Our members are dedicated to PASSION and PURPOSE without drama!

Use Math to beat Roulette/Baccarat

Started by Nickmsi, May 30, 2016, 04:43:02 PM

Previous topic - Next topic

0 Members and 14 Guests are viewing this topic.

Big EZ

24 more shoes to add to the testing ....


x
w
w
w
w
w
------ +4

x
w
w
w
w
------ +3

x
x
w
x
w
------ -1


w
w
x
w
x
x
w
------ +1

w
w
w
w
w
x
w
------ +5


w
w
w
x
w
------ +3

w
w
w
x
w
------ +3

w
w
w
x
x
x
w
------ +1


w
w
w
w
w
x
w
------ +5


w
w
w
x
w
------ +3

w
x
w
w
------ +2

w
w
x
w
w
x
w
------ +3

w
w
w
x
w
w
w
------ +5

x
x
w
w
w
w
------ +2

x
w
w
w
w
w
------ +4                               

x
w
w
w
x
w
w
------ +3

x
x
w
x
w
------ -1

x
w
x
x
w
w
------ +0

x
w
w
x
w
------ +1

w
w
w
w
x
w
------ +4

w
x
w
w
------ +2

w
w
x
w
w
------ +3

x
w
w
w
x
w
------ +2

w
w
x
w
w
------ +3                             
Quitting while your ahead is not the same as quitting.

MarkTeruya

Hi Nick

Thanks for the updated spreadsheet. 

I'll test further maybe later in the week.  I kinda of have an inkling with this as I've tested and played Eirescots Birthday Paradox Matching pairs at length also betting for UnBalanced V's Equilibrium over a given hand sample (6, 8 & 12).  The former can be so frustrating when it doesn't work, even though it has (suppose too) maths on its side, losing 4 bets on the bounce in a given 8 hand sequence, then losing 3 winning 1 and losing another 4, there is no way back unless you're incorporating something else.

Those methods which including this one are all non-random approaches, but present the same issue.  "Single win v's many losses" risk, I suppose the ultimate would be a betting method that is statistically expected to win more often than it loses but only involves a single bet :-)   

Repeating two's which it appears this option doesn't fair so well with, can be frequent within  the game of Baccarat.  I'll report back.

Do appreciate it's a work in progress and worth nutting out to see where it leads... 

For now....

soxfan

I still think that betting flat ain't gonna feed the buildog, but I'll eyeball this style and see if it is better than mine at pickin off the backs to backs win, hey hey.

plolp

Quote from: Nickmsi on June 04, 2016, 06:13:22 PM


Yes those 6 pattern out of 512 do show the losses of -3 or -4.

What do the other 506 show overall.  Do they compensate for these 6 patterns?

Let's review the statistical data and determine where to go if they are positive or if they are negative.

As I have said from the beginning, this is a FRAMEWORK that we can use in playing.  This is NOT the final version.  There is much more we can do to make it better.

In the history of the game, there has never been a NON RANDOM method that I know of.  Do you know of any?


"Do they compensate for these 6 patterns?"

Yes this is absolute equality.

You say it is non random ??    It's wrong .

   the theorem of van de waerden   proves that  all spins are random .

Rien de plus normal, tout est étrange .

Nickmsi

Yes Plop . . .You are correct, spins are random.

Bet Selection is not.

Cheers

Nick

MarkTeruya

Okay, I did some possible scenario testing;

First off, with all the losing columns and single unit winning columns the balance result was ZERO, as expected.

IMO is a waste of time testing data sources, for the simple reason the data is historic and is of no relevance to what you may encounter in live play, suffice to say guaranteed you won't encounter the same data you tested against at any gaming table, if offers nothing other than peace of mind to the beholder.   You are better off testing against "all possible scenarios"  you can achieve that by using truth tables.

For a given 9 hand sequence, 512 possibilities; 

8 columns produced 4 losing bets, a 1.56% chance of that happening
12 columns produced 3 losing bets, 2.34% of that happening
56 columns produced 2 losing bets, 10.94% of that happening
And there were 256 columns producing a 1 unit profit, 50% (less B tax)


In comparison, a 8 hand sequence betting that there will not be 4 of one-side and 4 of the other (equilibrium), carries a 72.7% statistical expectation to succeed (trust me when I say it simply doesn't pan out like this at the tables, it never does).   

