I see so many people on here talk about the "long run" that it makes my head spin. If you're trying to determine whether your results are within normal parameters based on "the long run" can't we just consider anything having a margin of error of less than 1% the "long run" since it is very unlikely our results will be outside of this. To get a margin of error of less than 1% I think you need a sample size of like 50,000 or so. I forget what the exact # is but I would then conclude that whatever your results are after say 50,000 hands, it's safe to say if you took the average of those 50,000 hands, that's what you could expect to continue to do on average.
the "Long Run"
- Thread starter Thunder
- Start date
ccibball50
Well-Known Member
Well putThunder said:I see so many people on here talk about the "long run" that it makes my head spin. If you're trying to determine whether your results are within normal parameters based on "the long run" can't we just consider anything having a margin of error of less than 1% the "long run" since it is very unlikely our results will be outside of this. To get a margin of error of less than 1% I think you need a sample size of like 50,000 or so. I forget what the exact # is but I would then conclude that whatever your results are after say 50,000 hands, it's safe to say if you took the average of those 50,000 hands, that's what you could expect to continue to do on average.
Makes my head spin too lol.Thunder said:I see so many people on here talk about the "long run" that it makes my head spin. If you're trying to determine whether your results are within normal parameters based on "the long run" can't we just consider anything having a margin of error of less than 1% the "long run" since it is very unlikely our results will be outside of this. To get a margin of error of less than 1% I think you need a sample size of like 50,000 or so. I forget what the exact # is but I would then conclude that whatever your results are after say 50,000 hands, it's safe to say if you took the average of those 50,000 hands, that's what you could expect to continue to do on average.
Unfortunately, what if actual results after 50,000 hands are 2 or 3 Stan Dev below or above EV?
It certainly can happen, and, if it does, your actual results are nowhere near within 1% of what EV "should" be. They actually are lots of %'s away from what one would expect.
You don't know what to expect just by playing 50,000 rounds in a consistent way - you know what to expect from telling a sim to run a couple billion rounds playing that same way.
You need a sim to tell what to expect so you can measure your actual results versus what it tells you to expect after so many rounds.
Expecting that 50,000 rounds actually played would represent anything close to what actually may be expected from 50,000 rounds is asking a bit much. In alot of cases lol.
Make any sense?
Well video poker is a different animal altogether since you have royal and straight flushes which make up a big part of your EV but very rarely happen.shadroch said:50,000 rounds in video poker is not all that much more than one cycle. Thats no where near the long run. I'd think the same applies to BJ.
Well, typically, "longrun" implies, to me anyway, a +EV game - the point at which one is more or less guaranteed to win money forever.shadroch said:50,000 rounds in video poker is not all that much more than one cycle. Thats no where near the long run. I'd think the same applies to BJ.
There aren't that many +EV video poker games, are there?
But if you do mean some full-pay Jacks or Better VP or something, tyhen, yes, you could figure out when EV would equal 1 SD and how much roll you'd need to have a chance of playing that many rounds.
I guess, strictly speaking, roll is irrelevant for how long it will take to achieve N0. It's only relevant for giving you how much chance to stick around long enough to achieve it.
Yeah it makes perfect sense. BJ I guess is a little different than say something where the variance is much smaller such as flipping a coin. I guess you'd need more like 1,000,000 hands or so to be within 1% of your expected EV. I know that after say 3 hours of play, you could easily be 3 standard deviations or more from your expected EV as evidenced by the fact that someone playing a 6 deck AC rules game with 75% penetration and a 1-10 spread probably shouldn't expect to make more than $10/hr counting at a $10 table, but yet there will be days where after 1 hour you're up $150.Kasi said:Makes my head spin too lol.
Unfortunately, what if actual results after 50,000 hands are 2 or 3 Stan Dev below or above EV?
It certainly can happen, and, if it does, your actual results are nowhere near within 1% of what EV "should" be. They actually are lots of %'s away from what one would expect.
You don't know what to expect just by playing 50,000 rounds in a consistent way - you know what to expect from telling a sim to run a couple billion rounds playing that same way.
You need a sim to tell what to expect so you can measure your actual results versus what it tells you to expect after so many rounds.
Expecting that 50,000 rounds actually played would represent anything close to what actually may be expected from 50,000 rounds is asking a bit much. In alot of cases lol.
Make any sense?
Figuring 85 hands per hour (since $10 tables are generally well populated), 3 hours of play spreading $10-to-$100 produces 1 standard deviation of just about $600 (depending upon the quickness of your betting ramp). Your EV for that period would be around +$45 (depending upon the count system). It means that about 1 session out of 6, you'll finish down more than $555 -- and another 1 session out of 6, you'll win more than $645.Thunder said:I know that after say 3 hours of play, you could easily be 3 standard deviations or more from your expected EV as evidenced by the fact that someone playing a 6 deck AC rules game with 75% penetration and a 1-10 spread probably shouldn't expect to make more than $10/hr counting at a $10 table, but yet there will be days where after 1 hour you're up $150.
Do this twice a week for a year and your EV becomes 100 times as large (+$4500), but 1 standard deviation becomes only 10 times as large ($6000). So now, 1 year out of 6 you'll be down more than $1500 -- and another 1 year out of 6 you'll be up more than $10,500.
