Dear friend, what I've written is the pure truth.

I'm deadly sure I'm playing with an advantage over the house, yet many times your suggestions helped me by transforming a 'normal' negative short term variance into positive or neutral situations.

as.

In a famous experiment (Confessions of a Coin Flipper and Would-be Instructor) C. Konold found out (contrary to the 'expert' common belief, including him) that after a fair number of coin flip trials the HTHHT sequence will invariably come out first vs the HHHHH succession, despite both patterns have the exact (theorical) same probability to appear.
His purpose was to emphasize one of the many Gambler's Fallacies affecting our brain, that is our 'overalternating' way of consider binary random outcomes.

Thus if you find someone willing to face your 1:1 bets where you'd bet that HTHHT sequence will come out FIRST than the HHHHH sequence, you'll get the sure EV+ side.

Obviously itlr every 5-hand coin flip pattern will come out with the same values (but by different frequency), otherwise gambling games wouldn't exist.
Moreover we won't find a game where all intermediate "no win/no lose" patterns will get us a neutral situation before getting a W (HTHHT coming out first) or a L (HHHHH showing up first).

Anyway, it's natural to think that if the HTHHT sequence will come out FIRST than the HHHHH sequence, many other heterogeneous patterns will follow the same propensity.

More importantly is the fact that along the way the HHHHH sequence getting the same 1/32 other pattern's probability to appear must 'catch up' in some way this 'deficit'. And of course it can do that just by coming out 'clustered', that is by a W/L ratio greater than 1/32.
This calls into question, albeit by an 'opposite' way of thinking,  the 'probability in decline' feature already discussed here.

it's just a matter of 'time', huh?


as.

Coin flip curiosities

Coin flip successions are unbeatable? Sure.

Does any coin flip pattern of the same lenght and different HT distribution have the same probability to appear by a 'time' factor? Nope.
That is some patterns are more likely to come FIRST than others, then obviously itlr all patterns of the same lenght will equal.

See you later.

as.

Glen, it's a very good post filled of ideas I've implemented in our play.
I don't recall the times when while waiting to place a dubious bet, your suggestions came to my help and believe me or not, I was more right than wrong (even by not betting at all, of course).
I even named them as 'Al spots'. Maybe I'll make some comments on that at my section.

But my basic approach remains strictly 'probability' based as I have to rely upon something that was demonstrated to be reliable beyond any doubt even though the short term variance could be a harsh obstacle to overtake.

So I continue to think that setting up a plan about your ideas is a very complicated task needing a lot of experience, I mean years and years of real play with an 'educated' mind. 

Even Giz replies are interesting, particularly when he wrote:

It's one thing to see a trend or a pattern. It's another to know if it is in a coincidental phase of working temporaraly or not. Knowing is everything.

BTW, I do not think randomness is beatable but the fact that he keeps making references about roulette ECs 'as a whole' where numbers are not perfect equally distributed (besides R/B numbers) on the wheel could be food for thought.

Anyway good thread.

as.

Why coin flip successions (EV=0) are unbeatable but baccarat EV- successions could be beatable?

There are several answers to this question, the main one relies upon the asymmetrical probability acting at every hand dealt as besides N9/N9, N8/N8, S7/S7 and S6/S6 points, every other hand will be hugely favorite to eventually win from the start. And this is due to the asymmetrical card distribution working per every shoe offered.

That's a quite of difference with a coin flip succession where the winning probability is always and endlessly symmetrical.

One could argue that we never know which side will be hugely favorite to win after two initial cards are dealt, but this is a completely wrong assumption as 'complex' patterns will help us to define the issue.

In poor words, most bac spots provide a kind of 0 sum (unbeatable spots), others are slightly shifted toward a precise outcome where math doesn't get any sensible role (for obvious reasons as itlr EV is mathematically negative).

Simplyfing, baccarat is just a matter of EV+ card distribution spots variance and where attempts to 'control' the vast part of EV- situations are worthless.
The clustering effect taken in its various forms is just one factor to take care of; there are other issues, some of them brilliantly discussed here by Alrelax and KFB in their posts.

