A personal test for bac randomness

Our group is made by frequentist probability lovers, in the sense that we like to collect data coming out from the same exact source and then building a probability theory.

Even the "same source" concept could be a volatile definition: think about shuffling machines operating at two alternate shoes lasting for a X time (number of shuffles per each shoe).
We've found important differences if the same shoe did undergo one or two shuffles or multiple shuffles.

Therefore if we want to exploit the "average" card distribution tool, we want to play at properly shuffled shoes.
Remember the comparison with black jack: low cards-neutral cards-high cards decks (in any LNH sequence) completely deny a card counter math advantage.
Of course such situation could easily happen for natural reasons, but we never know if it seem to appear for "too much" long.

At baccarat we've personally devised two valuable main tools to take care of in order to approximate whether a shoe is really randomly shuffled or not.

a) the math advantaged two-initial cards points losing "too many times" despite of their math propensity to win;

b) a higher than average ratio of hands resolved by 6 cards.

Of course those are the two main factors, there are other minor parameters to look for.

Realize that there's no way to win at baccarat itlr if our bets will get the inferior 2-card initial point as the number of drawouts will be underdog to get a long term edge.
Thus whenever the drawouts are coming out "too often", we theorized that that shoe was improperly shuffled. So unplayable.

Hands resolved by 6 cards is an additional factor to look for and is related to the high neutral card density (more than 30%) along with the 6s,7s,8s and 9s class (again more than 30%), then to  other less likely card combinations forming natural points as 5-4, 5-3, 4-4 or standing points as 5-A, 5-2, 4-3, 4-2 or 3-3.

Card distributions not forming those situations AT BOTH SIDES for long are relatively rare and when they're not (that is they are coming out too often) we could assume a kind of randomness bias.

Paradoxically it's better to move around a strong good or strong bad choice than navigating into a more undefined world where too many cards will dictate the actual result.
That's because an overalternating shifted world will be the least situation to happen.

as. 

IMO at baccarat the only reason why we could win is because of the more likely card distribution ranges.
The actual results do not necessarily be the by product of more likely card distribution ranges as (beyond the natural variance) there still exists the important factor regarding the shuffling more or less randomness.

Our data had taught us that a perfect randomness or a slight defect of randomness will go to our favor as best represents the "more likely card distribution" ranges.

Bad shuffled shoes need too much complicated algorithms to be resolved (approximated) and of course we never know how "bad" a shoe is shuffled and more importantly the more probable patterns to look for.

In poor words, we'll win a lot or lose a lot when shoes are badly shuffled with 0 impact of skills, whereas perfect random or near perfect random shoes will give us plenty of informations to draw on.

more later

as.   

At baccarat the definition of asymmetry and symmetry is particularly intricated for several reasons:

- the model is slight asymmetrical at the start (B>P)

- the model is affected by a huge first-step asymmetrical distribution of cards, specifically of key cards

- the model is affected by a huge second-step asymmetrical distribution of third(s) card(s)

- the model is finite and dependent, meaning that each situation (hand dealt) won't cross through the exact same parameters.

Overall and simplifying the issue, we might infer that symmetrical events are just "incidents" made along the way.
After all statistics give us plenty of examples where asymmetry will lead over the symmetry, the latter now intended as a steady expected probability happening for long.

Actually at baccarat there's no symmetry involved other than by coincidental factors and when some low levels of asymmetry are surpassed, more often than not a slight subtle force will shift the results in order to deny a kind of "balanced" results.

Therefore the rule to follow is expecting "low levels of asymmetry": whenever this rule seems to be "violated" best action to make is staying still (no betting) or to wager that the lowest levels of asymmetry will remain as silent.

After all we have strong reasons to think that cards are randomly arranged to get more asymmetrical patterns than symmetrical patterns and when this isn't true is just for a temporary and coincidental short term effect.

Examples of typical asym or sym patterns

ABAA = asym
ABA  = sym

AABBB = asym
AABBA = sym

AAABA or AAABBA = asym
AAABBB = sym

Overall we could accept the idea that asym=sym, yet we should be more interested about how many sym events will shift into asym events or vice versa and, more importantly, at which level of asymmetry or symmetry. Per every shoe played.

Now we might use a formula based upon that asym-asym > asym-sym; sym-asym > sym-sym for the most probable asymmetry/symmetry levels of apparition (0, 1 and 2).