For 256 possible options the chance of losing 4 bets worst case (as in BBBBPPPP or PPPPBBBB) for a 8 hand sequence is 2, so the % ratio is exactly the same as the VDW option. I think there is a marginal advantage as I'm only seeing 252 columns where no winning bet is achieved out of a possible 256 options (my figures could be wrong, too time consuming to double check), suffice to say not a recommended bet option regardless. 

I think The Birthday Paradox Pair method again was marginally superior given there are fewer losing columns, unable to confirm as I no longer have the data, plus you have various ambiguous modes, again risk making 4 bets to win 1 bet, too bad when you walk into back to back no-matching pair grids.

Not wishing to rain on anybody's parade, would like to post more promising news but "it is what it is", some days you'll win, other days you'll lose. 

You could restart after any winning bet, this changes everything both good and bad I expect..

sqzbox

Nice analysis. But I don't think anybody here is expecting that this approach will yield an actual edge against the probabilities in the game so no surprises with your results really.

What I think is being studied here are the LIMITS presented by this type of bet selection. "There will always be ..."  and "within such and such x will happen" and so on. This is the nature of the so-called non-random events. And, of course, how these limits can be utilised to advantage. Can a safe progression be worked out to match a certain property such that failure is unlikely - or even impossible? That is the real question here because, in my view anyway, a progression will be mandatory for long term profitability. So how can we make it safe?

Or perhaps can a combination of several bet selections from a palette of non-random methodologies be combined into a cohesive strategy together with a "safe"progression be profitable long term?

MarkTeruya

In my opinion the answer is probably not.  If it's impossible to handle Birthday Paradox Pairs, or Equilibrium v's UnBalanced by themselves, then it is extremely doubtful this will be any different. Introducing more bet options can work both for the player and against the player. 

One method loses, only to be followed by a second option losing straight afterward, more losses pilled on top of another, don't wish too be negative, but this is the stuff that will happen in live play.  IMO what needs to be achieved here and for similar betting methods is a reduction of the 4 to 1 ratio.  Risking four bets to win once, all these options present the exact same "worst case scenario, ditto any 4 column template with it's 99% mathematical expectation.  Either reduce the potential losses when things go bad or improve the win count when things are in sync.

What is required is a positive statistical option and when things don't work, it doesn't cost you 4 losing bets each time it fails  That being the case I could only suggest placing virtual losses in front whatever option you decided to go with.  Also you have to consider "what exactly is the pattern that this loses against".  in this case, repeating two's is one of it's nemesis, which IMO should remove it from  consideration regarding Baccarat.

Anyway sqzbox trust you are well, it's been  while since we communicated.

Nickmsi

Great discussion Mark(thanks for the statistics) and Sqzbox.

Sqzbox said:
"Nice analysis. But I don't think anybody here is expecting that this approach will yield an actual edge against the probabilities in the game so no surprises with your results really."

Mark said:
"What is required is a positive statistical option"

What if someone had a statistical option to yield an actual edge.

What criteria could we use to verify the claim?

If in profit after 100,000 spins? Probably not as any good 50/50 system could be ahead.

The Van Keelen test?

What would an edge look like?

What would a test look like?

What would the results look like?

Your thoughts would be greatly appreciated.

Cheers

Nick

from100

@Nick

One question:

In testing VdW on 8 th spin (or 6th or 7th)... have you waited 7 spins and then place a bet if there was opportunity? All other spins were just watching?

Thanks

MarkTeruya

Quote from: Nickmsi on June 05, 2016, 02:12:32 AM
Mark said:
"What is required is a positive statistical option"

What if someone had a statistical option to yield an actual edge.

What criteria could we use to verify the claim?

What I had in mind when I made that comment, are the 4 column template options, they "generally" present 12 ways to win and 4 ways to lose.  So statistically you have a 3 to 1 ratio in your favor, with the risk of losing 4 bets when things don't work.  Leaving aside the case of wins v's losses as everything balances out, as there is no possible mathematical escape from a 50-50 resolve.  Hence why VDW had a balance of zero when all potential possibilities were considered.

