Hi KFB!
We're looking forward to hear your report! 

as. 

At baccarat complexity is represented by numerous factors working together but at the end we just care about the destiny of our bet.

We know that some patterns are slight (or very slight) more likely to appear first than others, when not we may be interested to assess if they'd come out as 'second' (one losing step); in the event we missed both the first and the second attempt, we could reach the conclusion that the actual  'environment' doesn't fit the requisites making something 'more probable'.

In a word, we try to restrict and at the same time taking the best of the complexity by challenging a kind of binomial 'random' world to make 'too few' opposite environments along any shoe dealt.
Obviously when an environment had surpassed the empirical two step cutoff, we need a kind of 'signal' to stop (when profitable) or to restart (when unprofitable) the betting.

Recreational players and the vast majority of bac players hope that few environments appear for long in the same shoe or, worse, that the silence environment 'must' show up sooner or later and they try to force probabilities by increasing their bets.

Casinos confide upon the exact opposite thing: on average too many different (so short) enviroments are expected to show up making the baccarat outcomes a very volatile so unbeatable production.

That has almost nothing to share with the common notion to determine randomness by the number of runs.
But patterns runs are not the same as simple hands runs for a different complexity factor working, especially whenever we consider the clustering feature and back to back shoes.

It's of our interest to keep 'experts' to state that patterns runs will distribute with the same features than hands runs or that there are no valid points of intervention along any shoe dealt or series of shoes dealt.

More later

as.

Talking about a possible EV+ without precise math findings is just saying that a Ferrari runs slower than a Jeep Wrangler, but actually it's the environment that makes the difference.

I guess nobody will pick up a Ferrari 812 Superfast to drive through a rough desert road and only a fool would choose a Wrangler to compete with the same Ferrari in a long flat asphalted road.

So it's not the car which makes the difference but the environment does.

If the enviroment, equally splitted in rough and flat situations, is supposed to be 'random' produced, no car choice will get us a kind of 'profit' unless short rough segments are overcome by the Ferrari and/or short flat segments make a minor speeding impact for the Wrangler.

The 'illusion' of controlling randomness by raising the POS (probability of success) may be worthless but it surely increases the 'complexity' feature.

Therefore if a A=B proposition is unbeatable, a Ax(A1+A2)=Bx(B1+B2) proposition could remain 'less' unbeatable, as the number of outliers is someway restricted, so more likely roaming around the 0 origin point.

At the same token, if both Ax and Bx present a slight restricted number of deviations, Ax/Ax clusters and Bx/Bx clusters must take the same properties but now by a greater level of 'complexity' than the previous step, and it's reasonable to think that the more complexity is involved greater will be our POS (whether carefully assessed).

Yet we have reasons to think that the productions are not perfect randomly offered, so more likely taking a kind of biased one sided 'undetectable' direction.

Obviously if something is 'biased' it'll more likely produce clustered events than isolated events, so we do not need a lot of time to realize the most (even transitory) line any shoe will take.
We've seen that a long 'alternating' events course is the least likely to happen, providing a decent level of patterns 'complexity' factor.

Of course the BPBPBP succession (and many alike) cannot belong to the 'alternating' category as lacking of the 'complexity' factor.

See u next week.

as.

I mean that back to back patterns belonging to single shoes could be so biased (for a strong defect of random shuffling) to pose a real threat on our 'variance control' plan.
Single shoes being the fortune of recreational players but the nightmare for pros that for obvious reasons must confide on situations restricted within a wide yet 'affordable' range.

The fact that such rare 'single shoes' might symmetrically form profitable patterns (that is conforming too much with our plan) is not counterbalanced by the specular strong negative situations that put into the toilet our expectation.

It's like to toss a coin thinking it's unbiased whereas is harshly biased, but a single shoe cannot be a reliable source of consideration for obvious reasons.

Therefore it's way better to 'challenge' the system by spotting few positional events taken from endless shoe successions.
Yes, we need a kind of bias to succeed, but this must be limited in some way and at either way.

Naturally in the vast majority of the times what seems to be 'unnaturally' distributed at single shoe's Big Road will give spectacular successions at one of more derived roads and vice versa. But  the issue remains the same: We can't know what should be more likely to happen at a specific line,  unless that single line is considered after numerous trials and upon the same positional probability.
 
Casinos do not hope we'll lose or win 8 hands in a row so frequently, but they know we will.
Unfortunately what we had won after an 8 winning streak will be overcome by a specular 8 losing streak (for the vig or for the asymmetrical BP probability).
We have found that such situations happen more probably within a single shoe dealt.

As long as you consider patterns by taking care of multiple selected and different 'environments', probability to lose will be more restricted than at those (very rare) single shoes.

In addition and being the above conditions fulfilled, the more you'll wait for a unlikely deviation, higher will be your EV+.

as.     
 

WOW, you exchanged e-mails with Nickerson!

I agree with you that it's worth reading some coin flip published researches.

For example, Hahn and Warren and then Sun and Wang have clarified that the 'waiting time' is the longest for streak patterns. Yet the waiting time measures the average amount of delay in a pattern's appearance time and this differs from the frequency of occurrence of the pattern.

Some patterns present 'contradictions' when considering both the waiting time and the probability to show up first (Gardner).
Take THTH and HTHH patterns. THTH has a waiting time of 20 and HTHH a wating time of 18, yet the probability that THTH will appear sooner than HTHH is 0.64 vs 0.50.

On the same line, Nickerson (along with Gardner) stated that both HHT and HTT patterns have equal waiting times of 8, but HHT is expected to occur before HTT two thirds of the time.

Obviously such findings aren't directly related to the baccarat probability to encounter this or that first or later, mainly as the baccarat 'coin' is biased at the start and especially it's 'dynamically' biased for the ever changing card distribution that could be considered as 'neutral' just at the start of any new shoe dealt.

Whether previous patterns of the same shoe really have a slight effect over the next results is a more complicated issue that can't be disjointed by the general baccarat features coming out from large samples analysis.

More later

as.

A book is coming, something as 'Asymbacguy on baccarat'.

It's a strict technical book where I'll present first all the theoretical ideas we've based our strategy upon, then the precise and step-by-step plans that worked for us after playing innumerable HS sessions worldwide.

A book for advanced players for sure, illustrating that it's very harsh to erase and invert the HE but it can be done.

as.

HT patterns have the same long term frequency to appear but featuring a different rhythm of presentation

It's an obvious statement that makes 'unbeatable' any binomial production 'guessing'.
Casinos prosper about the 'uncertainty' of the outcomes offered, yet this uncertainty could be measured at least by interesting 'estimation' values that might shift the uncertainty world into a more detectable world.

If our 'estimation' really works and it's capable to overcome the HE, we'll play with an advantage otherwise we're simply belonging to the baccarat losers category (almost touching 100% of all players worldwide).

To realize whether we're fooling ourselves (or, worse, others) we need to test our strategy on very large samples, moreover by scrutinizing our results by simple distribution issues that have nothing to share with the betting amount we've employed unless belonging to the same category.

Therefore any unit class must get a long term W/L ratio, that is must win by flat betting, otherwise we should directly skip to the superior unit class which had demonstrated to be worthwhile. And so on.

In fact, there's no point to bet 1 if we know W=L and the same for superior betting classes.
So it's not the betting amount which cares, the W/L ratio does.

That is most part of the baccarat spots (W=L) are particularly susceptible of variance at either way: When we'll catch the positive side we'd think to play with an advantage and when we endure the negative side, we feel ourselves as unlucky or having made many wrong moves.

But the only pure wrong moves to make at baccarat are just represented by playing too many hands. 

Actually and besides of rare opportunities happening, the 'subjective' element cannot get any significant room to produce profits. After all the vast majority of players adopt this way of thought and we all know the results.

OoOoO

If you'd assign a +1 or -1 number to the innumerable back to back situations that bac provides, you'll see that no consistent values will overcome others, unless you set up 'limited' random walks movements.
Even by following this procedure, variance doesn't make 'easy' to get more wins than losses for the strong asymmetrical amplitude coming out from the card distribution, needing a fair number of decisions to detect what's is really more probable to show up at some point.
On the other end, what is supposed to be more likely by math remains more likely to show up, independently of the spot we've chosen to bet (then it could be taken randomly). 

So we may postulate that the 'uncertainty' is inversely proportional to the number of hands dealt.
What some authors name as 'RTM effect' or 'correction effect' and at the same time confiding that something is more likely to happen first by various reasons no matter what.

IMO it's the delicate (measurable) assessment of the aforementioned features that helps us to find the positive  or negative 'more probable' succession lenghts.

as.
 

Nice thread, indeed.

as.

Measuring the variance impact

We all know that by increasing the number of trials, both 'ideal' and 'terrible' situations will come out, yet even at a 50/50 coin flip situation some patterns are proven to show up first than others (Konold, Nickerson, etc).

Now we're posing the question about what can do an asymmetrical starting probability considered by certain 'more likely' patterns, well knowing that itlr a kind of 'exhaustion' and 'reversed situations' will take place.

Notice that if the probability of success will be homogeneously spread by exact probability values, our EV=0 (before vig) as per every 3 wins whatever distributed a single loss will balance the previous wins.

On the other end, odds to get a first W per every 4-hand sequence theorically are 3:1.
Actually and by taking advantage of the statistical features presented above (with many thanks to M.v. Smoluchoski), your bets are tremendously EV+.

Casinos can't do nothing about it, as long as there's a cut and a burning procedure (so discounting a very very unlikely illegitimate voluntarily card distribution negating this feature), probability after effects will get the best of it by a 1 billion of accuracy.

as.

That's interesting (I've read your whole thread on that), yet it's quite difficult to put some of those thoughts into practice, besides of a general 'emotional neutralism' attitude best accomplished after  several winning sessions and very few or 0 losing ones, a thing that it's very unlikely to happen for the most part of players.

Casinos treat baccarat the way really is: an endless series of 'finite' successions of binomial outcomes where it's virtually impossible to guess more winning hands than losing hands and not only because for each winning bet they unfairly pay 0.9894:1 or 0.9876:1.
In fact casinos are so sure about the impossibility to guess more right than wrong, that allow some very high stakes players to get 'rebates' (up to the point that losses are 50% decurted, for example).

Technically, casinos rely upon variance and not about the math edge.
But obviously math edge decreases the players' positive variance and at the same time worsen the players' negative variance.
So players profit lines will slowly be asymmetrically shaped toward the negative territory.

I like your 'consider the process and not the outcome' thought.
This thought could be interpreted in several ways, it would take a 1000 pages book to partially illustrate the many situations baccarat provides along the way. Yet it's an important feature to look for, IMO.

as. 

Btw our approach, albeit being strictly based upon long term datasets, might not be the best method to beat this game. Alrelax and KFB (among few others) have presented very interesting and in some cases pivotal ideas to play with a kind of advantage.
But it takes a fair amount of experience to know what to do in the intricate situations (that constitute 90% of bac circumstances). And of course 'no betting' is only a partial 'best move'.

Variance

Variance is both a 'quantity' and a 'quality' issue; whereas quantity is impossible to 'forecast',  quality is more affected by dependent variables as any shoe dealt will be the by product of an asymmetrical card distribution where key cards play a huge role over the final outcomes.
Not every time but most of the time. 

In the vast majority of the times, casinos collect their huge profits by exploiting the players' "variance's quantity' Fallacy, that is their 'hope' that things will change very soon while losing and that things will stand for long while winning.

Technically speaking it's just a 'permutation' issue that has very little to share with the HE.
Unfortunately most players don't give a damn about the 'permutation' issue, hoping for endless profitable situations.

We've seen that the permutations tool will affect the 7-step progression cycle, as per each cycle considered, out of 128 possible patterns 93 are winning and just 35 are losing.

More interestingly is the fact that by raising the probability of success, the probability to cross favourable situations considered by a 'quality factor' will be more and more predominant.
Up to the point where after a given rare deviation, our edge will be astoundingly high.

See you ina couple of days.

as. 

Hi KFB and thanks.

It's a very simplified concept related to a random walk working at a random (unbeatable) environment  and at a unrandom (beatable) environment where we still do not know which lines will be more probable to show up.

Since the winning or losing probability is symmetrical at random sequences but asymmetrical at unrandom sequences, we may get a better picture of the endless successions volatility just by 'randomly' betting some events by a 'time' factor decided objectively and not subjectively.

More on that later

as.

"I think every game can be beaten"

Richard Munchkin (professional gambler, author of "Gambling Wizards")

as.



 

Today is 06/13/2023 so I'll randomly take from my datasets the 6th and 13th 'event' of each shoe of the 2023-2032 ten shoes sample.

1)  WL
2)  WW
3)  LW
4)  WW
5)  WW
6)  LW
7)  WW
  WW
9)  LW
10) WW

So the sequence is: WLWWLWWWWWLWWWWWLWWW, that is a -++++++++ sequence (too good, probably!)

Let's take of the same sample the 5th and 12th 'event' (one event back):

1)  WL
2)  LW
3)  LW
4)  LW
5)  WW
6)  LW
7)  WW
  LW
9)  WW
10) LL

So the sequence is WLLWLWLWWWLWWWLWWWLL = ---+-++++++-

Now the 7th and 14th events (one event forward):

1)  WL
2)  WW
3)  WW
4)  WW
5)  WW
6)  LW
7)  WW
  WL
9)  WL
10) WL

So it's WLWWWWWWWWLWWWWLWLWL that is a -+++++-+- succession

Not a recent good run at final W clusters, so we'll take the 2033-2042 shoes sample just at the original 6th and 13th spots (that already got us a steady profitable succession, so somewhat fearing a kind of 'balanced' negative factor)

1)  WL
2)  WL
3)  WW
4)  WW
5)  WW
6)  WW
7)  WW
  WW
9)  WW
10) WW

It's a WLWLWWWWWWWWWWWWWWWW sequence. That is a  -+-++ sequence
Notice that W isolated came out four times consecutively isolated (considering the previous sample). It's a very very unlikely scenario yet balanced by the L isolated patterns. 
In addition, by just playing towards W clusters we'll waste the opportunity to win at the interminable final W succession. But itlr by hoping to get the minimum level of W cluster we'll save a lot of money.

