Could you maybe give an example for a "high variance play"?
I don´t want to beat a dead horse, but I want to get it clear for me.
1) As far as I understand it, "high variance" play does not correlate with bad play, correct? I mean, I could have a huge negative bb/100 and still have a low variance, right?
2) What plays do in general support high variance? Aggressive plays or passive plays? I always thought the more aggressive you are the higher variance gets - but I´m more on the passive side, so I´m amazed about my high S.D. Or is it in general playing for too big pots? Could you maybe give an example for a "high variance play"?
1) A high standard deviation doesn't necessarily equal bad play.
2) Thin plays generally carry with them more variance. If you look into the variance equation you can see that if you have plays that have a low mean as compared to the result and/or you don't have the positive result occur frequently then the variance will rise.
Mine is 97/100 over 50K hands. I'm also not particularly aggr - my 3bet is 7, flop CR 9, mid 30s overall AFq.
Based on my own leaks, I would guess it has something to do with taking call down lines where betting or raising is more +ev and/or calling too much.
Just finished this first episode, I made notes by pausing every slide and taking the important points down and I feel its really stuck in my head. I look forward to the rest of the series, especially the learning chunks, plans and routines.
Good video so far. Only about half way through and I like all of the information provided. I was just going to ask a fairly dumb question.
Are your presentations available for download? I'd like to have a hard copy that I can follow along with and jot notes for further reference. Being as I have been out of college for almost three years this plethora of information is a bit overwhelming as well.
Thanks in advance!
yes, i must say that this video is giving some light into statistics for someone who dont know statistics
I must recon the helpness of it, and thank the producer for that.
i must also comment that the producer drives some things into extremes, and i dont like that because i believe that i understand statistics alittle more than average people.
questions
why do you need to make all examples for 99.5 confidence ( 3 sd's from the mean) and cannot stop at 2 sd's at least ( 95% ) especially if you are dealing here with poker players who will not care ( be reasonable here) about if they dealing with 99.5% of their total result or only 95% of it?.
Make a parallel between this line and estimating their equity in any hand, do they care if they have equity of 99% or just 90%.
Now, the difference between 3sd and 2 sd is enormous in statistics ( some people will not find one sd being so enormous, but they are not understanding stats as they should if they play poker.
In one example you take 13.8% two times the extremes make it 30% and now, the fact that 13.8+13.8=27.2 doesnt matter anymore 2.8% away from the truth, but you push 3sd instead of 2sd's for 3,5% away from truth. From a statistical point of view you made a great error by adding statistical results of 2 different extreme results of an experiment + and - and turning it into one single explanation of the whole. but that's another story, of a man standing with one man on the burning stove and another man in the ice water and saying on average he is ok.
if you make that for your examples it would be 3.5% less accurate, but you would not scare people so much with the results with 2 sd's.
And i presume your intentions are to give people a real understanding not a scary understanding of variance in poker.
Not to mention the situations where some people would like to see a 68% confidence which is the bulk of their results in time
again, nice series, thak you
yes, i must say that this video is giving some light into statistics for someone who dont know statistics
I must recon the helpness of it, and thank the producer for that.
i must also comment that the producer drives some things into extremes, and i dont like that because i believe that i understand statistics alittle more than average people.
questions
why do you need to make all examples for 99.5 confidence ( 3 sd's from the mean) and cannot stop at 2 sd's at least ( 95% ) especially if you are dealing here with poker players who will not care ( be reasonable here) about if they dealing with 99.5% of their total result or only 95% of it?.
Make a parallel between this line and estimating their equity in any hand, do they care if they have equity of 99% or just 90%.
Now, the difference between 3sd and 2 sd is enormous in statistics ( some people will not find one sd being so enormous, but they are not understanding stats as they should if they play poker.
In one example you take 13.8% two times the extremes make it 30% and now, the fact that 13.8+13.8=27.2 doesnt matter anymore 2.8% away from the truth, but you push 3sd instead of 2sd's for 3,5% away from truth. From a statistical point of view you made a great error by adding statistical results of 2 different extreme results of an experiment + and - and turning it into one single explanation of the whole. but that's another story, of a man standing with one man on the burning stove and another man in the ice water and saying on average he is ok.
if you make that for your examples it would be 3.5% less accurate, but you would not scare people so much with the results with 2 sd's.
And i presume your intentions are to give people a real understanding not a scary understanding of variance in poker.
