Quote from: albertojonas on March 22, 2026, 10:43:51 PMMay you direct me into these posts please?

Suppose your trigger is a FRESH 3/3+ streak.
You want to classify the next pattern being either another 3/3+ streak (cluster) or anything else (that is a single or a double so making that 3/3+ streak as isolated).
Of course after a 3/3+ streak came out as clustered you don't want to go further so waiting until a new 3 streak shows up.
You register how many times a (1/2) or another 3 had shown up knowing that the expected ratio is 3:1.
Even though occasional deviations could be harsh in a way or another, there's a constant force acting toward a balanced ratio after a given deviation happened, especially after some clustered 3s and especially when the 3/3+ streak is at B side (making more probable short predominant patterns at B, so leaving few room to 3/3+ P streaks suddendly balancing the previous B same streak).

The same classification could be made for clusters of 2 vs superior clusters, clusters of 3 vs superior clusters and so on.

Normally here it's the back-to-back shoes registration that counts as single shoes may be affected by huge volatility and  no balancement.
 
as.

Quote from: alrelax on March 23, 2026, 02:04:57 AMYou wrote: "Probably those who constantly win have learned to think that the HE is the least problem to face, yet we don't know a single successful player confiding that his/her main profit comes out from wagering the side bets."

And you are absolutely correct.  But side bets do appear frequently, not rarely in the 5 Treasures game.  I have done and do well with wagering 3-4 sides at a time.  But the main source of winnings will always be the base B/P wager.

That's the answer I wanted to hear from you after reading your excellent post (#1501). 

oOoOo

Successful players consider any new session as a fresh one, forgetting what happened in the past, for the simple reason that we can't know precisely how the cards are shuffled. Actually we could even consider the scenario that the same or almost the same shoe will distributed again, an heaven if we have won and we keep taking the same strategy but a bad nightmare whether we have seriously lost and continuing to apply a preordered/adapting plan not fitting the previous shoe(s).

Winning is not "hoping for", winning is a complex process dictating what's more likely to happen at the spot we decide to wager (or to fictionally guess).
Fictional right guesses are not wasted opportunities and at the other end fictional wrong guesses are good spots to save our money as we'll expect more negative ROI situations than positive ROI situations.
Moreover the vast part of W/L results are distributed by a clustered fashion so negating a long overalternating scheme.

There are infinite situations to look for, especially if we take into account several sub random walks applied to the original BP sequence.

Suppose we are taking care of the 1/2 vs 3s ranges.
Obviously the 1/2 ranges cannot be zero at any of derived roads (meaning we won't get all 3s streaks along all the shoe at one line, let alone at different random walks) and it's 100% certain that very soon a 1/2 range will come out clustered at one or, more probable, more than one derived road.
On the other end it's particularly likely that at a given derived road, a long 1/2 pattern will stand for long.
If the general probability ratio of 1/2 vs 3 is 3:1 and knowing that the production is asymmetrically shaped, we may infer that most of the times a 3 value will shift to 4 (or more), especially whether one or two 1/2 ranges have formed an exact 3 gap.

3s streaks are moving by the same propensity, so more likely coming out as isolated than clustered, especially by the "clustered" isolated/isolated fashion.
 
Even though some colliding events come out quite often (meaning that opposite BP bets are making favourable opportunities at one side or the another one(s) ), we know that the vast majority of shoes will make more probable fair long 1/2 ranges so making our guess a kind of no brainer shot.

as.

Yep.
 
It's 1 billion % certain that itlr any progression in the world can't overcome a negative edge, otherwise the claimer would be hired by MIT or by NASA and getting payed millions of bucks without betting (so risking) any dime at bac tables.

The only way to verify a possible advantage at a negative edge proposition is to show a strategy capable to provide an ascending profits line by flat betting and the only way to do that is by theorizing a possible unrandomness of the production and/or to demonstrate that card distributions are affected by "limits" shifting at some points the results by values overcoming the HE.

