Any one done sims on this possibly new method?

I have been doing some programming and simulations on various things to do with blackjack and believe I may have found a new method of winning at blackjack - I don't know if anyone has pointed this out before.

Basically I was looking at the distribution of counts occuring in a 6 deck shoe. It is obvious that the count must be zero and the start and zero at the end of shoe. It is also therefore obvious that the maximums (highest high counts and lowest low counts) mostly appear at the middle of the shoe.

Further more we know that a card count increases or decreases the house edge. It is therefore statistically possible to calculate the most probable count at the middle of a shoe, based on the outcome (win/lose ratio) of previous hands in the shoe. eg. if we have won 20 out of 25 hands, its much more likly there is a good high count that if we got very lucky on a low count.

Most importantly for any shoe with a good count halfway though - it will have on average a good count for every remaining hand in the shoe. We know this is and it is the basis for the unbalanced counts like front count etc.

So if we have a high win/loss ratio at the beginning of a shoe, its is likly that the count is high, and that we will have a favourable count for the remainder of the shoe, so we should increase our bet. Like wise a large number of looses in a shoe is likly to have occured from negetive count that will stay on average negetive until the end of the shoe, so we should decrease bet or stop betting.

We have all seen 'hot shoes' and 'cold shoes' maybe there is truth to it after all?

I can imagine that the variance is going to be collosal with a method like this, but it will be interesting to see what reduction is house edge one can obtain by betting in the later part of the shoe depending on the number of wins in the early part.

dacium said:

I have been doing some programming and simulations on various things to do with blackjack and believe I may have found a new method of winning at blackjack - I don't know if anyone has pointed this out before.

Basically I was looking at the distribution of counts occuring in a 6 deck shoe. It is obvious that the count must be zero and the start and zero at the end of shoe. It is also therefore obvious that the maximums (highest high counts and lowest low counts) mostly appear at the middle of the shoe.

Further more we know that a card count increases or decreases the house edge. It is therefore statistically possible to calculate the most probable count at the middle of a shoe, based on the outcome (win/lose ratio) of previous hands in the shoe. eg. if we have won 20 out of 25 hands, its much more likly there is a good high count that if we got very lucky on a low count.

Most importantly for any shoe with a good count halfway though - it will have on average a good count for every remaining hand in the shoe. We know this is and it is the basis for the unbalanced counts like front count etc.

So if we have a high win/loss ratio at the beginning of a shoe, its is likly that the count is high, and that we will have a favourable count for the remainder of the shoe, so we should increase our bet. Like wise a large number of looses in a shoe is likly to have occured from negetive count that will stay on average negetive until the end of the shoe, so we should decrease bet or stop betting.

We have all seen 'hot shoes' and 'cold shoes' maybe there is truth to it after all?

I can imagine that the variance is going to be collosal with a method like this, but it will be interesting to see what reduction is house edge one can obtain by betting in the later part of the shoe depending on the number of wins in the early part.

Click to expand...

I don't think wins and losses can reliably predict the count in a blackjack game.

I think you said it opposite. If you have won 20/25 hands it is more likely for the count to be NEGATIVE. But even if you lose 20 hands in a row the count can still be negative or positive. A streak of wins or losses will not accurately predict the count of the shoe, in my opinion.

dacium said:

I have been doing some programming and simulations on various things to do with blackjack and believe I may have found a new method of winning at blackjack - I don't know if anyone has pointed this out before.

Basically I was looking at the distribution of counts occuring in a 6 deck shoe. It is obvious that the count must be zero and the start and zero at the end of shoe. It is also therefore obvious that the maximums (highest high counts and lowest low counts) mostly appear at the middle of the shoe.

Further more we know that a card count increases or decreases the house edge. It is therefore statistically possible to calculate the most probable count at the middle of a shoe, based on the outcome (win/lose ratio) of previous hands in the shoe. eg. if we have won 20 out of 25 hands, its much more likly there is a good high count that if we got very lucky on a low count.

Most importantly for any shoe with a good count halfway though - it will have on average a good count for every remaining hand in the shoe. We know this is and it is the basis for the unbalanced counts like front count etc.

So if we have a high win/loss ratio at the beginning of a shoe, its is likly that the count is high, and that we will have a favourable count for the remainder of the shoe, so we should increase our bet. Like wise a large number of looses in a shoe is likly to have occured from negetive count that will stay on average negetive until the end of the shoe, so we should decrease bet or stop betting.

We have all seen 'hot shoes' and 'cold shoes' maybe there is truth to it after all?

