
October 18th, 2011, 05:59 PM


Executive Member


Join Date: Apr 2006
Posts: 5,141


Quote:
Originally Posted by blackjack avenger
A point graph you plot the days in order of occurrence and then connect the dots?

so.... i wasn't doin it right then, no?
maybe more like this ...........?
but how's that ever gonna look like a bell even after decades of plays?

October 18th, 2011, 06:14 PM


Executive Member


Join Date: Apr 2006
Posts: 5,141


hmm, maybe this way?
http://support.microsoft.com/kb/213930
.......... ehhmm, that's gonna take me a while, lol................

October 18th, 2011, 07:15 PM


Executive Member


Join Date: Apr 2005
Location: Minnesota
Posts: 1,625


Late to the party, and didn't bring much, but...
sagefr0g:
Obviously, with BJ we can run a sim and get our expected win rate and standard deviation. With other games, like poker, and whatever it is you’re doing, we have to generate our statistics based on our own experience. As we’ve been reminded, with live play it’s unlikely we will ever have enough data to generate meaningful statistics. But still, we want to have something to look at.
I’ve never seen the use of STDEV recommended as a way to do this. In post #9 of this thread, statmanhal describes exactly how to set up a spreadsheet to calculate your hourly standard deviation. If you don’t want it to be hourly you can omit that part, or just plug in “1” for the session duration to pretend that each session was only one hour.
To get your bellshaped curve I think you’ll need to look at the number of days you get each kind of result. Say you have these kinds of results: Huge Loss, Moderate Loss, Average Outcome, Moderate Win, Huge Win. The number of days you have an average outcome should be the bulge in the middle. The number of days you have huge wins or losses should be the tails on the sides.

October 18th, 2011, 08:24 PM


Executive Member


Join Date: Apr 2006
Posts: 5,141


Quote:
Originally Posted by Canceler
sagefr0g:
Obviously, with BJ we can run a sim and get our expected win rate and standard deviation. With other games, like poker, and whatever it is you’re doing, we have to generate our statistics based on our own experience. As we’ve been reminded, with live play it’s unlikely we will ever have enough data to generate meaningful statistics. But still, we want to have something to look at.

exactly!
Quote:
I’ve never seen the use of STDEV recommended as a way to do this. In post #9 of this thread, statmanhal describes exactly how to set up a spreadsheet to calculate your hourly standard deviation. If you don’t want it to be hourly you can omit that part, or just plug in “1” for the session duration to pretend that each session was only one hour.

ahh ok, i haven't checked it entirely closely, but i think he's using essentially stdevp a function in excel, which is like the first example ie. (dividing by n) in this link:
http://standarddeviation.appspot.com/
stdev in excel would be like the second example (ie. dividing by (n1), ( i believe) where one doesn't know the entire population but just has a sample. so maybe i'd be best off using stdev.......
Quote:
To get your bellshaped curve I think you’ll need to look at the number of days you get each kind of result. Say you have these kinds of results: Huge Loss, Moderate Loss, Average Outcome, Moderate Win, Huge Win. The number of days you have an average outcome should be the bulge in the middle. The number of days you have huge wins or losses should be the tails on the sides.

there ya go, makes sense...... so technically i'd need one heck of a lot more data. and then would have to know how to do the math right in the first place, lol
edit: but anyway, some rainy day i think i'll try this method: http://support.microsoft.com/kb/213930
thank you
Last edited by sagefr0g; October 18th, 2011 at 08:33 PM.

October 18th, 2011, 10:18 PM


Executive Member


Join Date: Apr 2006
Posts: 5,141


Quote:
Originally Posted by Canceler
......
To get your bellshaped curve I think you’ll need to look at the number of days you get each kind of result. Say you have these kinds of results: Huge Loss, Moderate Loss, Average Outcome, Moderate Win, Huge Win. The number of days you have an average outcome should be the bulge in the middle. The number of days you have huge wins or losses should be the tails on the sides.

so, like for the image below, which is results over days.....
i can kind of just conceptualize, Huge Loss, Moderate Loss, Average Outcome, Moderate Win, Huge Win ... ect. , and at least get a 'handle' on how things have been going, sorta thing........ gives one an idea of what kind of dispersal of results has been happening, and maybe how it will tend to go in the future, and as time goes on and more data comes in, see how it all stacks up again, at least have a qualitative analysis if not quantitative, sorta thing, at least have something to look at and get a feel for how things are going, maybe stack that analysis up against what excel thinks is one standard deviation, which in this case was $265.27 and the expected value which is $65.83 ... to where for the time being at least, i can kind of say to myself, well as far as i know right now, i can expect to be within $200 or so plus or minus of a $65 result about 68% of the time...... no?
Last edited by sagefr0g; October 18th, 2011 at 10:26 PM.

October 18th, 2011, 10:43 PM


Executive Member


Join Date: Apr 2006
Posts: 5,141


Quote:
Originally Posted by blackjack avenger
....
If you don't know your SD hope your betting conservatively.
I dont know
1/4 Kelly comes to mind
Don't want the misses mad

along this line of reasoning, let me ask you a question, since you seem to know a lot about Kelly stuff.
well, first off, speaking of betting, Kelly stuff and bank roll, well the roll so far has more than doubled.
the question being, when you are 'properly' betting and making positive EV plays, errrhh well, i dunno, is there some significance to reaching the point where the roll is doubled, far as Kelly theory goes?
seems i've seen a lot of talk about the point where advantage players have doubled their roll, in conjunction with Kelly stuff. can you shed any light on that subject, ie. significance of doubling the bank roll and Kelly stuff?
have anything to do with N0, maybe?
edit: like here's an example where doubling the roll is mentioned:
That's why people advocate 1/2 kelly or 1/4 kelly. If you bet half of the kelly fraction, then you have way less chance of losing 50% of your bankroll. For example, a full Kelly better has 1/3 chance of losing 50% of his bankroll before doubling, while a 1/2 Kelly better has 1/9 chance of losing 50% of his bankroll before doubling. from this post: http://www.blackjackinfo.com/bb/show...04&postcount=2
just one of many examples where "doubling the roll" is emphasized when discussing Kelly stuff
Last edited by sagefr0g; October 18th, 2011 at 11:37 PM.

October 19th, 2011, 12:11 AM


Executive Member


Join Date: Apr 2005
Location: Minnesota
Posts: 1,625


I just used the Huge Loss, Moderate Loss, etc. to give you the concept of how to get your bellshaped curve. Not to imply that that kind of chart would have much value, but you did seem to have your heart set on it.
Nevertheless, here’s an example of what I had in mind. Being lazy, rather than make up fake data, I just used the outcomes of my last 100 sessions of 2/4 Limit Hold’em. (Astute observers will notice I was a loser at that game, as most people are, but never mind that!)
The result types are in $25 increments. Result Type A is a loss of $175 to $200. B is a loss of $150 to $174.99. H is a loss between $0 and $25. I is a win between $0 and $25. P is a win of $175 to $200.
The columns represent the number of sessions I had of each result type. With so few data points, the chart is only vaguely bellshaped, but you get the idea. Again, this type of chart is of little value. Your SD numbers will be more useful, I think.

October 19th, 2011, 12:38 AM


Executive Member


Join Date: Apr 2006
Posts: 5,141


Quote:
Originally Posted by Canceler
I just used the Huge Loss, Moderate Loss, etc. to give you the concept of how to get your bellshaped curve. Not to imply that that kind of chart would have much value, but you did seem to have your heart set on it.

yah, lol, it's virtually all i got, for now, lol
excepting, i do know the EV.
just hoping to get some kind of a handle on the swings and how they may come down, regardless of how hazy that handle may be.
but i get your point.
Quote:
Nevertheless, here’s an example of what I had in mind. Being lazy, rather than make up fake data, I just used the outcomes of my last 100 sessions of 2/4 Limit Hold’em. (Astute observers will notice I was a loser at that game, as most people are, but never mind that!)
The result types are in $25 increments. Result Type A is a loss of $175 to $200. B is a loss of $150 to $174.99. H is a loss between $0 and $25. I is a win between $0 and $25. P is a win of $175 to $200.
The columns represent the number of sessions I had of each result type. With so few data points, the chart is only vaguely bellshaped, but you get the idea. Again, this type of chart is of little value. Your SD numbers will be more useful, I think.

yup, got it, definitely makes it clear how the bell curve is arrived at, thank you.

October 19th, 2011, 04:07 AM


Executive Member


Join Date: Feb 2007
Posts: 2,267


floating point graph?
Quote:
Originally Posted by sagefr0g
so.... i wasn't doin it right then, no?
maybe more like this ...........?
but how's that ever gonna look like a bell even after decades of plays?

I am sure there is a real name for this graph, but it escapes me.
Froggie
Start the graph at starting bank in the middle of the Y verticle axis, then let the results be plotted & added & subtracted at each point. Then connect the dots. At the final dot your bank should be at its current level. A nice thing, this graph can be done with regular notebook or graph paper. Also, this graph can show results over time, an example if you change strategy.
I started a thread in "other games" you may find interesting.
Last edited by blackjack avenger; October 19th, 2011 at 06:55 AM.

October 19th, 2011, 11:52 AM


Executive Member


Join Date: Feb 2007
Posts: 2,267


[QUOTE=sagefr0g;257339]along this line of reasoning, let me ask you a question, since you seem to know a lot about Kelly stuff.
well, first off, speaking of betting, Kelly stuff and bank roll, well the roll so far has more than doubled.
the question being, when you are 'properly' betting and making positive EV plays, errrhh well, i dunno, is there some significance to reaching the point where the roll is doubled, far as Kelly theory goes?
seems i've seen a lot of talk about the point where advantage players have doubled their roll, in conjunction with Kelly stuff. can you shed any light on that subject, ie. significance of doubling the bank roll and Kelly stuff?
have anything to do with N0, maybe?
Why talk about doubling bank? Are you retiring? So assuming one is going to play on the risk of drawdown numbers approach the infinite numbers quickly.
Chances of losing % of bank with Kelly resizing:
50% chance of losing 50% of bank with Kelly
80% chance of losing 20% of bank with Kelly
Chances of losing % of bank with 1/4 Kelly:
.08% chance of losing 50% of bank with 1/4 Kelly
21% chance of losing 20% of bank with 1/4 Kelly
Begs the question. Why would anyone play a positive expectation game with an ror? Now, what if one figured the advantage correctly on an unsure game but bet kelly? They face a 50% chance of losing half, if that happens I would think confidence would suffer.
Kelly & N0? Kelly raises N0 by a factor of 9. However betting 1/4 Kelly does not raise N0 greatly though growth suffers.

Thread Tools 

Display Modes 
Linear Mode

Posting Rules

You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off



All times are GMT 6. The time now is 08:25 PM.