If a given flat betting strategy sucks, the same strategy applied by any kind of progression sucks even more.

as.

Definitely baccarat is a game of skills, the few who have found out why this game could constantly be beaten know very well this.

Probably those who constantly win have learned to think that the HE is the least problem to face, yet we don't know a single successful player confiding that his/her main profit comes out from wagering the side bets.

Successful players know that a strict mechanical betting applied to every production have zero chance to overcome the HE.

Successful players know that patterns, being positive or negative, could last for "very long" but negative patterns are asymmetrically more harmful than positive ones (for the HE).

Successful players know that the occasional ultra positive patterns will be soon replaced by undetectable patterns that again could last for long.

Successful players know that "less is more", meaning that the probability to be right is inversely proportional to the number of bets placed on the felt.

More later

as. 

Quote from: albertojonas on March 18, 2026, 11:14:39 PMCan you direct me to any information on these investigations, beside the books written on Marigny adaptations?

Unfortunately not, such tests were run several years ago at real roulette spins , I've tried to look for some links but without finding them.
Basically the trigger was set up at 3,5 sigma and even at 4 sigma then looking for a minimal correction under the MdG guidelines but no significant differences emerged than by using a random selection.

Anyway my posts abound of Marigny concepts as "isolated", "clustered" and the best of his philosophy that is the flat betting scheme suggested in his main book.
After all baccarat can't be compared with the roulette symmetrical productions, maybe here some of his ideas whether properly adapted might work.

as. 

You described very well the "statistical limit" concept but as long as the production is independent and therefore symmetrically shaped, no strategy will be able to shift a 50/50 probability in our favor no matter how deep a deviation will happen before starting to bet.

This topic was studied nearly one century ago by Marigny de Grilleau whose strategy was to wait for a 3 sigma before betting with the aim of winning just one (large) unit by flat betting.
Several years later, this MdG strategy was deeply investigated by testing real roulette outcomes and simulated pc spins but with no avail at both cases.

Correction surely will act but always in proportional terms (percentages) related to the number of hands dealt so needing a lot of time to show up, so making worthless any bet selection.

Anyway the main problem is not related to the A/B deviations but by the 0 impact that cannot be overcome by any bet selection or progressive schemes no matter how sophisticated are conceived.

At baccarat things are way easier as any deviation won't stand for long (strong correction), at least by considering the same shape of apparition of any pattern at back to back finite and dependent shoes.  If not, all bac player aiming for constant deviations of some kind will clean up every casino in the world and we know that's not the case.

To support our hypothesis let's take the back to back bac doubles distribution.  We want to bet that any double won't be followed by another double just one time (so letting go the multiple double clusters greater than two).

The double/double vs double/anything else ratio is 1:3, we'll wait for a situation where this ratio is moderately/heavily shifted towards the left (the strongest, the better), then starting to bet in order to get an isolated double apparition or isolated double sequence (isolated double correction).

That's one of the most "balanced" propositions we've found out in our long term trials.

as.

Maybe you are opening a new world about this game, no jokes.

Yes, getting a clear and focused mind is very important in every field, but successful gamblers made their fortune by exploting an advantage most people do not have.

Now the question specifically related to baccarat is:

Could a player find the situations where B bets will get at least a 51.3% probability to win and/or where P bets will get at least a 50.1% probability to win?
That's the basic undisputable form of advantage that must be measured after collecting several session results.

If no valuable triggers exist (I disagree on that, but that's not the point here) that means that some players show the ability to guess more right than wrong by a value capable to erase and invert the HE.

I'm aware some studies made at mere coin flip propositions guessing have shown that some people got a winning percentage above than 50% and those studies were driven by serious professors evaluating significant statistical data.
We know that when the HE=0 things might naturally take incredible long positive (or negative) lines, so what when we have to guess at a -1.06%/-1.24% negative edge proposition?
Now is a clear focused and prepared mind capable to spot the events where the HE will go down the drain?

I don't have the answer but Alrelax words seem to give us a positive reply, we ought not to forget that he played for real an astounding amount of shoes.
For that matter a couple of high end casinos floormen gave me a similar response when asked if they'd think some players have "more feeling" for the game than the rest.

So could a clear, focused, prepared and experienced mind be able to approximate at best when some spots are better than others?

as.

Good thread.

Al wrote: The bottom line is, while you cannot totally control what is happening, or about to happen in front of you, you must control yourself by the way you respond to what is happening or just happened.

That's a very good rule of thumb and our general answer, when in doubt, is to bet very few hands.
Obviously there are many specific guidelines to follow as well that more or less are framed in the "hoping for the best but expecting the worst" picture.
This seems to collide with the #2 point (The habit of your inner resistance) but it doesn't if one has verified a long term edge that must be adapted to the actual outcomes.
Laboratory tests are made by a way greater speed than live shoe results and this leads us to do a lot of mistakes at real tables, especially when we're losing.

Then:
1). The Habit of Expecting Things to be a Certain Way

As you sayed, this point is very subjective.
We'd guess that at least 95% of bac players hope for limitless positive deviations, 4.99% "wrongly" confide about 1-step long term math/statistical data (B>P, B/P average ratio, etc) but only the remaining 0.001% (an optimistic percentage, I know) are really able to understand and more importantly exploit the intricacies of the game.
"Expected things" need a moderate/huge hands volume to be exploited, providing one had carefully tested and measured why his/her edge comes from.

2-The Habit of Your Inner Resistance

Not a surprise knowing that 100% of keen bac players have lost their a$$es at the tables.
The game is conceived to make the players to lose no matter how smart or st.u.p.id or aggressive or cautious they are.
If a coin flip proposition (No HE) cannot guarantee a player to be ahead after X hands, let's imagine what happens after several trials of betting when the ROI is constantly 0.9894:1 or 0.9876:1.
In a word, this factor is important to be "ignored" after having assessed to play with a verified long term edge, even though we all know that most "no edge" players are particularly prone to raise their bets while losing and being particulary prudent while winning (an asymmetrical detrimental attitude).

3). The Habit of Focusing Only on What Is/Was Wrong

This is a very important factor to be constantly aware of.
When things are not going to our favor, we should evaluate what are the probabilities that future patterns will fit (or not) our plan and now our brain must be particularly focused about NOT losing more hands than promptly recovering the actual losing status.
It's here that most money is lost by the intervention of what we call as "compounding error" even if the first hands were lost by natural variance.

as.

By applying a 0.75% general probability to win at a sure asymmetrically card distribution and slight asym results game (B math propensity) we are taking into account a "biased" W/L 3:1 ratio.
We know that the B propensity won't get us any advantage whatever taken whereas the asym card distribution will.

Then quite frequently a part of results will come out "coincidentally" and confounding the picture, nevertheless following a long term more likely distribution that becomes "certainty" with the increasing number of shoes dealt.

So the "expected" general 3:1 ratio becomes more than a virtual value whether considered within few shoes where we expect more deviations than "balanced" events, of course the problem is always to estimate (approximating at best) which deviated line will be predominant over the other one (that is disappointing the 3:1 ratio).

Possible answers addressed to solve this problem.

1) The unlikelihood such 3:1 ratio will stand for long

2) The minimum requisites to get an event coming out by a more likely shape

3) The actual deviations happening at the shoe we're playing at (exploiting deviations)

4) The RTM effect working at multiple shoe distributions.


1) Any exact 3 W streak vs 3+ W streaks will be so balanced in its apparition that waiting for a fictional 2:0 or 3:0 ratio will get us a future edge, of course by betting that a 3 streak will become a 4 streak or longer streak.

2) If the W/L ratio is 3:1, we just need a W event to come out clustered once and again waiting for a 0:2 or 0:3 W event NOT showing up clustered is a good way to look for a possible advantage.