I can imagine that the variance is going to be collosal with a method like this, but it will be interesting to see what reduction is house edge one can obtain by betting in the later part of the shoe depending on the number of wins in the early part.

Click to expand...

Dacium in my humble opinion you've put forth a stroke of genius! no kidding.
there have been several authors (and i'm not refering to J. Patrick or looking for full ash trays that sort of thing but genuine AP authors) in the literature who have refered to such effects but i've never seen it layed out as you put it forth. perhaps you are wrong in your surmising as to how it specificaly is but your big picture is i believe right on. there are such effects as you allude to and i believe the potential is there to use the intelligence garnered from those effects to develope or enhance an edge. it think your correct in that such an approach would have a horrific variance that would need to be addressed. you'd need to tweak such a system mightly from your first rough thoughts before it would have any use. for one thing with respect to the ace/ten front count and the idea that multi-deck shoes tend to be sluggish and hold an advantage or a disadvantage once established through the remainder of a shoe as described by Renzy and others, well i wonder about that. it's not been my experience when applying hi/lo to notice such an effect as a predominate nature of shoes. it would be interesting to see or make a study of said affect in a quantitative and scientific way to see if there is some fundamental undelying theorum or principle there.
very interesting thoughts you've put forth there. thank you for sharing.
best regards,
mr fr0g

dacium said:

Basically I was looking at the distribution of counts occuring in a 6 deck shoe...It is also therefore obvious that the maximums (highest high counts and lowest low counts) mostly appear at the middle of the shoe.

Click to expand...

Perhaps the largest running counts occur in the middle of the shoe, but the highest true counts occur at the end. That is why deeper penetration is so important. The deeper they deal, the more high counts we will see. Also, the count is more accurate at the end of the shoe.

dacium said:

Further more we know that a card count increases or decreases the house edge.

Click to expand...

Well, not exactly. The house edge rises and falls on its own. Card counting is just a way to monitor where it is at any time. The count does not affect the house edge, it's the other way around. I think you understand that already but I just wanted to clarify for anyone else reading this post.

dacium said:

It is therefore statistically possible to calculate the most probable count at the middle of a shoe, based on the outcome (win/lose ratio) of previous hands in the shoe.

Click to expand...

Well, yes and no. It is entirely possible to calculate the most probable count at any point in the shoe: It is zero. The count may fluctuate, but I believe the average count anywhere in the shoe will be zero. Also, the win'loss ratio may not correlate to the count/house edge at all.

dacium said:

eg. if we have won 20 out of 25 hands, its much more likly there is a good high count that if we got very lucky on a low count.

Click to expand...

Perhaps, but we don't know how high the count is. The running count may be positive but we are still at a disadvantage. Estimating the count based on the win/loss ratio will not be accurate. Also, as Scott mentioned, you've written it backwards here.

dacium said:

We have all seen 'hot shoes' and 'cold shoes' maybe there is truth to it after all?

Click to expand...

There is definitely truth to it. After all, if there weren't "hot" shoes then the count would never get high. However, the win/loss ratio will not accurately tell you if the count is high or not, and it won't tell you how high it is. Therefore you cannot truly know when to raise your bets or how much to raise them. A similar method, called something like "count profiles", was published by Jerry Patterson in "A Winner's Handbook." Unfortunately it was of no practical use despite his claims that it could be used (with a betting progression of course!) to beat the CSMs. :laugh:

dacium said:

I can imagine that the variance is going to be collosal with a method like this

Click to expand...

The variance will be collosal with any method you try. That is just a factor of the game. Don't let it discourage you.

dacium said:

but it will be interesting to see what reduction is house edge one can obtain by betting in the later part of the shoe depending on the number of wins in the early part.

Click to expand...

There may be a reduction, but it will not be significant. With any situational betting scheme like this there are bound to be errors because of the generality inherent in them. They are oversimplified and make too many assumptions to be accepably accurate.

-Sonny-

Yes I had it the wrong way around.

If you can consider every possible way a blackjack hand plays out, I would expect to see for an average win the true count would be reduced, and for an average loss the true count would be increased (by an extremely small amount), after all it is the high cards that help you win and the lows that help you loose!

We all know that if you loose 5 in a row on roulette it means nothing, but in blackjack it may actually mean a win is more probable to come next, because unlike any other game past outcomes do affect the next hand, because they have withdrawn certain cards.

This has been examined before, and the correlation between win rate and the resulting count is too small to be a good predictor of deck composition. I don't recall any exact details, nor where I remember this being mentioned, but I know that the correlation fell far short of what would be required to overcome the inherent house edge.

