Hi KFB!

I think (you too, of course) that the best bac players are 'situational' players, meaning they like to bet toward some value of clustered events that happened so far than hoping that an event 'must' happen sooner or later.

So for example, if my strategy tells me to bet P I could place an additional Panda side bet, providing it came out by a higher frequency than probability dictates, that is following the 'rare events come out in clusters or stay silent for long' tool.

The 'betting P and F-7' is a more intriguing matter (the same about betting P side and placing a 'Lucky 6' wager).

Personally I very rarely play at EZ bac tables, so I can't tell you about the first attack, yet it could happen that I'll bet P side simultaneously wagering a 'lucky 6'.

The reason is because 'Lucky 6' is a well more likely possibility than F-7, at the same time knowing that Lucky 6 tables provide a better negative edge at Player bets (1.24% vs 1.46%).

In some way, my P bet must shift many Banker winning points but a 6. And those are surely coming out at least once in the vast majority of shoes dealt (differently to F-7). 

Many casinos have changed some bac tables from a pure 5% B commission game to "Lucky 6" tables as their math advantage will raise from 1.06%/1.24% to 1.46%/1.24%, forgetting that Lucky 6s are more constant to be distributed than F-7, for example (now the HE at B bets is around 1.01%).

The main error (obviously dictated by math experts losers and math probabilities) is that three cards are needed to form a F-7, whereas in the vast majority of the times a Lucky 6 needs just two. The same about Panda bet.

I think that any bac player should bet toward more likely situations, thus two-card happening will be more likely than three-card happening.
The same about simple B and P bets: we do not want to bet a side needing a third card to improve its point.

as.   

Nice job by any means!

Congratulations!

as.

More samples to look for:

WWW
WW
LWW
WWW
WWWW
W
LLLW
WWWL
WW
LLW
WWW
WW
LW
LWLWW
WW
WW
LW
WLW
LWW
WW
WWW
WW
WWW
LWW
WW
WWW
LWW
LWLWW
W
WWW
WL
WWLW
WWW
LWW
WLW
WLLL
WLW
WW
WL
WW
WWLW
WLWW
WW
WW
LLW
WWW
WL
WW
WLW
WL
W
WWW
WWW
LWL
WWW
LWW
WWW
WWW
LW
WW
WWW
WW
W
WW
LWL
LWW
WLW
WW
WW
LLLL
WWLW
W
LWL
WW
WWW
WWWWW
LWL
LLW
LL
LWW
LW
WW
WW
WWWW
WWLW
WL
W
WW
WW
WLW
WWW
WWW
W
LW
WW
WW
LW
LW
LLW
LWW
WW
WW
WWWW
WW
W
WW
WW
WWW
WLW
WL
WWW
W
WWWL
LLLWW
LWW
WWW
WW
no PATTERNS
LWWW
WWW
WW
WW
WLW
WWW
LW
W
WWW
WL
WWW
LLW
LWLW
WW
W
WL
WW
LWWL
WW
WW
LLW
WW
LWW
WL
WLW
WWLW
LLLL
WW
WWW
WL
WW
W
WW
WW
WW
LW
WWW
WWW
WW
WW
WW
WLW

At the end the same 'after a single L bet toward W' strategy will get a +11 unit result before vig.

For that matter a W-W clustering scheme applied at this last sample will get a +15 unit final result (before vig) that at the previous shoes sample won't provide any positive amount.
That's why is so important to patiently waiting for things more likely stopping after a given 'unlikely' pattern had shown up than hoping that a given pattern will prolong forever and ever.
Of course many intermediate situations may endorse a 'stop the betting up to a loss' strategy, anyway it's not the main tool real bac pros will look at.
And numbers displayed are the answer. 


as.

Even if I like Walt Whitman US writer, I don't want to be contradictory about what I've written so far: that is simple W or L situations happening when a strict mechanical strategy works out cannot provide a valuable edge.
Yet let's see what are the results of a 'bet W after a single L' plan did on the previous samples: 

It's a +23 unit (before vig) result.