3) Nowadays cards are so whimsically (and possibly unrandomly) distributed that strong deviations come out around any corner so giving a fk about expected probabilities.
I don't recall how many times we have collected additional important profits by following the Alrelax statement: "when it's there it's there".
Do not put a limit about a positive steady deviation happening. Most of the times cards aren't properly shuffled so affecting general (short term) probabilities.
On the same line whenever results are strongly deviating from the norm, do not continue to bet the "norm", stay put (or, at the very least, bet that the improbable stays improbable).

4) Unless you have verified that after very long trials same deeply selected events will provide more wins than losses after vig, the RTM factor will make worthless any unidirectional mechanical plan in a way or another.

as.

Quote from: Whatswhats on March 09, 2026, 10:48:25 AMASYM, I didn't understand well why you got the sequence of W/L , example why first is WL while other just W?

The plan is to make clustered 1 and 2 vs 3s (horizontally or, as in this example, vertically), so only singled 1 or singled 2 are a L, any 1-2 cluster (1-1, 1-2, 2-1 or 2-2) is a W.

For example a 1-2-2-1-2-1-1-2-2 sequence is just a W, 1-3-1-3-2-1-1 sequence is L-L-W, etc

What we should be interested about is not how many consecutive Ls or Ws we'll get along the way but the "waiting time" of the patterns.

Notice that a 3-3-3-3 or 3-3 or 3-3-3 sequence doesn't provide any classification (no W no L) as in order to have something clustered (1-2) we need one element to come out.

Then you can classify how many 1-2 clusters greater than two show up so now the "singled" losing event is a double appearance and true clusters are 1-2 successions longer than two.
Example: 2-1-3-1-1-3-3-1-2-3 that under the simple cluster classification is a W-W-W sequence now becomes a L-L-L succession; In the same way, a 3-1-3-3-2-3-1-3 sequence forming a L-L-L under the simple cluster classification now becomes a worthless registration as no 1-2 category got at least one cluster.

We ought to remember that at baccarat the "overalternating" movement is the less likely to happen so when we play the 1/2 vs 3 plan we are betting that the 3:1 ratio won't stand for long at either side of the deviation.

More later

as.

Thanks for your valuable comments.

Both replies focus about the importance of concentrating our action when things seem to go in our favor and/or when expected values are more likely to show up.

Casinos will make their enormous bac profits by exploiting the uncertainty but this will get limits of intervention capable to erase/invert the HE.
Only the "time" factor will help us to define such limits but at the same time long term data assessments (or long term experience) will direct us to play the right spots where the uncertainty is underdog to come out so favoring more likely situations to show up.

Example.

Probability that any shoe will present at least a 1-2 succession in any order (1-1, 1-2, 2-1 or 2-2) longer than one (enemies: 1-3, 2-3) is 100%, so it's just a matter of time to spot the shoe fragments where a 1 or 2 will be followed by at least another 1 or 2 before a 3 happens.
More importantly is assessing back to back shoes where precise pattern positions are less likely to form same back to back situations (ranges) for "long".

Here some shoes (just the first 14 patterns are displayed for simplicity):

1-1-1-1-3-2-3-1-2-1-3-1-2-3
2-1-1-1-2-3-2-1-1-3-3-2-1-3
3-3-1-2-3-1-3-3-2-2-3-2-2-1
1-1-3-1-3-2-3-1-1-1-2-2-1-3
3-3-3-3-3-1-3-2-1-1-1-1-1-1
3-2-1-1-1-3-3-1-2-1-1-1-2-3

So far how many 1-2 patterns longer than one are VERTICALLY showing up per each of the 14 columns? W= win and L=loss

01) WL
02) WL
03) W
04) W
05) L
06) LW
07) L
08) WW
09) W
10) LW
11) W
12) W
13) W
14) LL

Now, if you have to make some "guesses" about the next W/L patterns belonging to the simple 1/2 lenght what will you expect to get?