KenSmith said:

I don't recall any exact details, nor where I remember this being mentioned, but I know that the correlation fell far short of what would be required to overcome the inherent house edge.

Click to expand...

John Gwynn did a study of this and published it in 1986. The gist of it is:

1) After a win the player advantage decreases by 0.10%
2) After a loss the player advantage increases by 0.12%
3) After a push the player advantage decreases by 0.15%

Leon Dubney then created similar "situational" betting strategies based on these factors and published them in his book "No Need To Count." As Arnold Snyder mentioned in "Blackjack Wisdom" the bet spread required to make this type of system even mildly profitable ($1-$2 per hour) would be much bigger than anything that a card counter would dare use in a casino and the variance would be immense (because the system is so inaccurate, as Ken mentioned).

I think this road has been well traveled over the decades, but I would recommend any of the above books if anyone wants to pursue this path. Just be aware that many people have investigated this concept and all have come up empty-handed. :sad:

-Sonny-

KenSmith said:

This has been examined before, and the correlation between win rate and the resulting count is too small to be a good predictor of deck composition. I don't recall any exact details, nor where I remember this being mentioned, but I know that the correlation fell far short of what would be required to overcome the inherent house edge.

Click to expand...

check this link perhaps its where you saw it mentioned:
http://ourworld.compuserve.com/homepages/greenbaize21/prog.htm (Archive copy)

what interests me about this issue is not trying to use it as a sole perfected system but to use such awareness in conjunction with the standard AP systems especially when caught in a marginal position or when camo is called for.

best regards,
mr fr0g

I just stumbled accross this full analysis by MathProf:

http://www.bjmath.com/bjmath/progress/progloss.htm (Archive copy)

His conclusion, which confirms Snyder's, was that you would need a 1-46 spread just to break even.

-Sonny-

Some of the ways some people have simulated the method are borderline pathetic.

The one you just posted is terrible. He waits for a loosing streak, then bets down the rest of the deck, which proves nothing. I think he has mistaken having a 1% edge on the next hand, for have a 1% egde for the remaing deck. Its like getting a count of +1 then stopping the count and just playing down the whole shoe, rediculous.

dacium said:

He waits for a loosing streak, then bets down the rest of the deck, which proves nothing. I think he has mistaken having a 1% edge on the next hand, for have a 1% egde for the remaing deck. Its like getting a count of +1 then stopping the count and just playing down the whole shoe, rediculous.

Click to expand...

But that's exactly what someone using that system would do. Since a win only decreases the advantage by 0.10% it would take 10 hands to wipe out a 1% advantage. Someone using this system would still be making big bets during those next 10 hands. Since you will never see a string of 5 losing hands followed by 10 more hands in a SD game (or even a DD game for that matter!) this system is almost useless. This system is not effective at all in multi-deck games.

Another reason this system is so weak is that it makes very general assumptions about the deck that my not correlate to the actual advantage at all. You will be raising your bet into a disadvantage almost as often. It is simply not accurate enough to establish a winning betting strategy.

-Sonny-

Sonny said:

John Gwynn did a study of this and published it in 1986. The gist of it is:

1) After a win the player advantage decreases by 0.10%
2) After a loss the player advantage increases by 0.12%
3) After a push the player advantage decreases by 0.15%

Leon Dubney then created similar "situational" betting strategies based on these factors and published them in his book "No Need To Count." As Arnold Snyder mentioned in "Blackjack Wisdom" the bet spread required to make this type of system even mildly profitable ($1-$2 per hour) would be much bigger than anything that a card counter would dare use in a casino and the variance would be immense (because the system is so inaccurate, as Ken mentioned).

I think this road has been well traveled over the decades, but I would recommend any of the above books if anyone wants to pursue this path. Just be aware that many people have investigated this concept and all have come up empty-handed. :sad:

-Sonny-

Click to expand...

one way of putting to use such limited feedback might be a form of wonging in. lets say you stand across the room from a table. you can't see the cards values so it's reasonable to the pit that you don't need heat. but all the while you are tallying up the win to loss's. if you see that the players have really been getting beat up on it would be likely that a lot of low cards have been eaten up. come in at the half way point or before of such a shoe and start the count at zero. if there is infact an advantage you would be doing fairly well and if the count does go up you would bet the optimal. this way your escaping making a lot of waiting bets at the begining of say a six or eight deck shoe. you are more likely to be comming into a shoe richer in high cards than if you were starting a fresh shoe.

best regards,
mr fr0g