Again let's see what happens at this new shoes sample by spotting 5th and 12th 'event'.

1)  WW
2)  LW
3)  LW
4)  LW
5)  WW
6)  WW
7)  LL
  WL
9)  LL
10) WW

So it's a WWLWLWLWWWWWLLWLLLWW = ++-+-++---+.

Now the 7th and 14th events spots:

1)  LL
2)  WW
3)  LW
4)  LL
5)  WW
6)  WW
7)  LW
  WL
9)  WW
10) LW

It's a LLWWLWLLWWWWLWWLWWLW succession = -++--++++++.

'Fortunately' such results cannot be labeled as suspicious as L>W and in one occasion (first 6th/13th sample) Ws were distributed in the worst possible way (according to the plan we're talking about).

But there are still many issues to take care of regarding the probability of success.

as.

Baccarat random walk(s)

Let's consider the 0.75% A probability vs the 0.25% B probability. It's a 'so called' symmetrical probability as itlr A=B, yet the A and B distributions are asymmetrically shaped in the vast majority of the times.

Take a 4 'event' sequence of W (p=0.75) and L (p=0.25) that could only form such patterns:

WWWW
WWWL
WWLW
WWLL
WLWW
WLWL
WLLW
WLLL

LLLL
LLLW
LLWL
LLWW
LWLL
LWLW
LWWL
LWWW

Of course 3/4 of the times we'll get the first eight patterns starting with a W and just 1/4 of the remaining times eight patterns starting with a L.

It's interesting to notice that by betting toward W clusters and L isolated events and adopting a 1-2 mini progression (vig ignored for simplicity) we'll get:

+4
+1
+2
-2
-1
-5
-6
That is a overall -7 unit loss

-3
-3
-6
-2
-5
-1
+2
+2
That is a cumulative -14 unit loss

This is a very difficult concept to grasp as obviously those 16 patterns get a quite different probability to happen, strongly shifted towards W rich patterns.
In an ideal world where the W/L ratio is 3:1, out of the possible scenarios (albeit getting a very different probability to appear) just WWWL, WWLW, WLWW and LWWW patterns will follow precisely the above 3:1 ratio. Yet among the four possibilities, the WLWW sequence will get us a -2 loss (-3 and +1).
The remaining three cases are: WWWL (+1), WWLW (+2) and LWWW (+2).

In summary, the W clustering/L isolated strategy applied to sequences following the exact W/L 3:1 expected ratio will provide us 3 units of profit (minus vig).

Let's see what happens at the patterns where the number of W/L is 2:2.

Those patterns are: WWLL (+1, -3 = -2), WLWL (-3, +1, = -2) , WLLW (-3, -3 = -6), LLWW (-3, +1 = -2), LWLW (+1, -3, +1 = -1), LWWL (+1, +1 = +2).
Overall it's a -11 unit deficit but besides of the WLLW pattern, losses are restricted within the -2 limit.

Then there are the W/L 1:3 sequences and they are: WLLL (-6), LLLW (-3), LWLL (-2).
Again only WLL(L) sequence got us a -6 deficit.

Finally the homogeneous W or L patterns where WWWW (+1) and LLLL (-3).


Splitting outcomes in four 'events' categories was just an example, we have learnt that an asymmetrical (and unrandom) system remains asymmetrical independently of the precise spot chosen to be bet or considered.
For example, if the previous shoe had dealt a L in the examined spot, odds remain to make more probable a W than another L and the same is true about a single W towards getting another W (so forming a W cluster).

In the endless 'random' sequences we'll have to face, the W/L 3:1 patterns must take the more probable platykurtic curve where we are strongly favorite to win.
Of course there are harsh negative outliers and they are: WLLW and WLLL sequence; no other patterns 'rich' of L can get us a stronger damage, that is going too far from our expected more likely scenario.
Fortunately such bad situations ar quite 'balanced' by 'occasional' W coming out independently of the shoe we're playing at as 75 is always more likely than 25.

See u next week for a an interesting strategy I've recently been aware of.

as.