1-2-1-3-2-3-1-1-1-1-1-2-3-1
2-1-3-3-1-1-1-3-1-1-3-2-1-2
2-1-3-3-1-2-1-1-1-1-1-3-2-1
2-1-3-1-2-2-3-1-2-1-1-2-3-2
2-2-1-3-1-1-1-3-2-1-3-3-3-3
3-3-2-3-1-1-2-1-1-1-1-3-2-1

Again considering vertically those outcomes we'll get:

01) W
02) W
03) LW
04) L
05) W
06) W
07) WW
08) LW
09) W
10) W
11) LW
12) WL
13) W
14) W

There are infinite statistical considerations to be made, for example that the only column providing ALL 1-2 patterns is just the column #9.
On the other end, all columns getting a loss (L) on the first six shoes dealt (#5, #6, #7, #10 and #14) got a W on the next six shoes dealt.
(L happening at columns #1, #2 and #14 after the first shoes sequence didn't get a second result to be compared at the second shoes succession).

We mean that "limits" should be evaluated not only by the simple single shoes' texture but even by the back-to-back shoes sequence (vertical shape) that cannot provide harsh deviations for more than one column in most cases.

as.   

That's an interesting list, never heard of "bilking" at bac tables.

as.

Hi whatswhats!

About your question:
Let's take the casino's counterpart: does it make any difference if any player will bet 1 hour long for 24 days or 24 hours straight just once?

Mathematically it doesn't make any difference but in practical terms it's more likely that each player will lose more money by playing for 24h straight than by 24 separate 1 hour sessions as at short sessions players will try to preserve more their winnings (for example by not raising too much their bets, lowering the profit goal, betting few hands).
Of course such a consideration doesn't stand if a player will always and invariably play a rigid scheme (as a robot as you sayed): results will be in line with the tested strategy.

Variance is in direct relationship of the number of hands played (or observed) and not about time fragments, so as players we can easily expect many 1 hour losing sessions in a row and maybe a 24 hour session could be a terrific winning session not comparable with many 1 hour winning sessions.

So as long as we play with a verified long term advantage, we'll expect to win money regardless of the time sessions (and actual short term results) so the more we play and the more we win (in the exact same way casinos are confident to separate money from their customers' pockets). Hence the attitude of not raising our bets or preserving temporary profits or lowering profits now become an additional negative factor for us, providing a proper bankroll. Of course betting few hands remain the core of the advantage (or a good tool to lose less if we play without an edge).

oOoOo

Finding a possible edge

In order to find out a possible edge we reckon some factors are particularly important (decisive):

1) Data must come out from the same source (the one we'll play at)

2) The number of shoes examined

3) SD values of a possible propensity

4) Falsifying hypothesis around any corner


Data must come out from the same source

The idea that every baccarat distribution is equivalent to each other is totally wrong; Yes, itlr B and P will approach more and more the 50.68%/49.32% ratio but that's not sufficient to say that different productions are so randomly placed (or whimsically biased) to be unbeatable.
Some bac results will show up by a "biased" probability at one production but not at another one and vice versa.
As mentioned several times here, we should be focused about "ranges of apparition" applied to the same production.

Could a given production be voluntarily manipulated to get a possible propensity to stop (or even inverted)?
Not likely but possible.
Our countermeasure of betting very few hands and tracking the results under our lens will help us (factors #3 and #4) to discard a possible too frequent strong deviation(s) from a natural variance impact.


The number of shoes examined

At gambling games there are no precise answers to state how many trials we need to get some valuable conclusions for the simple reason that it's almost impossible to run innumerable propositions under the same conditions.
Yet at baccarat univocal or steady status of something won't stand for long, otherwise high stakes players would clean up many premises worldwide.

Therefore it's "incredibly" probable that what worked in the past won't happen in the next future (Alrelax quote) so in some sense we don't need very large samples to assess a possible propensity and anyway we're destined to bet a miniscule amount of the "infinite" world (quoting  Alrelax again) where very long term values cannot provide a substantial edge.