Summary.

W= +1, L= -3, triggers are any L not following another L and of course we'll bet one time toward a W.

I'm not stating this is an 'optimal' strategy to look for, I'm just saying that this is one of the wonderful opportunities to get the best of it at baccarat. Of course getting the least impact of variance.
Why should that be true?

Again the answer relies upon the average card distribution happening at every shoe dealt. making 'unrandom' certain pattern successions showing up at some points of interest, that is such spots are fully beatable.
Add this to the fact that the vast majority of shoes dealt are affected by an intrinsic defect of randomness: yesterday, now and in the future.

oOoOo

Playing baccarat is just as wagering into a 'biased' coin flip finite succession, where there two kind of 'biases':

- a general steady math bias (general asymmetrical probability favoring B);

- a more important actual 'bias' coming out from the actual card (unrandom) distribution, privileging some patterns than others.

Needless to say that such features cannot mock an independent 'coin flipping' game unless dictated by coincidence.
Thus, we know without any shadow of doubt that the coin will be heavier or lighter in some points of the shoe, naturally knowing that at the most situations our coin will be 'equally' balanced considering all opposite forces, meaning it will be unbeatable.

So forget to try to guess every hand dealt as most hands fall into the 'pure coin flip' unbeatable category. 
Whenever you'd think a given line (and there are innumerable r.w.'s to look for) seems to reach a valuable deviated ratio (especially after many no valuable patterns had shown up), bet this line by a 0.75% probability in order to prolong it just by one step.
But the better feature to look for is the 'complexity' of such patterns, so the more complex a searched pattern is, better are the probabilities to catch a winning hand along the way.
It's now that the 'clustered' factor will take its primary role over the possible outcomes.

as.

Take the #7 point depicted above applied to a 180 shoes sample derived from REAL live tables.
That is we'll put in action a RW considering streaks of consecutive values after an apparition trigger had come out.
As always when taking into account a 0.75% probability and simplifynig things, W +1 and L= -3.
Instead of looking at final W/L ratios, let's focus about WL dispersion values.

LWLW
LWW
LWW
WL
LLW
WWL
WW
LW
WWWW
WW
LW
LWW
LWW
WWW
WW
WW
WW
WLWW
LLW
WWW
LWLL
W
W
WW
W
LWWW
LWW
WWW
LWW
WWW
WL
WW
LLLW
LWL
LW
LWWW
WW
LW
LWW
WW
LWW
WW
WWL
WLWL
WL
WWW
LW
WW
WW
WW
WW
WL
WW
WWLLW
LWW
WW
LLLW
WWW
WWW
WW
WWW
WWW
WWW
WWWL
WLWL
WWL
W
WWW
WW
WWW
WLW
WWWW
WLW
WL
LLLW
WW
WW
LWLW
LLL
WW
WW
WW
WW
WWLW
WL
WLW
WW
WW
LWL
WLW
WL
W
WW
WW
WWW
WW
WWW
WW
WWW
L
WWW
WW
WW
LWW
WLWW
LWW
LL
LWLW
WLW
WLW
LWW
WLW
WW
LWW
WW
WW
WWW
W
WLW
WLW
WW
WW
WLW
WWW
WW
WW
WLLL
W
LLW
LW
WLW
LWW
LLL
WWWW
LWW
WLW
WLW
LWWW
WWW
WW
WW
WW
L
LW
WWWWW
LWW
WWW
WWL
WW
WWW
WW
LW
LWWL
WW
LWWW
WLW
LWWW
WLW
LW
LWLW
WW
WW
WW
WWL
W
WLLL
WWW
WWW
W
WLW
WWW

We see that per any shoe dealt, such streaks categories are not going to provide many WL ratios greater than a 4 hand pace succession.

Patience is the key factor to win at games

Even though we know to play with a verified EV+, long losing streaks could come out and we can't do anything other than accept them and move forward.
Think about those formidable black jack counters playing with a math edge yet possibly falling into the negative territory even for months.
The same about best poker players, where months could easily become one year or more.