After 21 months of this, your EV and 1 standard deviation will both be about $7900. Now you've got 5 chances out of 6 (1 standard deviation) to be ahead of the game.
ccibball50
Well-Known Member
Ok so if you are off more than 1% after 50,000 hands, start all over again. the chances of 100,000 is 1/2 do it again and the fractions keep getting smaller. eventually your odds will be minute, that being outside that one percent would be equivelent to winning the lottery 3 times in the same month on the same machine. Its just not going to happen like that. Therefore the longrun is abtainable. If you are down 3%, you are probably tipping too much, or messing up on your strategy.Kasi said:Makes my head spin too lol.
Unfortunately, what if actual results after 50,000 hands are 2 or 3 Stan Dev below or above EV?
It certainly can happen, and, if it does, your actual results are nowhere near within 1% of what EV "should" be. They actually are lots of %'s away from what one would expect.
You don't know what to expect just by playing 50,000 rounds in a consistent way - you know what to expect from telling a sim to run a couple billion rounds playing that same way.
You need a sim to tell what to expect so you can measure your actual results versus what it tells you to expect after so many rounds.
Expecting that 50,000 rounds actually played would represent anything close to what actually may be expected from 50,000 rounds is asking a bit much. In alot of cases lol. I played over 700 hours last year as a part time player. at 100 hands an hour, thats 70,000 and I rarely play unless i am getting 100 to 180 hands an hours.
Make any sense?
Last edited:
ccibball50
Well-Known Member
You cant easily be 3 standard diviations away. It is possible, but rare. YOu have to know what a standard diviation is. It is a probability percentage that the outcome will come a certain amount higher and a certain amount lower than the EV.Thunder said:Yeah it makes perfect sense. BJ I guess is a little different than say something where the variance is much smaller such as flipping a coin. I guess you'd need more like 1,000,000 hands or so to be within 1% of your expected EV. I know that after say 3 hours of play, you could easily be 3 standard deviations or more from your expected EV as evidenced by the fact that someone playing a 6 deck AC rules game with 75% penetration and a 1-10 spread probably shouldn't expect to make more than $10/hr counting at a $10 table, but yet there will be days where after 1 hour you're up $150.
you have the same chance to be outsid 2 standard diviations in the long run as you do in the one session. How many times have you been outside 2 standard diviations? I have never been outside 2 SD and I played over 700 hrs. (this is highly unusual I know) But you should only do it a couple times out of every 100 trips.
This is true. You should be outside 2 standard deviations on the minus side once every 44 times -- be it a session, a year, whatever. You should be outside 1 SD on the minus side once every 6 times, and outside 3 SD's once every 770 times on the minus side.ccibball50 said:You have the same chance to be outsid 2 standard diviations in the long run as you do in the one session. I have never been outside 2 SD and I played over 700 hrs. (this is highly unusual I know) But you should only do it a couple times out of every 100 trips.
There is a 10% chance you'll be outide 1.28 SD's on the minus side, and a 5% chance to be outside 1.64 SD's on the minus side.
Renzey can you elaborate on how do you get these numbers?Renzey said:This is true. You should be outside 2 standard deviations on the minus side once every 44 times -- be it a session, a year, whatever. You should be outside 1 SD on the minus side once every 6 times, and outside 3 SD's once every 770 times on the minus side.
There is a 10% chance you'll be outide 1.28 SD's on the minus side, and a 5% chance to be outside 1.64 SD's on the minus side.
My understanding was that 2SD refers to the 95% confidence interval and 3SD to the 99% confidence interval.
Thus 68% of sessions fall within EV +/- 1SD
95% of sessions fall within EV +/- 2SD (i.e. 27% of time you will be between 1SD and 2SD)
99% of sessions fall within EV +/- 3SD (i.e. 4% of time you will be between 2SD and 3SD)
Am I seeing this incorrectly?
Don't forget that there is a minus side and a plus side to the bell shaped curve. It's true, 2 SD's covers 95.4% of all results, but the outside remainder gets split evenly between the plus and minus sides. Of course, in risk taking, you're not concerned with the plus side, so it's often forgotten about and mistakenly thought of as being grouped in with the minus side. In reality, it leaves 2.3% each for the plus and minus sides outside of 2 SD's.matt21 said:Renzey can you elaborate on how do you get these numbers?
My understanding was that 2SD refers to the 95% confidence interval and 3SD to the 99% confidence interval. Thus 68% of sessions fall within EV +/- 1SD 95% of sessions fall within EV +/- 2SD (i.e. 27% of time you will be between 1SD and 2SD) 99% of sessions fall within EV +/- 3SD (i.e. 4% of time you will be between 2SD and 3SD) Am I seeing this incorrectly?
With 1 SD, it's 68.2% in the middle with 15.9% each on the plus and minus sides. With 3 SD's, it's 99.74% in the middle with 0.13% each on the plus and minus sides.
Statistics books also have tables of "Z" Scores, telling what fraction of SD's are equal to a particular probability of occurance. That's where the "1.28 SD's equals 10% probability" and "1.64 SD's equals 5% probability" comes from.
Last edited: