Another long run question

[Stylee]

This may be a simple matter of correcting me on my math that I'm not very good at. I very well might not have even used the calculator right so be gentle.:laugh:

On CVCX online risk calculator I entered 5000 bankroll, and used the example of the KISS1 count from pg. 150 of Blackjack Bluebook 2 and to enter 5.5 win rate per 100 hands (assuming $5 unit and $11/hour or 100 hand win rate) and standard deviation of 54 units. After 50,000 hands played my ROR is 0.000%.

Now, if I enter the bankroll as 1000 and hands played at 10,000, my ROR is .670%. Does the original non-existent ROR eliminate the possibility of only having 1/4 of my bankroll after 1/4 of the hands played since my ROR is then existent, or am I missing something? What am I not understanding?

 

[Kasi]

This may be a simple matter of correcting me on my math that I'm not very good at. I very well might not have even used the calculator right so be gentle.:laugh:

On CVCX online risk calculator I entered 5000 bankroll, and used the example of the KISS1 count from pg. 150 of Blackjack Bluebook 2 and to enter 5.5 win rate per 100 hands (assuming $5 unit and $11/hour or 100 hand win rate) and standard deviation of 54 units. After 50,000 hands played my ROR is 0.000%.

Now, if I enter the bankroll as 1000 and hands played at 10,000, my ROR is .670%. Does the original non-existent ROR eliminate the possibility of only having 1/4 of my bankroll after 1/4 of the hands played since my ROR is then existent, or am I missing something? What am I not understanding?

Which calculator are you using? Maybe win rate is 2.2 units not 5.5? If unit is $5 and roll is $5000, maybe try putting 1000 in for the number of units in roll?

 

[EasyRhino]

Okay, the key indeed was the change in bankroll amount. (I went to the site and ran the same numbers)

http://www.qfit.com/blackjack-calculator-c3.htm

Let's refer to your figures at "FullBR" and "20%BR". With "FullBR" indeed, the calculator spits out a 0% (or at least below three decimals) chance of losing it all

But, as noted, there is still a (small) chance of losing 20% of that bankroll in 20% the number of hands.

This difference in behavior between 10,000 hands and 50,000 hands is indeed the "Long Run" asserting itself. Numbers that big are where it really becomes apparent.

 

[Kasi]

Okay, the key indeed was the change in bankroll amount. (I went to the site and ran the same numbers)

http://www.qfit.com/blackjack-calculator-c3.htm

Let's refer to your figures at "FullBR" and "20%BR". With "FullBR" indeed, the calculator spits out a 0% (or at least below three decimals) chance of losing it all

But, as noted, there is still a (small) chance of losing 20% of that bankroll in 20% the number of hands.

This difference in behavior between 10,000 hands and 50,000 hands is indeed the "Long Run" asserting itself. Numbers that big are where it really becomes apparent.

What do you think he meant by "5.5 win rate per 100 hands (assuming $5 unit and $11/hour or 100 hand win rate) and standard deviation of 54 units. "?

The calculator I'm looking at wants one's roll in units and one's win rate per 100 in units? What game wins 5.5 units an hour? Doesn't make sense to me.

 

[Stylee]

What do you think he meant by "5.5 win rate per 100 hands (assuming $5 unit and $11/hour or 100 hand win rate) and standard deviation of 54 units. "?

The calculator I'm looking at wants one's roll in units and one's win rate per 100 in units? What game wins 5.5 units an hour? Doesn't make sense to me.

You're right, that should be 2.2 units per hour, dark keypads make great mistakes

Let's refer to your figures at "FullBR" and "20%BR". With "FullBR" indeed, the calculator spits out a 0% (or at least below three decimals) chance of losing it all

But, as noted, there is still a (small) chance of losing 20% of that bankroll in 20% the number of hands.

I guess the part I'm having trouble with is that after loosing 80% of my full bankroll, the remaining 20% IS my full bankroll now which now has a different ROR. Looking back I might have used the wrong calculator, but it kinda serves my purpose. If your ROR changes evertime your bankroll changes, doesn't the "long run" goal also change or start over? If it takes 50,000 hands for variance to even out when betting a certain way with a certain bankroll, does a change in betting or bankroll require a new 50,000 hands for variance to even out in that set of variables?

 

[Kasi]

You're right, that should be 2.2 units per hour, dark keypads make great mistakes

I guess the part I'm having trouble with is that after loosing 80% of my full bankroll, the remaining 20% IS my full bankroll now which now has a different ROR. Looking back I might have used the wrong calculator, but it kinda serves my purpose. If your ROR changes evertime your bankroll changes, doesn't the "long run" goal also change or start over? If it takes 50,000 hands for variance to even out when betting a certain way with a certain bankroll, does a change in betting or bankroll require a new 50,000 hands for variance to even out in that set of variables?

Now it makes sense to me lol. That's what I thought.

But, yes, if you now have a 200 unit roll instead of a 1000 unit roll, your ROR is different from that point forward. It'd be like you had started with a 200 unit roll in the first place.

All adding more units to a roll does is lower ROR. As long as you bet the same dollars at the same points, your 2.2 units per 100, stays the same. Your EV per 100 stays the same. Your N0 stays the same. Your stan dev per 100 stays the same. Your avg bet stays the same. Naturally it will take longer to double roll with a larger bankroll since you only will make $11 per 100 no matter what size roll is if you bet same dollars at same points.

In other words if you had started with $100K instead of $5K, you still would have lost the same $4K in however many hands. That result, any result, would still be the same number of stan dev above or below expectation after the same number of hands played. 50,000 hands will have an EV and stand dev associated with 50000 hands whether you started with a million or a thousand bucks. But your chances of being able to reach 50000 hands of play changes.

And, of course, your ROR over a finite number of hands will always be less than your original lifetime ROR.

The only difference is that in one case you have $1K to recover with vs $96K to recover with assuming no re-sizing of dollar bets. Which is the same thing as assuming a different number of units in roll. So, if you could change your unit to $1 with $1000 left, you would have same ROR as $5 with $5K since both rolls have 1000 units. You would still make 2.2 units per 100 but now it would be $2.20 per 100 instead of $11.

Hope that helps answer your question.

As an aside, there are ways to calculate the chances of losing x percent of your starting roll at some point.

No don't ask me lol. I just know there are ways lol. But I think a Kelly better probably has a 20% chance of losing 80% of roll at some point.

Do you know what your original ROR was? Guessing a little higher than Kelly?

If this is for real and you know how many hands you lost your $4K over, you can figure out how "unlucky" you were. If that turns out that you were very very "unlucky", maybe look elsewhere to explain the results.

 

[EasyRhino]

If your ROR changes evertime your bankroll changes, doesn't the "long run" goal also change or start over?

A great number of calculators, including those CVCX ones, assume a bet spread and game that never changes. So there, the ROR is measured from your initial starting point.

So, if you lost, say 90% of your bankroll, and kept the same bet spread, then your "localized" ROR would indeed be much higher. However, you'd also need to acknowledge what the odds were to lose 90% of the BR in the first place.

Another way to manage things, though, is to resize your bets with every bankroll change. If you were to bet a precise full-Kelly fraction every hand, this would mean that your chance of losing it all would be theoretically zero, but your chance of losing a significant portion would be fairly high.

 

[Stylee]

In other words if you had started with $100K instead of $5K, you still would have lost the same $4K in however many hands. That result, any result, would still be the same number of stan dev above or below expectation after the same number of hands played. 50,000 hands will have an EV and stand dev associated with 50000 hands whether you started with a million or a thousand bucks. But your chances of being able to reach 50000 hands of play changes.

Thanks for the input. Makes more sense now, and it would probably make even more looking into Kelly betting.

Another way to manage things, though, is to resize your bets with every bankroll change. If you were to bet a precise full-Kelly fraction every hand, this would mean that your chance of losing it all would be theoretically zero, but your chance of losing a significant portion would be fairly high.

That kinda agrees with my original thoughts that to be constantly perfectly accurate you would need to calculate ROR after each hand.

It seem with this information that the long run is kind of theoretical. You can play one way for 100,000 hands and reach the long run, but if your play style changes, it's a new set of 100,000 hands you are playing towards to reach the new long run. Am I close on this? Am I relating ROR and the long run variance eliminator too closely?

 

[Kasi]

It seem with this information that the long run is kind of theoretical. You can play one way for 100,000 hands and reach the long run, but if your play style changes, it's a new set of 100,000 hands you are playing towards to reach the new long run. Am I close on this? Am I relating ROR and the long run variance eliminator too closely?

I guess "long run" is theoretical and could mean different things to different people.

To some it maybe is when your EV equals 1 stand dev. This might be 20,000 hands or 100,000+ hands depending. To others maybe it's 2 stan dev. Maybe to some when roll is doubled because if your original ROR was 1 in 20 after doubling now it's 1 in 400 and the chances of ever losing all your roll become increasingly unlikely.