We know that an asym/sym/asym sequence lasting for long is the least occurence to happen and the same is about long sym/sym successions.
The remaining probability world is what we should be interested to focus about as proportionally taken (remember the 0.75 probability to happen) asym/asym..., sym/asym and sym-sym/asym patterns are way more probable to naturally come out so maybe enticing (I've sayed "maybe" as a simple flat betting scheme will get the best of it) a multilayered progressive scheme actioned by one or more two losing sym sequences happening at different sections of the shoe.

Assigning a 0.75 general winning probability to a asym/sym independent model, the average expected W/L ratio is 3:1, so unbeatable for the vig or P asymmetrical probability.
In reality baccarat slightly endorses the asym patterns formation in the way that each asymmetry level will be more probable than the symmetry counterpart.
Naturally the sure asymmetry will make coincidental symmetrical patterns along the way, that's why  we have to restrict their appearance by assigning or not them to a more likely sequences category.

I know, that's all rattlesnake.sh.it.
Fortunately. 

as.

Hi lp!

Actually things are much more complicated as I talked about A and B events and not about mere B or P hands.

For example a BBBBBB or BBBBBBBBBBBB or PPP or PPPPPP patterns should be considered as asymmetrical (or symmetrical) in relationship of the previous pattern and not by their shape alone.
Therefore BBBBBB could be either an asymmetrical or a symmetrical pattern by what came out previously.
The same about blue/red derived roads or any other random walk you want to utilize.

Then, since each shoe is a world apart, levels of confidence should be approximated by the number and ranges of asymmetrical or symmetrical situations just happened.
Low symmetrical patterns are a general rule, but the actual route must be carefully defined as just one hand could transform an asymmetrical pattern or sequence into a symmetrical pattern or sequence; obviously such thing might happen by an opposite fashion, anyway not constituting the propensity we're really wanting to exploit.

More later

as.

"WITH MUCH APPRECIATION AND GRATITUDE"

Yep, I cannot find better words than Vic's

Actually I hope you'll find the time to keep posting regularly.

as.

We're deadly sure that the asymmetry/symmetry issue is the main tool to extract an advantage from.
Simplyfing:

S-S = stop the betting

S= what to bet next depends about the previous As/S texture

As-S= bet As

S-As= bet As

As= almost always bet As

As-As= caution, sky's the limit but what we have secured should remain in our pockets.

So any hand dealt is a new hand my a$$, cards are arranged to make more probable asymmetrical situations than symmetrical ones.
Situations seemingly belonging to a kind of steady symmetrical world are just springing from incidents, natural variance or weird unlikely card distributions.

Now a careful multilayered progressive plan cannot be wrong as:

First level of confidence:

As-As > As-S

S-As > S-S

Second level of confidence:

Any S-S happened previously somewhat reduces the S-As propensity at a new pattern.

Any As-As and S-As events happened previously endorse the As-As and S-As propensity at a new pattern.

Third level of confidence:

- As-As coming out clustered twice previously is not a good indicator to bet again towards another As-As pattern;

- S-S coming out clustered twice is a good indicator that the next S will be followed by As and not S.

- S-As coming out clustered twice is an excellent indicator that next pattern will be of the same shape (that is that S will come out as isolated again).

To provide a vulgar example say the shoe went as:

BB
PP
BBBBBB
PPPPP
B
P
BB

Such fragment will form a S-S-S sequence (BBPP, BBB../PPP.., BP).
Run the derived roads and such symmetry will go down the toilet.
In fact only the sr will form one symmetrical pattern. Everything else will produce asymmetrical patterns either at sr and at the remaining roads.

as.

As long as cards are asymmetrically distributed along any shoe dealt (100% of the times),

As clusters > S clusters and As isolated events < S isolated events.

Unlikely distributions, incidents and natural variance will make the overalternating mood as the least probable happening so possibly inverting the above propositions.

Therefore sometimes S clusters will predominate over As clusters; at a lesser degree S isolated events will be more restricted in their appearance than average so privileging S clusters.
That further denies the overalternating mood appearance.

When in doubt go either for low/moderate levels of asymmetry better by trying to by pass the lowest losing levels or tell the casino to wait for inferior players action by not betting a fkng nothing.

If properly executed and by choosing the right random walks,  this plan is equal or superior than the edge sorting technique.

as.

Asymmetry/symmetry considerations about bac successions

Symmetrical situations are slight less likely than asymmetrical situations and of course asymmetrical situations cannot stand for long most of the times.
More precisely asymmetrical situations will come out more probably (so more clustered) than symmetrical situations, anyway we have to respectively approximate at best the asym/sym appearance by setting up a "cutoff" value as just one hand might easily change a more natural (expected) flow into a moderate/strong deviation pattern. Obviously we're way more interested to avoid moderate/strong symmetrical distributions than stopping the asymmetrical counterpart.

Suppose we have two fighting A and B events (patterns, situations) forming an original succession and several derived sub successions.

We might empirically consider as symmetrical (s) those patterns:

1s) ABAA.. and BABB..

2s) AABBA and BBAAB

3s) AAA..BBB.. and BBB..AAA..

On the other end we'll get the asymmetrical (a) patterns counterpart being:

1a) ABB or BAA

2a) AABA or BBAB and AABBB.. or BBAAA..

3a) AAA..BA or AAA..BBA and BBB..AB or BBB..AAB

Once a pattern had surpassed the 2-step level we have taken as a "cutoff" value and dictating whether a pattern is either labeled as asymmetrical or as symmetrical, we're not interested to know what the fk happens next, unless a new opposite category shows up.

Now we have transformed a BP succession into an Asym/Sym sequence that is slight less likely to produce strong overalternating moods, especially if we are taking care of the different 1-step and 2-step probability situations.

That means that asymmetrical or symmetrical spots are more probable to produce clusters at any side of the spectrum and when they're not the asymmetrical subtle force itlr will deviate the results by forming low-level more likely ranges.

What is really interesting to notice after thousands and thousands of real live shoes tested is that whenever a given precise Asym/Sym pattern level hadn't shown up so far, we should consider it as a kind of "not existent" pattern so increasing the probability of other patterns coming out.

As a side note we've found particularly useful to read and study the Yerkes-Dodson law (1908) as without the use of a software baccarat remains a strong human challenge vs a "machine".

as.

Hi lovepreaks!

Basically 5-card hands are the core of the asymmetry whereas 4-card and 6-card hands constitute the symmetry. Of course we are interested about the "ranges" of such different situations transformed into actual results so we're obviously forced to consider asymmetry and symmetry at the real BP sequence and derived successions.

It's a fact that at baccarat the least distribution to show up is a kind of "overalternating" mood of two fighting opposite events; an example is when consecutive doubles come out in a row thus the CFS is 1 per every two hands dealt (+-+-+-..).
The most part of the remaining distributions are made of univocal distributions getting low/moderate levels of deviation (++, +++ or -- and ---).

When complex patterns are considered, better will be the prediction as whenever many hands are needed to produce a pattern, the overall asymmetrical strenght will get rid of both some "incidental results" and natural variance.

Progressively wagering B after P and PP is not an option that I recommend as those are patterns too easy to come out so lacking the "complex" requisite.   
   
I have lost a lot of money by betting purely based on emotions

We all did.
Casinos prosper about players' emotions a lot more than by exploiting the math edge.

Baccarat is a strict technical game where successions are somewhat "biased" by more likely ranges that have nothing to do with mere math; it's up to us to define what are the more probable ranges intervening in the process and without any doubt in many instances best move to take is to stay still.

More later

as.

BTW, thanks for your interest in reading my pages and keeping to read this wonderful site.
400k views are a very good accomplishment, thanks.

If anyone has a bit of hope to beat this game (and taking back the money lost), well, you are in the best place to get some hints. By far. And, more importantly, for free.

There are very illustrious players posting here: Alrelax and KFB. Both are very experienced players and at baccarat experience and proper knowledge is everything.

We know that as long as a verified math propensity isn't devised and proved, well we're talking about bighorn.stuff, but that's an advantage for us.
At baccarat we can play mathematically "wrong" and betting thousands by being labeled as clowns, no problem with that.

After all, not every baccarat supervisor will be 100% certain that the game is unbeatable.
And they are right. 

as.

That's good Al, we'll wait for your link

HERE: ((With reasoning, not armchair quarterbacking).

https://betselection.cc/actual-baccarat-shoes/think-and-be-smart/msg72689/#msg72689

Baccarat results are distributed by a kind of an on/off model but the average card distribution is not, meaning that some distributions are more likely than others.
That means that in the short/intermediate terms a slight but constant force will polarize the results by more likely ranges that of course will be "almost" neutralized by those not average shoes deviating at various levels from the norm.

In a nutshell our betting options are restricted into three different categories:

a) A pattern or two patterns are going to be clustered at least one time (a1); then we'll get a2, a3, etc. clustered scenarios;

b) A pattern is going to be isolated (I), that is followed by an opposite pattern;

c) A pattern or (more unlikely) two patterns had never shown up so far (N).

We've seen that average shoes will more likely include a1, a2, a3 and I patterns, whatever arranged (permutation issue); N is the pivotal factor to get most of our edge as an asymmetrical world is destined to get "silent" the remaining elements so endorsing the probability that all possible other patterns will come out.
Now we shouldn't be worried about the precise pattern's shape (a1, a2, a3 or I) coming out, we'd just try to get rid of the pattern that had never shown up so far, meaning that it's just a "level" consideration to be made and obviously it's better to wait a given pattern to be built before thinking to place a bet.

By doing this we're limiting the randomness world and exploiting an average asymmetrical card distribution as "silent" elements will let the remaining patterns to be more clustered than average.

Obviously up to a point as low numbers (I, a1, a2 and a3) will be slight more probable than huge numbers.

Notice that even if casinos virtually want to deal "all patterns" being balanced, such thing comes out at our favor as now pattern ranges are more limited than average.

So the main group of fatal mistakes to do are chasing what didn't appear so far and hoping that a given univocal pattern will come out "too" long.

Examples.

Shoe #1 (many hands were discarded at the start of the shoe--precious hint).

I-I-I-I-I-I-a3-a3-I-a1-I-a1.

Think of every element coming out, not knowing what will come out next: Are we going to bet towards a2?

Shoe #2.

a2-a1-a1-a3-I-a1-I-I-I-a1-I.

In this shoe every element had come out, anyway just one deviated pattern happened (a3) and at the intermediate/final portion of the shoe only low number patterns happened (I or a1).

When in doubt go either for the silent pattern to prolong its absence or towards low numbers (that is anything but a a3).

Mix every possible pattern whatever you want, itlr either one or two silent patterns will happen for a decent time or, more likely, low numbers (I and a1 OR a1 and a2) will come out clustered at least one time.

Such propensity doesn't work ALL OF THE TIMES but MOST OF THE TIMES, so capable to reach and surpass the famous B 51.3% and P 50.1% probability cutoff values worth of playing a sure fkng indeniable EV+ game.

See you next week

as.

Al wrote:

I have capitalized very nicely as well as missing the opportunities being presented because I was desiring something and not realizing at the instant time, what being presented cannot be changed to what I desired, etc.

Yep, as humans we make plenty of good plays and a greater amount of mistakes, that's why almost nobody will lose the expected EV- but way more than that.


What are we really going to exploit?

Once seated we should have a clear vision of what we're going to do. 

Each bet remains EV-, no mechanical plans can beat it, let alone a mere trend following strategy or other human shoe's compliance approaches.

Maybe a mix of the two could be helpful but the main issue we should be focused at is the classification of the actual shoe needing several factors to be ascertained.

More later

as.

Good quote.

Here's the quote.

Just because you are winning does not validate your edge any more than losing invalidates it.

DO NOT LET YOURSELF SLIP INTO GOD MODE

it often precipitates a fall. Your results are a product of the EV you generate.

as.

Asymmetry and symmetry

Baccarat is a sure asymmetrical model as card ranks cannot be distributed proportionally along any portion of the shoe dealt. And there are many cards forming a bac shoe.
Moreover when cards are properly shuffled, different ranks are getting a "more likely distribution" producing a large number of "low" asymmetry and few sections of "moderate/high" asymmetry.

Virtually the symmetry doesn't exist unless for a coincidence; And we can't forget that the two fighting events (B and P) will get an asymmetrical probability at the start.

But if you want to try to make a living at this game you should understand that there are many different shoe productions incorrectly considered as equal when they are not.

Randomizing a 416 cards shoe is not an easy task and actually we players cannot have a bit of knowledge about how and how much shoes are really randomized.

The only situation to get more informations is when a new shoe or a same shoe is either manually or repeatedly shuffled under our eyes and this thing happens quite rarely.
In fact nowadays the vast majority of bac shoes are:

- machine shuffled (mainly by a SHFL machine), utilized more than one time with two alternating decks;

- preordered shuffled, utilized just one time.

Honestly and besides Vegas and some other US casinos, we do not trust any casino in the world so we're prepared to assume that cards cannot be properly shuffled at any shoe dealt.

The utmost interest of any casino in the world is to offer random situations, yet strong unrandom distributions itlr tend to favor the house and not the players.

Think about those "weird" blackjack infamous shoes coming out in a row when the distribution sounds as low cards/neutral cards/high cards and where it's almost impossible to extract an edge.

Believe me or not, our algorithms had taught us to realize when a shoe is profitable (so enticing a larger number of bets at that shoe), neutral (so itlr producing a slight loss for the EV-) or unprofitable no matter how deep we'll select our betting.

