[Closed] How to effectively lose allins.

try new room ♦ ...and when you are allin don't think negatively

How did you calculate the distribution function given that various different situations create completely different equity values and therefore probabilities of winning?

Originally posted by StoneJ
How did you calculate the distribution function given that various different situations create completely different equity values and therefore probabilities of winning?

Well, for each game I have an equity. I have 130 games. I create 130 random results of those games based on the equity. And so I get a possible number of games won for a given simulation. Then I run a lot (like 100000) simulations. At that point I can (I hope I can) estimate that if there are N simulations where I would won # times, then probability of winning # times given my set of games is about N/100000. So this is on the first graph.
Then on the second graph (which I called a distribution function, but I'am not sure that it is the correct name for it) each value is an estimated probability of winning the given number of games or less. It is calculated, by summing up all values from the first graph up to a given point. On the second graph points are connected, but it has no meaning, I've just chosen a wrong chart type in Excel, there should just be points, no lines.
Hope this clarifies. What is a bit astonishing is the uniformity of the graphs, I'am not really sure why do they look that way.

the graph is uniform because it is a gaussian distribution aka normal distribution, or the bell curve. I understand the thought process but i think the application is wrong. I would have thought a binomial distribution would be more appropriate given the response data is binary (success/fail).

For the simulations where you calculate randomly how many success/failures out of 130 you have do you simulate 130 different equities based on the 130 hands you played all in or have you calculated an average probability and run 130 with this probability?

At first I was suprised, but now when I think about it seems clear that the distribution would be normal. I don't quite understand the remark about binominal distribution it does not seem to fit the situation.

Anyway I did not calculate an average probability, but for calculating an outcome of each game I used the corresponding equity. At the moment when I think about it the result would be the same. Below is a graph that shows how my allin decisions are distributed over equity ranges.

The last datapoint actually does not describe a range, but a single point with games where I have 100% equity and cannot lose. The graph actually shows that I'am not a complete looney, but also that it is not like I only go allin with nuts (maybe I should work on that). Not all information is available here because the graph does not say anything about the pot odds.

And here another graph which shows how my percentage of winning looks in various equity ranges.

Definitely I'am not doing well when I'am 60 to 90 percent favourite. Maybe I should not go allin with QQ against AQ than?

stop saying (or thinking) "one time!", problem solved.

Post a graph with EV versus money won/lost in different relative equity ranges.

As to distribution I'am not so sure any more. Actually after asking some people and reading some Wikipedia I'am more inclined to guess that distribution of number of games won is binominal. Anyway what I've calculated and graphed is actually an empirical distribution not a teoretical one, but of course the teoretical question is still valid.

>stop saying (or thinking) "one time!", problem solved.
I don't quite get that advice, could you please maybe write a sentence or two of explanation?

>Post a graph with EV versus money won/lost in different relative equity ranges.

I don't know what is "relative equity" and how to calculate it, but I've created a graph which shows EV versus money won in *equity* ranges (version of Excel is different than previously and so is the graph).

It actually shows that my earnings are below EV in every equity range besides the 100% point :-). With that one it would be really hard to achive a loss. Anyway all that graph making helped me to get some optimism and stop thinking about bad luck. But I have to admit I did not have a good feeling for how big variance in poker can be.

That's what I meant.

Still it's not what I think I want. What would be useful would be a graph where you could see what the average effect of a decision is compared to its expected value. I expected this graph to show it, but that's not true, because the 0% EV could include hands where you go all in with a full house against quads and the 10% EV range could include spots where you get incredible pot odds.

Your graphed distribution is theoretical, not empirical. You simulated 100,000 different outcomes. Asymptotic theory show that as n tends to infinity certain distributions arise out of certain data. Binomial distribution is definitely the appropriate distribution to use to model such data. By taking 100,000 values you take a sample size that is so big that it removes the skew and you get an approximate normal distribution.

your graphed distribution would be empirical rather than theoretical if you only graphed your actual results, not simulated.

Agree with StoneJ and binomial is the way to go. Without being derogatory you seem to have confused yourself a bit.
We have a saying in Aikido; 'A blue belt knows just enough to get himself into trouble, but not enough to get out of it.' This would appear to be the problem here. I suggest a rethink about method, no offence meant.

Well guys(girls?), thanks for comments and discussion. I posted my results because I was quite astonished when I did some calculations and saw how exactly unlucky I had been and that it was not just my bad feeling. It is important to realize (and I'am a living proof) that you can play quite some hands, make good decisions and still be about 20 buy-ins below what one would expect. And it is just an inherent property of the game, maybe a bit extreme, but not that much unusual.

As for the distribution name and properties, sure I got confused, but that is not a big deal. It helped me to rethink the issue, read a bit, talk with my friends who have better grasp of statistics and to get better understanding of those things and refresh some information that was long gone from my memory. And of course last but not least to get my mind off the playing and losing thing.
But that does not change the fact that whatever the name of the distribution of the number of allin games won is, I find myself at the very left side of it. I just hope that after some time I can switch places with somebody at the right side :).

if i were you, i'd spend some time working on my game then trying to analyse my luck.
well you certainly would have brushed up on your mathematical skills

Originally posted by grtwrsw
Well guys(girls?), thanks for comments and discussion. I posted my results because I was quite astonished when I did some calculations and saw how exactly unlucky I had been and that it was not just my bad feeling. It is important to realize (and I'am a living proof) that you can play quite some hands, make good decisions and still be about 20 buy-ins below what one would expect. And it is just an inherent property of the game, maybe a bit extreme, but not that much unusual.

As for the distribution name and properties, sure I got confused, but that is not a big deal. It helped me to rethink the issue, read a bit, talk with my friends who have better grasp of statistics and to get better understanding of those things and refresh some information that was long gone from my memory. And of course last but not least to get my mind off the playing and losing thing.
But that does not change the fact that whatever the name of the distribution of the number of allin games won is, I find myself at the very left side of it. I just hope that after some time I can switch places with somebody at the right side :).

20 BI's below cant be quite extreme. I mean if you read other blogs and posts you can see people have worse cases.