Fortunately baccarat isn't a 'forced one-sided' game, meaning that we can take the side we wish anytime we wish even though we can't confide on a strict math edge.
Anyway at baccarat the main potential advantage to look for is to stay away when things seem to go toward the wrong way, so trying to get the best of it by distorting our plan consitutes a huge mistake.

Suppose to flip a biased coin where we know heads come out 51% of the times and tails the remaining 49% and no vig is applied.
Itlr we'll get a 2% pure edge, right?

Right of course, but the fkng variance could get us losers for a quite long time and no progressive plan could get the best of it  unless carefully determined by a very diluted multilayered betting scheme.

In this thread I've illiustrated 'general' strategies that can get you a statistical edge up to 6% or more, nonetheless the fkng losing 47% counterpart may easiily showing up for long or by clustered forms.

I recall my poker tournaments history where at key all-in spots my AK lost vs AQ just 8 times out of 9, despite being nearly 73/37 favorite to win every hand.
Surely it happens, so let's figure out about what lower favourable probabilities could do.   

Notice the 73% winning probability being so close to the 75% probability I've stressed about on my pages.

The difference is that at baccarat we could exploit and manage '0.25% losing situations'  way better than at poker and reasons are widely described on my pages.

So at baccarat the 'patience' factor should be confined just to one or two shoes played, differently than poker where every hand is completely independent from the previous one.

as. 

That's what could happen when you roam with a kayak in the wonderful Silver Springs Park (FL) waters...

@klw: you got a personal PM about your question 

as.

Hi KFB and klw!
Thanks again, I truly appreciate your interest!

@KFB

I've made a mistake, odds are 10.62:1 and not 11.62:1. Sorry.

"2) When in doubt or unless your strategy dictates to bet B to form or limit some patterns, do not bet a dime at Banker side as you'll be 10.62:1 underdog to cross through an asymmetrical hand favoring B).

Once I've heard a high end casino floorman whispering "as long as most bac players like to bet Banker, we'll eventually make more money than if they would prefer to bet Player".
Is this a mathematical heresy or it was just an ignorant and/or misleading thought?

Actually he was completely right and we'll see why.

1- At commission games, yes, the ROI difference between always wagering Banker vs always wagering Player is 0.18%, so itlr B is less worse than P. Not a awesome percentage to look for by any means. It would be at other constant betting games as black jack, for example, where every tiny percentage favoring players adds up.

2- Some no commission games make Player side the best bet to make: think about tables (Lucky 6) where any winning Banker hand by a 6 point is payed 1:2. At those tables, the HE is 1.46% at B side and remains 1.24% at P side.
The difference is 0.22%, now less worse at P side.   

3- Besides the "Lucky 6" tables, other no commission games make B side less disadvantaged than common commission games, think about EZ baccarat tables (HE at B bets= 1.01%) or rare tables (Holland and Spain) where B winning hands by a 5 point are payed 1:2 (HE at B bets= 0.93%).

4- Overall no commission tables where B bets are less burdened than common commission tables do entice the side bets wagering, that is those side bets 'covering' the situations where B will be payed 1:2 (F-7, for example). And we know how huge is the HE on those side bets.

But the decisive factor why, generally speaking, B bettors are more likely to lose more money than P bettors (or to win less) is because B aficionados think that the Banker advantage will be steadily distributed along each hand dealt whereas such advantage is just concentrated on very few hands. 

In fact, per every hand wagered math probability to get an asymmetrical hand favoring Banker is 8.6/91.4, a 10.62:1 ratio.
To put things on another perspective, on average it's just that whenever we'll bet Banker we'll be economically 'wrong' more than 10 times (as we'll be payed 0.95 to 1) and astoundingly right just less than one time (now getting an average potential 57.93% winning probability to rely upon).   

Think that at baccarat a huge percentage of winning hands are made of naturals and standing 7s and 6s, and those are pure symmetrical probabilities that must come out here and there, but the payment is strongly different being 1:1 at P side and 0.95:1 at B side.
It's true that at some no commission games such symmetrical possibilities are payed 1:1 no matter the side wagered, yet many strategies cannot allow to get a Banker key spot to win other than by at least a 0.95% ROI. Meaning that in the vast majority of the times we shouldn't put ourselves in the position to 'ensure' our main B bet by wagering a side bet.
This is a strong long term EV- move.

Of course there are times to bet those side bets, but we ought to do that independently of our main bet, otherwise we'll lose money itlr.

as.

Ok, now it's time to move on other games. Baccarat lacks of suspence.

as. 

Imo the clustering effect is the best tool we should exploit to beat baccarat for the relative inherent difficulty to distribute moderate or long 'hopping' opposite event sequences.
Of course longer are such more likely 'clustered' events better are our probabilities to win and possibly to win huge.
Whenever a clustered situation of fair lenght happens, we might win one unit or two coincidentally, meaning we can take triggers at different values and not necessarily by a simple 'no hopping' distribution.

An example is when 3s are silent making a single-double betting plan particularly appealing and the same is true about doubles (now singles and 3s become good triggers). At both situations very rarely winning streaks stay at a 2 minimum clustered level for long.

Anyway even taking into account the general probability things will distribute on average, very often what seems to be 'unlikely' tend to come out strongly 'clustered', meaning there's no point to 'force' general probabilities to show up as the actual card distribution denies it.
In such occurences we have to choose whether to 'chase' an unlikely world or simply wait for better opportunities (aka next shoe/s).

Mathematicians correctly state that those natural 'flows' are just a by product of math values, stressing on the unimportance of choosing triggers that might erase or invert the HE. As they simply do not exist.
But they are right only when they consider a baccarat production as completely random, at the same time not considering a kind of 'conditional probability' coming out from an actual card distribution affecting different random walks applied to the same source of results.

That's why Alrelax advices (that in no way suggests a simple trend following strategy or complicated progression schemes) take an important value.
In poorer words no one 'mechanical' system would be superior than a careful observation made at very large real live shoe samples.

How to link a 'general probability with an actual probability' strategy

Keywords are to think the game as 'probability ranges' (probably it's the same concept of Al's sections and turning points') and as defects of randomness. 

There are infinite 416 card combinations (ok, but 128 cards have the same 0 value), yet every shoe dealt in the universe will fall into more likely ranges of probability.
Now either by a general probability point of view and, more importantly, by an actual point of view (randomness flaws), any sequence will show up by a slight different degree of apparition than what expected values dictate. In a way or another, of course.
We know that when we bet B side we must get at least a 51.3% winning probability to make profitable such bets and we need just a 50.1% winning probability to get an edge over the house while betting P side.

Of course nobody is going to give a fk about those values (just hoping to win no matter what), but they should as this is the only sure factor to know whether we're really playing a EV+ game or being just temporarily lucky (or naturally 'unlucky').
That's why baccarat is the second best game casinos make more money with.

Hence patterns distribution ranges matter, there's a general probability and an actual probability to assess both getting more likely situations to show up, especially when considered by a back-to-back probability.
That is we must discard many extremely deviated situations potentially or actually coming out, unless we have reasons to ride  them.

This is a very simplified list to look for:

1) Consider baccarat ranges just in terms of 1s, 2s and 3s. An exception is displayed later.

2) When in doubt or unless your strategy dictates to bet B to form or limit some patterns, do not bet a dime at Banker side as you'll be 11.62:1 underdog to cross through an asymmetrical hand favoring B).
It's true that it's more likely to encounter long B streaks than long P streaks but itlr such attempts are worthless.

3) Besides side bets, the only edge we can rely upon to beat baccarat is exploting the 'clustering effect'.
Math 'experts' will teach us that this effect will be proportionally distributed with opposite (losing) situations but they are deadly wrong.

4) The clustering effect must be exploited either by a simple 1-level back-to-back property or by concidentally betting an already moderate or long 'same events' sequence. Of course at both scenarios the main aim to look for is just one win (what baccarat pros do).

5) Many sections of the shoe are unplayable and this doesn't necessarily mean that after a given series of unplayable sections a kind of favourable opportunites will arise.
Do not rule out the possibility that cards are so whimsically arranged to get few clustered profitable events and the idea to bet on such 'unwanted' scenarios is out of order.

6) Always take into account the doubles or 3s pace of apparition considered at various portions of the shoe and in relationship of their average final value.
Of course singles make a huge impact of such doubles or 3s gaps.

7) A corollary of the #1 point, it's a kind of another invincible plan:
Set up two different random walks (SRW) applied at streaks clusters of precise lenght.
SRW #1 bets toward doubles and triples (exact triples) clustered at a given level, so any 3+ streak is a sort of boundary.
SRW #2 bets toward exact triples and 4 streaks clustered at a given level getting as 'boundary' any 4+ streak.

Those are 0.75% probability random walks that starting at a 0 point must move toward the left (lòsing side) or toward the right (winning side) by a 3:1 W/L pace. But there's a general propensity regarding which clusters of such events we'll take as triggers. (Differently to brownian motion, for example).
And differently to the common random walk concept applied to an independent (and symmetrical) source of results, odds that any class of events will be affected by relative low dispersion values at losing side are astoundngly high and, for that matter, itlr such RWs will infinitely, albeit slowly, direct toward the positive right side.
And I'm not implying to always consider more valuable P streaks lenght to stop than B streaks lenght to stop for obvious fkng reasons.

Providing the valuable triggers to look for, such 'streaks' plans cannot be wrong by any means, mainly as it works wonderfully even at pc simulated shoes (yes, yes, yes!).
Sure, a substantial amount of patience is needed but I do not recall any long term winning player lacking this important element to succeed at games.

as.

Thank You Al!
Your post Is really good, tomorrow I'll try to make additional comments.
(Today we are too busy to destroy a fkng casino, no jokes)

as.

Nice post!

I have an idea in order to lower the unit disboursement.
Instead of using full amounts, why not adding a proportional percentage your Plan dictates?
So instead of 3, let's add a 30%,  20% for 2 and so on.
Now your bets are:
$100, $130, $120 and $160.

Imo surviving takes a primary role over the urge of winning fast.


as.

Hi klw and thanks for your interest!

Itlr and just considering infinite and simple 4-hand sequences, the probability to get a final neutral result (W=2 and L=2) will be always underdog (unfavorite) vs a kind of light (+2) or strong (+4) unbalancement happening at either W or L side.
Of course we could easily face long series of WLLLWLLLWLLL or LWWWLWWWLWWW opposite situations making difficult to grasp the final outcome knowing the quality nature of the very first result, yet a kind of 'clustering effect' must work at baccarat as key cards (and/or 'miracle' cards) cannot be distributed proportionally along any shoe dealt.
Otherwise this simple plan could beat other games as roulette, for example, where each spin is completely independent and randomly ruled.

Raising the winning probability

Say we make an infinite series of 4-bets getting an overall winning probability of 0.75% by a 1-2 unit limited progression.
Now 'breaking even' 4-hand sequences are (vig ignored for simplicity):

WWWL (+1, +1, +1, -3) = 0

WWLW  (+1, +1, -3, +1) = 0

WLWW (+1, -3, +1, +1) = 0

LWWW  (-3, +1, +1, +1) = 0.

Any other possible W/L four-hand combination makes very different final scenarios ranging from the best possible scenario being WWWW (+4) to the very 'unfortunate' situation that is a LLLL spot (-12).
Naturally the LLLL spot is three times less probable than the WWWW pattern.

In some way we know that a mechanical 4-hand W/L pace will get more probable a W clustered situation to show up at some point and of course we can't get a W cluster if the very first hand is a L.
At the same time, without considering the actual 4-hand pace, we might fall into the mistake to take any single W or, worse, any single L as a valuable trigger to consider our options. If this should be the case, the game wouldn't exist at all.