Not to mention the situations where some people would like to see a 68% confidence which is the bulk of their results in time
again, nice series, thak you
My intentions were actually to show the extremes, so that why I decided to go with 99% confidence intervals in general. The idea was to show players who have little background in this information how bad/good things can get because I feel that most player's have an inaccurate assessment of how extreme variance can be. So, for education purposes, I thought it was best to show the extremes so that it would be more dramatic and make the point stick. Extreme examples are easier to conceptualize and remember. Also, most players tend to underestimate variance so to counter that, I wanted to show the extreme variance. As a whole, the community underestimates variance, so I felt it better to show the other side of the coin to mitigate that. In your particular case, this series was slightly missing the target audience since you have a background in this material.
The example you gave of rounding 2*13=26 to 30 is true, but I favored simplicity and rounding in that particular example as to make things a little easier for the listener. Again, this is for educational purposes and was oversimplification/too inaccurate for someone with a background. Maybe that's a little bit too much liberty to take (and I can see that), but I don't think it detracts from my mission; which was to show players the math behind variance, show players how extreme variance can be (as mentioned above, this is why I showed the extremes), and to show players how to deal with variance. As for my examples as a whole, I feel like I hit the point really well. Of course, it's hard to be perfect, so I'm sure there are things I could have done better.
Furthermore, I've shown how the math is done, and provided many examples so if people want to use one of my 99% CI examples and look at a 68% CI example, it's a pretty easy to and/subtract 2SD's from my results. I think I mention in the series "if you don't like my numbers, try it with your own" and that is one of my main goals of the series - to encourage people to these calculations on their own. So, the 68% is easy for anyone to calculate and didn't really work with the theme I was trying to hit, to show the extremes, so I didn't bother with it. I think my examples were mostly fair and pointed. I didn't think it was necessary to do that for each example when I wanted to look more at the extremes that the average poker player (imo) underestimates.
This is really interesting. I just looked at my current database which is smaller, but I just got rid of my old one which was 250k hands plus. And my S.D. is 98. I play all 6max NLHE. This makes sense to me because It has always seemed like I was on a sick heater or playing horribly and losing. It's seems as I'm either winning for 20k hands at 45bb/100 and then losing at 44bb/100 for the next 20k hands. Now I just need to figure out what the high variance plays and I'm making are! Thanks again.
Time Link to 00:20:18
Is the s.d/50k calculation right here? or is it supposed to be .02684? Thanks!
Is the s.d/50k calculation right here? or is it supposed to be .02684? Thanks!
nevermind.....total retard here.
Been watching this for my "Math Attacks" series.
I have a question, in your presentation you treat your winrates as all equally likely. However, this is not the case, as they caveated in Chapter 3 of the book "mathematics of poker". It is much more likely that you win at a lower winrate than a higher winrate. So while the Gaussian tells us that it's just as likely that "the dude" wins at 10bb/100 as 6bb/100 this is simply not true via Bayesian inference. This also helps with the fact that we don't have to play such huge amount of hands to be certain we are winning players, stats converge much quicker.
Thoughts?
Been watching this for my "Math Attacks" series.
I have a question, in your presentation you treat your winrates as all equally likely. However, this is not the case, as they caveated in Chapter 3 of the book "mathematics of poker". It is much more likely that you win at a lower winrate than a higher winrate. So while the Gaussian tells us that it's just as likely that "the dude" wins at 10bb/100 as 6bb/100 this is simply not true via Bayesian inference. This also helps with the fact that we don't have to play such huge amount of hands to be certain we are winning players, stats converge much quicker.
Thoughts?
Yeah, I'm actually thinking of doing another episode to tie together some of this stuff with Bayes', but it was a little bit out of the scope of what I was trying to show in this series. I was more trying to show the extremes of variance and decided not to wade into anything Bayes' related for sake of simplicity. I was trying to give more black or white examples that people who don't know much about variance would get a lot from. Maybe I overstated it, but I'm kind of ok with that given that so many poker players are severely underestimating variance. But yes, good point that I probably should have mentioned as an aside.
I was more trying to show the extremes of variance and decided not to wade into anything Bayes' related for sake of simplicity.
As someone who only understands math in baby steps, I thank you for the simplicity
Yeah, I'm actually thinking of doing another episode to tie together some of this stuff with Bayes', but it was a little bit out of the scope of what I was trying to show in this series. I was more trying to show the extremes of variance and decided not to wade into anything Bayes' related for sake of simplicity. I was trying to give more black or white examples that people who don't know much about variance would get a lot from. Maybe I overstated it, but I'm kind of ok with that given that so many poker players are severely underestimating variance. But yes, good point that I probably should have mentioned as an aside.
No problem, great videos so far from what I have seen!
Time Link to 00:35:21
The dude doesn't worry about whether his winrate varies. The dude abides and I, for one, take comfort from that.
Great first episode.