I've chosen the simple P singles / P streaks problem just because is a kind of 'math' less disadvantaged' proposition where a sort of 'sky's the limit' (but unnecessary) plan will get the best of it.

Remember that casinos want you to guess the fkng unguessable whereas professionals like to 'guess' what is slight more entitled to win, at the cost of waiting and waiting and waiting.

So let's act as black jack card counters do with the paramount difference that at baccarat there is no need to bet a dime at the vast majority of EV- spots coming around (at black jack 88-83% of all hands dealt are EV-). At the same time giving a role to the actual card distribution.

Next samples are taken randomly from live shoes collected at various LV premises:

1- S,S,S,R,R,R,R,R,S,R,S,R,S,S,S,R,R,R,R,S,S,S,S,R,R

(+ + + - - - -) (- - + - + + +) (- - - - + + +) - -

W, W, L(-1)

2- S,R,R,R,R,R,S,R,R,S,S,R,S,R,S,S,S,S,R

(+ - - - - + -) (+ - + + + + -)

W, W.

3- R,S,S,S,R,R,R,R,R,R,R,S,S,R,R,S,S,R.

(- + + - - + -) - + -

W, -1 at the final uncompleted 7-event sequence

4- R,S,S,R,R,R,R,S,S,R,R,S,S,S,S,R,R,S,S.

(- + + - - + -) (- + + + - - +)

W, W.

5- S,S,R,S,S,R,R,R,R,S,R,R,S,S,R,R,R,R

(+ + - + + - -) - - + - -

W and a -3 uncompleted L cycle.

6- R,S,S,R,S,R,R,R,R,R,S,S,R,R,R,S,R,S

(- + + - + - -) + - - - +

W and a -1 uncompleted cycle

7- S,R,S,S,R,R,S,R,S,S,S,S,R,R,S,R,S

(+ - + + - - -) (+ + + - - - +)

W, W.

8- S,S,R,S,R,S,S,R,R,R,S,R,R,R,R,S,S,R,S,R,R,S

(+ + - + - + +) (- - - - + - +) - -

W, -3 L.

9- S,S,S,S,S,S,R,R,S,R,R,S,S,S,R,R,R,R

(+ + + + + + -) (- - - + + - -)

W, -3 L

10- R,S,S,R,R,S,S,R,S,S,S,S,S,R,S,S,S,R,S,R

(- + + - - + -) (+ + + + + - +) + - + -

W, W.

11- R,R,R,S,R,S,S,R,S,R,R,S,S,R,R,R,S

(- - + + - + -) - + - -

L -1 and another uncompleted L -2

12- S,S,S,S,S,R,S,R,S,S,R,R,S,R,R,R,S,S

(+ + + + + - +) (- + + - - - -) +

W, W.

13- R,R,S,R,R,R,S,S,R,R,R,R,S,S,R.

(- - - - + - -) + -

L -5 and an uncompleted 0 situation.

14- S,R,S,S,S,R,S,R,R,S,S,R,R,R,S,S,R,S,S

(+ - + + + - +) (- - + - - + -) + +

W, L -3. (To get things at the worst scenario, we'll ignore the uncompleted ++ sequence)

15- S,S,S,R,R,R,R,S,R,R,R,R,S,R,R.

(+ + + - - - -) - -

A W and an uncompleted -2 L

16- S,R,S,S,S,R,S,S,R,S,R,R,R,S,S,R,S,R,S

(+ - + + + - +) (+ - + - - + -) + - +

W, W

17- R,S,R,R,S,S,R,S,R,S,S,S,R,R,R,S,S,S,S,S,S,R

(- + - - + - +) (- + + + - - +) + + + + -

L -1, W (and an uncompleted W sequence)

18- R,R,R,R,S,S,R,R,R,S,R,S,R,S,S,S,R,R,R,S

(- - + - - - +) - + + - -

L -3 and an uncompleted W sequence

19- S,R,R,R,R,S,S,S,S,S,R,R,R,S,R,S,S,R

(+ - - + + + +) - - - + + -

A W and an uncompleted -2 L sequence

20- S,R,R,S,S,S,S,R,R,R,R,S,S,S,S,S,S,S,S,S,R

(+ - - + + + -) (- + + + + + +) + + -

W AND W (and an uncompleted W sequence).

Despite that such randomly taken 20-shoe sample overall formed 696 B hands and 746 P hands (ties ignored), the P single vs P streak plan got no problems to endure the possible negative variance.
No matter how were the permutations. 

Again, if the less likely world keep happening, it will clustered shaped and after having surpassed the 1 cutoff value, we're not interested to chase a more likely propensity unless it comes out once.
Since we're stopping the betting at less likely situations coming out clustered but prolonging the more probable counterpart until it'll lose (or getting a kind of +1 ratio), we know to play with a kind of long term advantage.

It's quite easy to falsify this hypothesis: setting up the same plan by wagering the opposite: P streaks  vs P singles.

as. 

Quotere: clusters

--I also find that after catching a couple wins from said cluster it is beneficial to not keep looking for the same exact cluster or "grouping" (in that shoe). The initial concept is still valid in the residual of that shoe. However, the residual of that shoe now possess one less of the Previously Caught Cluster.

This is an excellent consideration ignored by the vast majority of bac players, one of the main reasons why they keep losing.

For example, it's very important that the 7-event cycle would start with a W or, at worst by a L followed by a W, yet those situations will be more likely whether the previous WL register applied to any of the 128 possible permutations have shown a kind of unbalancement toward Lx or LL.
The same about every other 'intermediate' WL 7-cycle succession that obviously can't show for long the same features.

We ought to remember that itlr P singles > P streaks, yet volatility will easily make many shoes to produce many P streaks at any point of our 7-event cycle.
So now the problem shouldn't be solved by chasing the mere number of P clustered singles but also by studying their 'shape' that more often than not continuously changes in its 'density'.

Moreover even a 'clustered' clustering effect is always in order, of course needing some shoes to be dealt: Now a FB approach will make the best of it without using any progression.

See u later

as.

Thanks KFB and Adulay!!
Just finished a very long session , about 14 straight hours.

See you later

as.

Out of curiosity next shoes are:

S,R,S,R,R,S,R,S,S,R,S,S,R,S,R,S,S,R,S

(+ - + - - - +) (+ - + + - + -) + + - +

Win and Win.

S,S,S,S,R,S,S,R,R,R,S,R,R,R,R,S

(+ + + + - + +) - - - -

Win, maybe the - - - - streak could 'easily' come out at the very start of this shoe.
Nevertheless the W/L ratio is still 6/5.

R,S,R,R,S,R,S,R,R,S,S,S,R,S,R,R,R

(- + - - - + -) (- + + - + - -)

First sequence is a -3 unit loss, next sequence provides a W at third hand.
Let's assume the second streak wasn't in action, so our next bet will be 4.

S,S,R,S,R,S,R,R,R,R,S,R,S,R

(+ + - + - + -) - - + -

A W.
But since W=5 and L=6 and considering another 4 unit loss (so we're behind of 7 units) let's see what happens at a new 8 unit bet:

S,S,R,S,S,S,R,S,R,R,S,S,R,R,R,S,S,R,R

(+ + - + + + -) (+ - - + - - +) - -

A steady W at the first sequence, another W at the second one even if the WL ratio is 3/4.

as.

More shoes taken randomly from our live bac samples:

S,R,R,S,R,R,R,S,S,R,S,R,S,S,R,S,S,R,S

(+ - - - - + +) (- + - + + - +) + - +

First 7-hand sequence: ok, easy 'lucky' W, yet we couldn't be any worse than losing just 1 unit.
Second sequence got a W at 5th hand)

S,R,R,S,S,R,S,S,R,S,R,S,S,R,S,R,S,R

(+ - - + + - +) (+ - + - + + -) + - + -

No 'lucky' spots here: both 7-hand sequences got a 4/3 WL ratio. So two Ws and at the very first betting spot.

R,S,R,R,R,R,S,S,S,S,R,R,S,S,R,S,S,R

(- + - - + + +) - + - + + -

A W at the 7th hand

S,R,S,R,R,S,R,S,R,S,R,R,R,R,S,S,S,S,S,S,R

(+ - + - - - +) (- + - - + + +) + + -

Again a fortunate win in a -1 overall permutations scenario. Second sequence got a W at the very end of it)

S,S,S,S,S,R,S,S,S,S,S,S,S,S,S,R,S,S,R,S,R,R

(+ + + + + - +) (+ + + + + + +) (+ - + + - + -) -

Easy wins

R,S,S,R,R,S,R,R,R,R,R,R,R,S,R,R,S,S

(- + + - - - -) - - +

Again we were 'lucky' as the -++ sequence got us winners but let's consider the more probable scenario where we have a -3 unit loss, so we'll raise our next bet to 4.

R,S,S,R,R,S,S,S,R,S,S,S,S,R,R,R,R

(- + + - - + +) (- + + + + - -)

First sequence is a winning one no matter the permutations. The same about the second one.

R,S,R,R,R,S,R,S,S,R,R,R,S,S,R

(- + - - - + +) - - + -

A -1 unit loss, so let's raise our new bet to 2 units.

S,R,R,S,R,S,R,S,S,S,S,S,R,R,R,S,S,S,R,S

(+ - - - + - +) (+ + + + - - +) + - +

A prompt W despite of the unfavourable WL ratio, second sequence is a piece of cake.

R,S,R,S,S,R,S,R,S,R,R,R,S,R,R,R,R,R,R,R,R

A pretty bad sequence as S=6 and R=15.

(- + - + + - +) - + - - - -

We win at hand #5.
Notice that such bad sequence produced 5 Ws and 8 Ls. So a gap of 9 (S6 and R15) actually got us just a -3 unit loss.

See you next week

as.

Say we set up a progressive plan by wagering towards P singles vs P streaks by adopting the 7-step  progression method already discussed.

Once a new P hand comes out, we'll wager Banker by FB just one time (hoping a P single will show up) and we'll do that for 7 consecutive hands.
At the end of such 7 betting hands, results can only be those (vig ignored for simplicity):

+7 (all bets were winning P singles)
+5
+3
+1
-1
-3
-5
-7 (all bets were losing P streaks)

Without any shadow of doubt, itlr +7>-7, +5>-5, +3>-3 and +1>-1.
Of course vig will erase and invert such favourable propensity in terms of units won, yet those ratios stand yesterday, now, tomorrow and forever.

If after 7 bets made we'll be ahead by FB, we collect the winnings and restart a new 7-hand cycle.
If we are in the losing territory, we'll set up our new betting unit by adding one unit to the previous deficit and so on.

Example.

First 7-hand cycle: only 2 P singles and 5 P streaks = -3

For the next 7-hand succession we'll place a 4 unit bet stopping the play until we erased the previous deficit but letting the 7-hands going into completion without betting.
We know that the math probability to win is always 93/35, so strongly shifted at our favor.
Anyway say that despite the math, also the second 7-betting sequence will get us a -3 final deficit, so now we're behind 3 units (first 7-hand sequence) plus 12 units (the actual sequence).
So our new standard unit will be 16 units.
And so on.

In the example posted, we got just 4 P singles and 10 P streaks but arranged in negative (for us) permutations.

The favourable three-fold propensity cannot be disregarded for long, the problem is that we have to employ a large progressive bankroll to 'cover' the negative natural variance.

So we may use the 'clustering/isolated' effect to try to reduce variance (and disboursement), hence we'll bet toward a P single: a) at the start of the shoe, b) after a previous P single came out and c)after a previous (single) P streak came out.
Now long P singles are always good, P streaks followed by a P single are good and of course clustered P streaks will get us a way lesser damage than stubbornly wagering against them.

Let's take shoes randomly from our datasets.
S= P single and R= P streaks (runs)

1- S,R,S,S,S,S,S,R,S,S,R,R,R,S,R,R,S,S,S,R

(+ - + + + + +) (- + + - - - -) + + -

First 7-hand sequence got a W, second 7-hand sequence got a W (-++).


2- S,S,R,S,R,S,R,S,R,R,S,S,R,R,R,S,R,S,R

(+ + - + - + -) (+ - - + - - -) + -

Two W.

3- S,S,R,S,R,S,R,S,S,R,S,R,S,S,S,S

(+ + - + - + -) (+ + + - + + +) +

Two W

4- R,R,S,S,S,R,R,R,R,R,S,R,S,S,S,S,R,S

(- - + + - - -) (+ + + - +)....

First sequence got -3 L and the second one a W.

5- S,R,S,R,R,S,S,R,R,S,S,S,S,S,R,R,S,R,R

(+ - + - - + -) (- + + + + - -) -

W and W.

6- R,R,S,S,S,S,R,R,R,R,S,R,S,R,R,S,S,S

(- - + + + - -) - + - - + +

Despite of having 4 losses and 3 wins we'll collect a win at the first 7-hand sequence.

7- S,R,R,R,R,R,R,S,S,S,S,S,R,R,R,S,S,S,S,S

(+ - - + + + +) - + + + +

A W.

8- R,R,R,S,R,R,R,S,S,S,R,S,S,S,R

(- - - - + + -) + + + -

Finally a -3 L.
Now we have to raise our standard unit (so our new bet will be 4)

9- R,S,S,S,R,R,R,R,S,S,R,S,R,R,R

(- + + + - - +) - + - -

A W, so we go back to our 1 unit.

A note: arrange the + and - signs in the most bad sequence and in the worst scenario you won't be behind more than 1 bet.

10- R,R,S,R,R,R,R,R,S,S,S,R,R,R,S,R,S,S,R

(- - + - - + +) - - - + + -

It's a -1 unit L, so new bet will be 2 units.

11- R,R,S,R,S,S,S,S,S,R,R,R,R,S,S,R

(- - - + + + +) + - - + -

A W. So going back to the initial unit.

12- S,R,S,S,S,R,S,R,S,R,S,R,S,R,S,R,R,R,R

(+ - + + + - +) (- + - + - + -) + - - 

A W at the first 7-hand sequence and a -1 L at the second one.
So we'll raise our bet at 2 units for the next series.

13- S,R,R,R,S,R,S,S,R,S,S,S,S,S,R,S,R,R,S,S

(+ - - - + + -) (+ + + + + - +) - - +

A W at the first 7-hand sequence (even by accounting the worst permutation the final loss would be just -1 unit); second sequence is just a piece of cake and recovering easily the first sequence loss.

14- R,R,S,R,R,S,S,S,S,R,S,R,S,R,S,S,S,R,S,S.

(- - - - + + +) (- + - + - + +) + - + +

First sequence is a -1 L; second sequence bet by a 2 unit was a winner just at the final 7th hand.

15- S,R,S,R,R,R,S,S,S,S,R,S,R,S,R,S,S,S,R,S,S

(+ - + - - + +) (+ - + - + - +) + + - + +

W and W. At both sequences there was no way to lose a dime as + > -, so no 'bad' permutations couldn't come out along.

Comments

This 'basic' plan relies upon several distinct math and statistical features:

- RVM definition of randomness;

- Marian V. Smoluchoswki 'probability after effects' studies;

- The baccarat very slight general propensity to get the opposite result already happened;

- Marigny De Grilleau works;

- The math solid assumption that B>P.

- Other issues considered worthless (fortunately) I don't want to discuss here for obvious reasons.

as.

8- He survived cancer

Doyle was and remains the best Ambassador of Poker, a true legend!

I agree with 8OR9: a gentleman who deserves a great movie!!!

RIP Doyle!

as.   

Welcome back Giz!!

as.

1- Itlr any event or series of events getting a greater than 50% probability to appear will more likely arrange in clusters.
Obviously such propensity is related with the probability value, so betting B after a B is a too tiny value to set up a plan upon, so affected by a huge level of variance.

2- On the other end, the less likely event or series of events show the obvious general propensity to come out isolated, yet they are way more affected by volatility than the above 'clusters' propensity as in some way they must catch up the 'normal' clustering effect.
After having studied large live shoes datasets we have reached the conclusion that it's better to 'chase' clusters than hoping to get 'less likely' isolated situations.