To some maybe they don't care about long run so much as chances of increasing roll by 50% in the next 40 hours.

Sure if your "play style" changes like you go from play-all to wonging or switch to a different game, etc it's a brand new ball game. You could still likely bet your money to the same ROR as in the first game but EV and stan dev change too making the "long run" by one of the above definitions different too.

Anyway, best is know what to expect from betting before you do it.

 

[Stylee]

Yeah thats what I figured, just wanted to make sure. I see a lot of people talking about the long run as if it were some magical point when you stop losing

 

[blackjack avenger]

The Long Run, Run

If you change your style of play you are not really starting over. The two styles of play become blended.

The long run is sort of a magical number when your expectation (skill) overcomes your standard deviation (luck).

If you could play a 1000 hands a second with proper play you would love this game.

 

[Kasi]

I see a lot of people talking about the long run as if it were some magical point when you stop losing

I know what you mean lol

. At any point in time there is always some liklihood of losing entire original roll no matter how far ahead you might be at the time if you played forever. Easily measurable at any point in time so you know if it's 1 in 1000 or whatever. Or maybe one chooses to ask how likely is it I could lose it all or some portion in the next 50,000 hands, etc.

So long run is basically a pretty vague term to me lol.

 

[Kasi]

If you change your style of play you are not really starting over. The two styles of play become blended.

The long run is sort of a magical number when your expectation (skill) overcomes your standard deviation (luck).

If you could play a 1000 hands a second with proper play you would love this game.

Not sure I understand how 2 different styles of play become blended?

I guess 1000 hands a second would be nice of course but 50000 hands later nothing has really changed except that 50000 hands happened faster.

 

[blackjack avenger]

NO NO NO Long Run!!!!!

If you play different games you have one global NO or long run. It is a blend of the various NO's of the different games you are playing. The previous hands are booked. The results would cluster around the game you play the most and/or probably the game with the most variance. Even if you once played small but because of a cash windfall you can now play big. Those small hands still matter, though not much and will diminish with time.

The more hands you play the more skill overcomes variance. If you played a 1000 hands a second correctly you probably go home every weekend a winner.

 

[Kasi]

If you play different games you have one global NO or long run. It is a blend of the various NO's of the different games you are playing. The previous hands are booked. The results would cluster around the game you play the most and/or probably the game with the most variance. Even if you once played small but because of a cash windfall you can now play big. Those small hands still matter, though not much and will diminish with time.

The more hands you play the more skill overcomes variance. If you played a 1000 hands a second correctly you probably go home every weekend a winner.

Well I guess I just view it as each game has it's own unigue N0. One day you're playing a game with an N0 of 600,000 hands and the next day one with an N0 of 5000 hands. Just seems simpler to keep track of each different game separately.

If you play a game spreading $10-$160 with $10K, the next day play the same game spreading $50-$800 with the same $10K, the next day play either spread, or another, to a $100K roll because you got rich the N0 for that game is the same number of hands in all those cases and hasn't changed at all as long as you bet the same 16 units in the same place.

If you play each weekend at 1000 hands a second with a 10% ROR, you'll lose your original roll 1 weekend in 10. If with a 90% ROR, 9 weekends in 10.

But at least you'll know which has happened in a half-hour lol.

 

[rukus]

Well I guess I just view it as each game has it's own unigue N0. One day you're playing a game with an N0 of 600,000 hands and the next day one with an N0 of 5000 hands. Just seems simpler to keep track of each different game separately.

I think all BJAvenger is saying is that if you played for instance one game/style with N0 of 600 hours 50% of the time and one game with N0 of 300 hours the other 50% of the time, your real N0 should be a weighted average of the two, or 0.5*600 + 0.5*300 = 450 hours. Yes it makes sense to track each game/style's N0 (or 4N0 or 9N0 or whatever you prefer) for your own purposes to choose between games/styles to play/use, but in the end what counts when determining skill vs luck is your total overall expected N0, which would be the 450 hours in the example above.

If you play a game spreading $10-$160 with $10K, the next day play the same game spreading $50-$800 with the same $10K, the next day play either spread, or another, to a $100K roll because you got rich the N0 for that game is the same number of hands in all those cases and hasn't changed at all as long as you bet the same 16 units in the same place.

just want to clarify for others that a 1-16 spread in "play all" has a MUCH different N0 than 1-16 spread while wonging in/out at say +1. so it is not just spread that determines your N0.

 

[Kasi]

I think all BJAvenger is saying is that if you played for instance one game/style with N0 of 600 hours 50% of the time and one game with N0 of 300 hours the other 50% of the time, your real N0 should be a weighted average of the two, or 0.5*600 + 0.5*300 = 450 hours.

I guess you and he mean if he played 50% of the N0 hours at a 300 hour N0 game (150 hours) and 50% of the N0 hours in the 600 hour N0 game (300 hours). Not 50% of my total hours played devoted to playing each game.

If I played 300 hours at each game, splitting my time played between them equally, I have N0 in one game but still 300 hours away in the other. If I never play that game again for the rest of my life, I will always be 300 short of "blended" N0 and never attain it by this approach even if I attain N0 in the next 50 different N0 games I play. I'd much rather know I achieved N0 in 50 of 51 games than think I'm still somehow 300 hours short of my lifetime N0.

I shudder to think of the calculations involved over a lifetime playing hundreds of different styles perhaps lol.

Not to mention maybe I'm playing one of those games with a lower ROR and one with a higher one while playing equally skillfully in both. So never achieving N0 in a game isn't even necessarily a reflection of the skill of your play.

 

[EasyRhino]

If I played 300 hours at each game, splitting my time played between them equally, I have N0 in one game but still 300 hours away in the other. If I never play that game again for the rest of my life, I will always be 300 short of "blended" N0 and never attain it by this approach even if I attain N0 in the next 50 different N0 games I play.

That would seem to be unnecessarily bucketizing things. Assuming it's your money, it's all basically one long session and one long game. And reaching N0 on a game you were playing five years ago for six months at small stakes isn't going to matter a hill of beans compared to a game that have been played for longer times or higher stakes.

 

[Kasi]

That would seem to be unnecessarily bucketizing things. Assuming it's your money, it's all basically one long session and one long game. And reaching N0 on a game you were playing five years ago for six months at small stakes isn't going to matter a hill of beans compared to a game that have been played for longer times or higher stakes.

No big deal - I'd just care about N0 for a particular game with a particular spread kind of thing. If I play a different game with a different N0, I'd try to keep track of hands played for that game. So if you meant achieving N0 in a game 5 years ago is irrelevant to a different game played today, I guess I agree. If you are still playing that same game in the same way, or a game even similar, 5 years later then "longer times" and "higher stakes" doesn't really matter since NO is a number of hands played and going from a $5 unit to a $500 unit doesn't really matter either.

But if you imply by "it's all basically one long session and one long game" as if you've been playing the same game the same way, or maybe even similar games, if you go for some "blended" thing, that all would have about the same N0, say just for example a range of 40000 to 60000 hands, then if you've played 60000 hands, hopefully you know then, one way or another, whether or not your EV has overcome 1 standard deviation. Playing another 600,000 hands of those games in a lifetime won't change that.

Want to know if your lifetime N0 has been achieved or not? Just add up the dollar EV and dollar variance of every hand you've ever played I'd think.

That's my view on N0 and it won't change until tomorrow lol.

Love to hear from you/anyone, the people actually doing this stuff on a regular basis, go about determining N0, if at all, how important they think it is as something to measure results by etc etc.

 

[rukus]

I guess you and he mean if he played 50% of the N0 hours at a 300 hour N0 game (150 hours) and 50% of the N0 hours in the 600 hour N0 game (300 hours). Not 50% of my total hours played devoted to playing each game.

.........

Love to hear from you/anyone, the people actually doing this stuff on a regular basis, go about determining N0, if at all, how important they think it is as something to measure results by etc etc.

no, i meant 50% of your total hours played devoted to each game. your overall N0 would be 450. if you split your time equally between these two games in overall hours, you will have some overall expectation (some combination between the two games types), standard deviation (some combination between the two games types), and thus N0. it would take you 450 hours to reach one blended standard deviation in EV. now that i think about it, the blended N0 is not 450 hours because you need to weight EV by 50% each and then VARIANCE by 50% each (and then take the square root of this variance) before calculating the blended N0... but still my overall point still stands - you weight your various game types by % time you play them out of your total playing time, and then you calculate a blended EV, Std Dev, and N0.

as for me and what i use N0 for, i pretty much use it for count system, game and playing style selection. even way before i reach N0 i am constantly comparing my results with EV and do not wait until i reach one std deviation of EV just to say, "Ok, ive now reached the long run and only now is it time to compare my play vs EV".