In a word and for the considerations made above, a propensity should work at "limited" ranges of intervention where no natural variance could destroy it.
Obviously if such a propensity exists (and it does) it'll be slight distributed as miracles cannot happen.
In conclusion, just hundreds of shoes dealt could tell us whether a possible propensity affects the results, especially (that's our "model control") everything else will be somewhat balanced.

SD values of a possible propensity

Those values should be assessed by a negative standpoint: so we are interested about how many negative situations a given strategy infinitely run at the same situation will show up.
It's true that a propensity must come out more clustered than isolated, yet and since the game remains a general  EV- proposition, we confide more in the probability that things will change soon after some (low) levels of negative apparition.
So if we have ascertained that A>B, volatility will be lower while betting that B will be followed by A than by constantly wagering that A predominates.
When a B-B pattern comes out things are more confusing, despite being shifted towards one side again.

Falsifying hypothesis around any corner

Average card distributions themselves provide some short/medium/long term propensities, carefully measured and controlled by rigid statistical standards, yet every serious bac player should be aware that casinos are not there to lose money.
So whenever we suspect something doesn't run "properly", we've considered (among others) some factors to look for before betting.
Number of naturals and number of ties, for example. The lower, the better.
Then the number of standing points (6s, 7s) losing vs a drawing hand. Again, the lower, the better.

Whenever you suspect the card distribution seems to be too much unfairly deviated, stick to the 1-2 vs 3 general plan or, even better, do not bet anything.
Especially when cards are shuffled by a machine.

as.

Whatswhats wrote:

For me isn't just the bet selection or the money management so not just where and how much to bet, but also the entire PLAN to follow.

Yes I agree, but at the end the entire plan must be conceived to place more EV+ bets than we can and that can only come out from a valid bet selection.
That means that to win itlr we need to collect more wins than losses (after vig) being our EV totally insensitive to the betting amount.
 
Easier sayed than done, of course.


oOoOo

Random biased shoes and unrandom biased shoes.

Basically all shoes dealt present a "bias" due to a natural (random) asymmetry, then with the introduction of shuffling machines many shoes dealt are affected by a bias we have classified as "unrandom".

The difference between natural random biased shoes and unrandom biased shoes is that the former category will generally get lower deviations of expected asym patterns, so making a preordered plan more enticing than the latter class where it seems that very strong deviations are easier to come out around any corner by a weird strong asymmetry or strong (unlikely) symmetry.

Paradoxically and without touching intricated issues, unrandom shoes will get better probabilities to cross winning (or losing) clusters by simple strategies than by exploiting asymmetry at random shoes; asymmetry being the paramount factor why we should get a long term advantage.

Our conclusions came out from comparing the same amount of real random shuffled shoes with machine shuffled shoes by assessing several factors as the back to back patterns shape, the naturals back to back distribution, the asymmetrical hands results, etc.

Whereas the asymmetry at real randomly shuffled shoes seems to be homogeneously distributed, at machine shuffled shoes asymmetry or symmetry become harshly oriented toward one side per every shoe dealt, at least from a proportional 3:1 ratio point of view.

Obviously the asymmetry/symmetry concept is strictly related to the specific succession we want to take care of.
Make your experiments and you'll see what I'm talking about.

as. 

Here we go.

If we think some shoes are affected by a kind of bias, then a possible MM could have some merit so lowering or even inverting the long term math percentages.

In reality ALL shoes are affected by a bias but it's impossible to track all the factors merging towards it.
And even though we're able to spot the bias, we never know for sure how long it will last unless we had run the same situations for many many trials (so getting us more precise sd values).

Therefore the bias isn't a potential factor shifting the results at some points but is an actual one, yet we have few legitimate means to spot it in the infinite world of random walks.

On the other end, there's a controversial thought about setting up a plan on "no biased" successions applied to the common bac successions registered.
It seems that now we're playing a "random" (so unbeatable) game but we're just taking a kind of "losing" part that can become a winning one.
More later

as.