This topic becomes TL ; DR, but screw it, I want to leave my 5 cents - excellent content and the best animations I have seen so far, it was super nice to watch, especially now, when I am suffering from a really big downswing (I hope it is a downswing and in reality I am The Dude ! )
Time Link to 00:45:08
Last sentence; Does this mean that the probability of running good/bad is related to two things; 1) uncontrollable things like dealt cards, winning/losing flips etc. and 2) our self controllable own play, which is partially effected by 1)?
Last sentence; Does this mean that the probability of running good/bad is related to two things; 1) uncontrollable things like dealt cards, winning/losing flips etc. and 2) our self controllable own play, which is partially effected by 1)?
Yes, there are many variables, of course, but that is part of it.
I just got a chance to watch your tolerance series, I found it extremely helpful. Especially because I have been playing PLO lately and it gave me the tools to analyze crazy PLO variance.
I have been adjusting the mean and standard deviation to see what the chances are for different winrates situations. I have the following question: at what minimum sample size are the chances for the different winrate situations accurate??
I have looked at two indicators so far. The first one is the signal to noise ratio, (mean/sd), wikipedia suggests a 1:1 ratio. But I don't know if this is the right number for poker. For example, at a mean of 4 and sd of 70. It takes 30k hands for mean/sd to be 1:1. Which means 30k hands is a good sample size for assessing the chances of different winrate situations. But for PLO where the sd is 140, it takes 120k hands for mean/sd to be 1:1. Does this mean that a good PLO sample size is 120k hands?? Or can we go for a smaller ratio which would decrease the minimum sample size required???
Another indicator is the coefficient of variation (sd/mean).
Could you please tell me which indicator is best and what ratio should I look for to get the minimum number of hands I need to play??
Is there a program and I'm thinking of Hold em manager in particular that can tell one whether they are running good or bad. We all know players that we think are really lucky, but is there any way of proving this one way or the other.
Not that I know of. There are so many variables in determining what is "standard luck". I'd imagine something like that would a piece of software in and of itself.
Not that I know of. There are so many variables in determining what is "standard luck". I'd imagine something like that would a piece of software in and of itself.
Just a thought, like starting hand rankings, sets hit, % time staying ahead compared to % winning chance.
Anyway I know I should not be concerned about such matters. One of my biggest leaks is losing too much when I am hitting nothing for hours and even days.
Looking forward to watching the remainder of this series.
amazing video!
very, very well produced and thought-out!
i look forward to watching the rest of this series asap.
Just starting out with your tolerance series. Quite amazing, helpful, and content that will never go outdated
Time Link to 00:23:19
First of all nice video, bit late to the party but least i made it!
Just a quick and maybe silly question, will someone explain how 1 standard deviation and 2 standard deviations etc are useful and what the relationship is between them and the % value?
I didnt fully understand it in the video and i'm thinking i might be missing a really simple point somewhere haha
First of all nice video, bit late to the party but least i made it!
Just a quick and maybe silly question, will someone explain how 1 standard deviation and 2 standard deviations etc are useful and what the relationship is between them and the % value?
I didnt fully understand it in the video and i'm thinking i might be missing a really simple point somewhere haha
It is useful to the degree of accuracy you need. In statistics very often you look at the 95% interval. Using 2SDs would be fine if you want to calculate for yourself. 1SD is not recommended. You should only use 3SD (imo) if you want a degree of certainty. E.g. would you rather know that it is 95% certain that you over 100k hands will win between -5bi and +20bi or would you rather know with 99,7% certainty that you will win between -25bi and +65bi? These are just random number but to give you an idea.
@threads13
I think this is a good introduction to statistics for people who might not know much about it. You do a good job explaining it. Just have one comment. Remember there is no such thing as negative variance. In stat terms that does not exist. You can be negatively affected by variance but the variance can never be negative, but for instruction purposes I do understand what you are doing.
Home → Poker Videos → Tolerance → Episode One
Questions or comments? Email us at support@deucescracked.com or connect with us on Twitter.
You are here. DeucesCracked videos provide a structured way to increase your online poker winnings and reduce tilt.
We have a FREE 7 day trial, so you can see for yourself.
101-level poker videos for free. Tournament and cash game videos.
keepyourchips.com
One on one coaching from DeucesCracked instructors.
deucescracked.com/coaching
The easiest way to share hands on forums, IM, or email.
thehandconverter.com
Get money back for playing online poker.
Earn free DeucesCracked Premium.
rakeback.deucescracked.com
Terms of Use &
Privacy Policy
| Affiliates
© 2013 DeucesCracked.com All Rights Reserved.
Reply to Topic
Reply w/Quote