3- A considerable part of L isolated spots come out clustered after single (so not clustered) W events, so the succession looks as WLWLWL...

4- At baccarat the least possible effort to raise the winning probability from 0.5 to higher levels is to bet two hands in a row by 'chasing' a given outcome (0.75).

5- There are reasons to think that betting towards singles and doubles in terms of overall clustered or isolated successions is the simplest way to set up a plan.

6- It's not a coincidence that in my above posts I've focused the attention about the very two starting events that seemingly appear to get the highest impact of 'randomness'.

7- Progressions cannot erase or invert the HE, but if a slight propensity goes toward one side of action, it's just a matter of time to get the profits we aimed for.

Progressive plan.

Since we have to wager two times in order to get a profit but a progressive plan is in action (so lowering and diluting the disbursement will take a primary role), we set up the simplest scenario: wagering 1 and, in case of loss, 1. Both bets account for just one overall bet, so when we win at singles we profit 1, when we win at doubles we'll break even and when we lose both steps (a 3 streak appearance) we'll lose 2. (Vig is ignored here for simplicity).

We'll split our betting action within 7 distinct frames and per each frame we'll flat betting until we'll get a profit.
If a loss come out after betting all 7 two-step events, we'll raise the next standard bet by adding one unit to the temporary deficit and we'll bet this amount for 7 hands long until we'll get a profit. And so on. 
If a profit will come out BEFORE the 7-event frame ends up, we have to wait the 7-event termination, then restarting the betting with the standard unit.

Per every 7-event (two-step) frame we'll bet, out of 128 possible patterns (permutations) we have 93 positive winning patterns and only 35 losing patterns (0.7265% W probability) and this ratio will happen for every 'new' 7-event frame considered.

Example.

Following the W clusters and L isolated spots by betting toward singles/doubles at the very two initial shoe events (and considering them as a endless sequence), we'll lose 2 units whenever we'll encounter a L (3 streak), then winning 1 unit if the W is a single and breaking even if the W is a double.

The worst scenario per every 7-event (two-step) result is to lose ALL spots in the form of W isolated situations and L clustered situations, so the sequence will look as WLLWLLWLLWL where L could be longer than LL but we can't care less.
In such very unlikely instance our deficit will be -14 units so for the next 7-event frame we'll have to bet 15 units.

Of course losing 14 hands in a row is possible but it's a very very unlikely scenario, after all the problem relies upon the 'positional' W distribution in terms of clusters and L distribution in terms of isolated patterns, already more likely by intrinsic factors.

It's true that when things are going wrong, we need to employ a large bankroll, but eventually and unless very very unlikely permutations come out so negating our advantages, we'll know to come out ahead.

Summary

Things might change when we try to be ahead even after a first loss was made to chase a single. So also doubles constitute a win.

Anyway here's a kind of guideline:

Flat betting 7 times (by one or multiple slight progressive steps) the distribution of more likely probability events: at the end if you'll be ahead stay at the same unit amount for another 7 times. And so on.
Whenever you'll collect a loss (it can be just -1, -3, -5 or -7 units) for the next 7-event frame raise the betting amount by adding 1 unit (or more than that considering vig) to the actual deficit and so on.
Do not forget to let it go until the 7 events completion before betting again when ahead or to increase the bet while losing until seven hands are completed.

In terms of probability, this procedure should be solved by Markov transition matrices, to simplify the issue let's say that the expected probability to fail is 1:654.

Actually and at least in the terms prospected here (where itlr the bet selection makes a slight but significant role) the Black Swan (I like the Alrelax definition of a very bad situation showing up) will come out with odds close to 1:810 so getting the player a huge advantage over the house.

See u next week.

as.

Such method relies upon three different math and statistical propensities having the principal effort to put in the least corner the negative variance.

as.

In a couple of days I'll show you how a math progressive approach will get the best of the game, for once it's a only Banker betting method needing very very unlikely sequences to fail.
Actually it's just a matter of money...

as.