So the perceived edge / criteria would be the 3/1 even 2/1 statistical expectation of a win over a series of bets, as well as avoiding 4 losses in a row when things don't work, as it puts too much strain on the players money management.   You'll find this a mathematically impossibility.

The glaring issue with VDW is it only offers a statistical 50% expectation of success over 512 columns (forget the zero balance, as this applies to every bet option).   However if VDW offered for argument sakes, 510 ways to win and only 2 ways to lose, therefore you had a statistical expectation of 256/1, I would say hey, go for it, it's gonna be a cold day in hell before you got hit, leaving aside the obvious problem in terms of progression.

QuoteIn testing VdW on 8 th spin (or 6th or 7th)... have you waited 7 spins and then place a bet if there was opportunity? All other spins were just watching?
"Virtual losses" not applicable in this case, because the statistical expectation is only 50% over a series of 9 bets.  Therefore the player only has no perceived advantage after any prior losses or column failure.

Compared to betting UnBalanced over Equilibrium with a 8 hand sequence, '186 winning columns v's 70 losing columns', the expectation which is proven mathematically is 72.7%, ditto Birthday Paradox Pairs.

Trust that makes sense...


Bayes

Quote from: Nickmsi on June 05, 2016, 02:12:32 AM

What criteria could we use to verify the claim?

If in profit after 100,000 spins? Probably not as any good 50/50 system could be ahead.

The Van Keelen test?

What would an edge look like?

What would a test look like?

Hi Nick,

I assume you're talking about raw bet selection here and not any kind of MM/Progression. According to the "long run" a series of EC bets will only make a profit  in roughly 1 in 20 sessions of 5000 placed bets, so that would be a good starting point. If it's in profit after this you could get the z-score which is :

z = (w - n*p) / sqrt(n*p*(1-p))

If this is steadily increasing then there's a good chance you may have something. A score of +3 would only occur about 0.3% of the time, and anything higher even less, so for example if you have placed 5000 actual bets (not just spins) and have 2600 wins and 2400 losses, the score would be

z = (2600 - 5000 * 0.4865) / sqrt (5000 * 0.4865 * 0.5135) = 4.74 standard deviations above the mean, which would be pretty impressive.

pitrak

Hello,
i test the baccarat tracker with the zumma shoe I just started it and will post the result here
I finished the first shoe and used 4 diferent types of MM.Here are the resuklt of the first shoe


martingale: +14
flat betting= +10
MM: +13

zumma shoe 2 a very ugly shoe

martingale:+10  1time it went to -16
flat: +2
MM: +1

shoe 3

martingale: +8  higest bet was -4
flat: 0
mm:+4


shoe 4

martingale: +13 biggest bet -4
flat: +7
mm:+12


so after 4 shoes the profit is:

martingale: +45 units
flat:+19
MM (up and down management): +30


martingale most profitable but very risky drawdown within shoe
flat bet survives for now an ugly shoe
mm up and down  also survives

I keep testing



Nickmsi

Thanks Bayes . . .

That is exactly what I was looking for.  A definitive way to tell if you have an "Edge" or not.  Yes you are right in that I am considering only Bet Selection, Flat Betting, no MM at all.

On next update to my bot I will have Vic add the number of Wins, Losses and Placed Bets so it can calculate the z-score.

"If this is steadily increasing then there's a good chance you may have something"

I assume this means if you have a steadily increasing graph that would indicate we might be on to something.

Take a look at the attached 4 graphs.  They are for about 90,000 spins each.  Assuming these were for 90,000 placed bets, would this be the type of steadily increasing profits that you would associate with an "Edge"?

Thanks Patrick for your testing, results look promising so far.
Cheers
Nick

BetJack

Hello
I have been following with interest the topic
you will help me a lot if you share the following information

Quote from: MarkTeruya on June 04, 2016, 10:33:45 PM

For a given 9 hand sequence, 512 possibilities; 

8 columns produced 4 losing bets, a 1.56% chance of that happening
12 columns produced 3 losing bets, 2.34% of that happening
56 columns produced 2 losing bets, 10.94% of that happening
And there were 256 columns producing a 1 unit profit, 50% (less B tax)

You really will help me a lot
if you have all 512 combinations win / loss results

thanks in advance

with respect BETJACK