In fact any shoe we're dealing with is made by 'expected W/L ratios' and 'actual W/L ratios', and we know that utilizing a 0.75% winning probability the general ratio to look for is 3:1.
Naturally the very first hand of any new 4-hand pattern coming out will make more or less probable to get a final neutral, winning or losing sequence more likely prolonging or stopping at any level.

It's like we are fairly advantaged to guess which 'clustered form' will more likely take place along the way.

In a couple of days we'll see why.

as.

Winning and losing flows

Winning and losing successions move around precise general probabilities, that is anytime we're joining a table the probability to be 'right' or 'wrong' or 'no right no wrong' aren't placed symmetrically at each of those three scenarios.
In some sense we are destined to either win or lose, making the 'breaking even' the least scenario to happen.
We may get a better idea about that by considering a simple four-hand betting strategy that of course could be prolonged infinitely.

Within a four-hand betting, 5 patterns will get more Ws than Ls and they are (vig is ingnored for simplicity):

WWWW (+4)
WWWL (+2)
WWLW  (+2)
WLWW (+2)
LWWW (+2)

It's a 5/16 (31.25%) probability

The specular 5 losing patterns are:

LLLLL (-4)
LLLW  (-2)
LLWL (-2)
LWLL (-2)
WLLL (-2)

Again a 31.25% probability.

The remaining 6 patterns out of the possible 16 patterns, those making a 'break even' scenery are (neutral patterns):

WWLL (0)
WLWL  (0)
WLLW  (0)
LWWL  (0)
LLWW  (0)
LWLW  (0)

Therefore per every 4-hand sequence we'd attack, general probability to be ahead or behind vs breaking even is 62.5%/37.5% (odds 1.66:1).

In some way, if we were able to 'guess' whether any new 4-hand sequence will fall into the final W or L category, we might get a substantial edge over the house, at the same time knowing that a kind of back-up plan could act along the way (neutral patterns).

Moreover, we are not forced to bet all the four hands forming the final pattern and/or to bet the very first hand of the new pattern.
For example, if the first hand of the new pattern is L, we know there's only one possibility out of 8 to get a final winning pattern and 'backup' neutral patterns will be placed by a 3:5 ratio.
And the same features happen at patterns starting with a W when considering final losing patterns.

Anyway the fact that we may 'artificially' making neutral or even losing 4-hand sequences into winning ones (and vice versa) shouldn't shift the natural flow of the outcomes made of winning, losing or neutral occurences.

So far I've considered perfect 50/50 propositions, thus we may apply the same concept to a greater probability and you know what I'm referring to.

oOoOo

It's the hands not played that make you a long term winner

Even the few very talented bac players liking to bet a huge amount of hands know that when the 'environment' changes they must put a stop at their betting.
After all, less bets made=less mistakes made and no progressive plan could erase or mitigate the general EV-, especially when things seem not going in their favor.

Back again to the 4-hand sequence results.

An 8-deck shoe provides around a 16 or 17 four-hand sequences each providing final winning, losing or neutral results no matter what's the strategy employed.
Obviously a steady betting Banker strategy will get more W patterns than a steady betting Player strategy could get, anyway the ROI difference is 0.18%.
Now let's take into account how the W, L or N sequences will be distributed per every shoe, not giving a fk whether a winning or losing sequence got a relatively rare +4 or -4 units win. I mean just the quality matters.

So a first degree of 'guessing' is made upon the probability to get W, L or N events coming out clustered or not and at which degree.  So being considered by a real final quality.
Example: W, L, L, N, L, N, W, N, L, L, N, N, W, L, N, W, L.

Putting things into a W/L  'gap' concept and considering N as 'not existent' we get: 1, 3, 1, 2, 1, 1, 1.

Then we could make a 'more educated' guess by knowing the first hand of the new 4-hand pattern.
If it's a L, odds that the final pattern will be a W are 1:8. And N happenings are 3:5 underdog. 
The same about a final L pattern starting with a W.

A 0.75% probability will move around the same concept but by different tools we'll see next week.

as.