At the end the common denominator is the asymmetry level: When asymmetry reaches too high values at the shoe played (B category), we're navigating the "tourists' hope", meaning that this shoe is too much affected by a NOT average distribution enticing a possible negative multilayered progression. Anyway best option remains to stay still.

On the other end, those more probable shoes belonging to the average category might be exploited even better by a progressive multilayered positive plan, especially towards clustered events of low levels of happening.

Example.

No matter what random walk utilized, suppose we're taking care of doubles vs superior streaks.
Thus 0=no double (so a superior streak), 1=one double intertwined by two superior streaks, 2=two doubles between two superior streaks, and so on. 

More probable asymmetry levels about doubles move around 0, 1 and 2 levels.

The actual (real) shoe went as 0, 0, 0, 3, 0, 1, 2, 1, 0, 2.
A derived road went as 0, 0, 0, 0, 1, 0, 3, 1, 0.

Another shoe:

0, 0, 0, 4 (so 3), 0.

Derived road: 4 (so 3), 1, 1, 1, 2.

Now let's take the opposite situation, that is superior streaks vs doubles.

3, 0, 1, 1, 0.

Derived road: 4 (so 3), 1, 1, 0.

Tourists hope that "huge" numbers will come out for long or that a given same number will show up as long as the shoe could.
Professionals will take the same route whenever LOW numbers come out shortly but when "huge" numbers (dictating a greater than average level of asymmetry) seem to be predominant (suggesting a moderate/high asymmetry level) are simply not interested to chase a kind of "too deviated" world, maybe discarding from the possibilities very low numbers as 0.

We're fully aware that those considerations are directly falling into the Gambler's fallacy world and actually we have no reason to let "experts" to think otherwise.

All baccarat players are pure donators, period. 

as.   

Hi KFB, thanks, it's the same for me. Even if I don't reply very often, I really read and reread every post you present here.

Distribution of pattern numbers

That given numbers alone cannot get us an edge by predominating over other numbers is sure as hell, yet each shoe dealt will present a "more probable" numbers distribution for the finitess of the elements producing the results and for the related math features.
For example, when a huge number (3) shows up, we have to "guess" what will be the more likely next number to come.

Long tests have taught us that after a 3 number, there's a very slight propensity that next probable number will be 0 or 1, then 2.
Thus the least probability is assigned to another 3. It means that back-to-back "huge" number patterns are not coming out around any corner.
This shouldn't lure us to bet for any number different than 3 after a 3 even though a same succession won't form many simultaneously derived lines having the 3-3... shape.

Even worse is thinking that after a 3-3 succession, best bet to make all the time will be against one more 3 number.

Actually any 3 single number should be considered as a sign of a moderate/strong asymmetrical distribution deviating from the more likely "light" natural asymmetry.
Good news is that an interesting part of total shoes won't perform a single 3, so giving us a kind of "freerolling" by betting any of the other numbers.

On the other end of the spectrum 0s vs any superior number or 1s vs (2s-3s), will constitute the core of the light asymmetry.
Now differently than other mentioned techniques getting a 0.75%/0.25% general probability, here we are talking about a kind of 50/50 probability propositions.
Naturally linking 0s and 1s vs anything else will merge into a 0.75 p. 

Interestingly and obviously, the light asymmetry (0 and 1 numbers) tend to come out either  clustered at some levels or rarely distributed along any shoe*.
Most of the times single shoes do not produce balancements of a previously silent number, paraphrasing it's the classical example of "very good shoe" (no balancement) or "very bad shoe" (many balancements, thus chaotic undetectable flow).

*: Chasing the light asymmetry to be clustered is a way minor mistake than chasing a number never happened or few happened so far, especially if it's a huge number.

Labeling a shoe into a more probable category ASAP

Schematically and even knowing that things could (!) change along the course of a shoe, we'll have just two shoe types:

A- Light asymmetry predominant shoes (average shoes)

Patterns are consecutively short, huge numbers come out rarely or even not at all.   

B- Moderate/strong asymmetry predominant shoes.

One or two long patterns apparition is a long term reliable tool to look for, two huge numbers coming out rapidly are a fair sign of strong asymmetry somewhat affecting next shoe parts.

There's another important technical factor helping us to approximate at best which A or B category each shoe dealt belongs to that I can't discuss here.

At the end, average shoes entice a low numbers betting placement; Conversely B category should orient us to get rid of just one number: 0.
That means to encourage the use of a multilayered positive progression at A shoes and a multilayered negative progression at B shoes.

See you next week

as.