Roulette Martingale alternative (Carsch)

Below is a graph showing my strategy being flat bet until spin 190 then I switched to the Marty to spin 228 with a win of 210.00 (42 units). The other thumbnail shows the patterns I was up against. This was a tough one, usually they are a little easier.

 

Wrong graph. This one is correct:

 

Again exactly my thoughts as it really is the only thing that superficially doesn't appear to violate the math and could be an angle for systems player.

I would suggest counting the occurances of each streak in the W/L registry and plotting the distribution (use negative counts for losses and +ve for wins), then perhaps looking at the kurtosis or simply compute it with excel. A more narrow curve than a normal distribution would be what one is looking for as you want the mass closer to the mean and less in the tails.

Can it be done? Excellent question.

I haven't seen @David Gregory bet selection but you have confirmed he is getting excellent W/L registry with a reduction in loss streaks of 6 or higher.

My thoughts are that certain patterns have a shorter waiting time than others. Conways leading numbers tells us this, as do a number of other methods including Markov chains.

If you target patterns with shorter wait times and restart the matching process you can also reduce the effective occurrence of those patterns, even though from a stream perspective all patterns occur with the same probability.

For example take RRR vs BRR the RRR pattern has a longer waiting time and the longest of all patterns same as BBB. BRR trounces RRR (7/8 vs 1/8).

Now I realise that is comparing two patterns but it simply and example that some patterns form more readily than others. This does not mean they occur more often, all patterns have the same probability to occur when we examine the stream, their distribution counts are the same.

So we break the stream though bet selection, and therfore increase the occurance of patterns that we bet on or reduce the exposure to nemises patterns.

Returning to matching the pattern RRR once you do get a RRR (or BBB) the next result from a stream perspective can be another full match due to the self overlap in the pattern. So whilst RRR and BBB have a longer waiting time we see that overlap in the pattern itself makes up for it, so occurances of the RRR and BBB patterns HAVE TO BE CLUSTERED extending to 4iar, 5iar etc for the expected probability distributions to be realised.

So we have some options, avoid overlap and start matching over from scratch or bet on the continuation of RRR or BBB.

Starting the matching from scratch after a pattern match increases the distribution count of the patterns that you bet on given their waiting time is less. The BRR pattern will match 7 times more readily than BBB by starting matching again. Again nothing has changed in what the wheel produces, the stream of outcomes is still there but bracketing sequences through bet selection rules creates a different distribution, kind of like a filter. Sputnik uses what he calls gap theory (leaving a break) to isolate and alter the distribution counts of certain patterns over others and perhaps in doing so alters the dispersion in the W/L registry.

I am not saying that pattern matching improves our ability to get more wins, but what I am saying that the win loss streaks or the gaps between wins may be tamed but still within what probability dictates.

 

The Martingale is a horrible progression, generally reserved for new and naïve gamblers.
1. The initial bet must be silly small, which renders puny wins.
2. The bankroll risk to win ratio is absurd.

For the raw gambler, the Reverse Labby is a much more appealing progression. Especially when you have a gambling pool of friends out for a night at the casino.

1. The initial bets are larger than the puny Martingale bets.
2. Small losses are more common, but catastrophic losses are avoided.
3. The probability of enormous satisfying wins are more likely than with other progressions. And after all, isn't that what gambler's are after!!!
4. The player doesn't have to win at a rate that vastly exceeds expectation in order to win big!




Obviously the best progressions are betting a percentage of the bankroll. Something like edge/expectation x confidence level. However, this ideal progression requires that the player have the advantage over the casino.

 

Nice post and I agree, I'm definitely a fan of positive progressions.

Reverse Labouchere and also Contra D’Alembert are decent positive progressions.

From a study you can see they do reasonably well with nice exit points but as a result they have more volatililiy vs their negative counterparts.

 

It's insane and anybody who plays this seriously is insane. It is stupidity on display to risk so many units to make 1 puny little unit. Yet it is true that the Martingale is an absolute winner 100% of the time if you have a bottomless bankroll and the casino has no betting limits. You might be Bill Gates but you are never going to find a casino that has no betting limits.

 

Classic fool.
Admits martybation is bad and unsustainable but continues to “kitchen test and post” only a fool waste it’s time one a losing endeavour.
Numerous “respected” posters have offered better alternatives and your post actually shows how much you ACTUALLY RESPECT them.Zero respect shown by you , keep milking the teat. Sounds like you just jerking everyone off.
You got something try using some of “ your respected ( rlmao ) posters” advice . Stay at the table, keep batting off and suck that smoke out of your own arse.

 

I see what you are saying ..

I already asked you once, namely about Conway numbers with a reply received that applies only to EC.
But the principle must fundamentally apply, which you implicitly confirmed with the mention of Markov chains & trounces ..

So I am gonna ask again .. how does this apply to ds, given that mostly 2, sometimes 3ds are covered with various positions (3DS when exposition bordering 0, 2DS up to -10, 2Q & 3Q thereafter).

.

What patterns are we looking at? I guess that would be sounces, sixies?
How does the fundamental principle (of waiting time) apply to the pattern when applied to ds?
How do we further maximize hitrate by including all ds1-11 in the pattern, still withing 6-spin ds cycle?

 

Also, I've missed Sputnik's gap theory, but since many of you have mentioned it, it must fairly recent.
Can someone print the link.

 

Yes a Markov chain is the long way to work out the probabilities of success (absorbing states) regardless of the transition probabilities (bet types and their probabilities).

Draw a Markov state diagram and derive the transition matrix and then solve all the equations to compute the absorption probabilities and the mean time to absorption. It is a LOT more effort than the Conway numbers and this is left as an excercise for the reader.

Here are some videos and papers that might help to understand the process:

https://cs.nyu.edu/mishra/COURSES/09.HPGP/scribe2​

 

I don't have a link can't find it maybe @Sputnik will chime in.

 

Actually I address the issue of the zeros sleeping and the zeros coming awake and even swarming. You are right that it can take thousands of results for these extreme phases to go back to looking like what normal is. Psychologists have done peer reviewed experimentation on people and their impressions of what they think random is and what is not random. These extremes are considered not random by most people and they proved it. Yet, people have no difficulty seeing the perfect sequence of results that take out anyone playing a Martingale Progression. There is no difference between a perfect sequence and a swarm of zeros. Now what you are saying about the zeros might go far beyond the possibility of happing on a real live wheel because it is an RNG. But swarms of zeros is a natural occurring part of randomness.

 

This explains the waiting time of different patterns when looked at from a block perspective vs a continuous stream which allows overlaps. If our bet selection splits the stream then occurances of some patterns (nemises patterns) that you are exposed to are going to be less as a result:

 

You could use this valuation method to create your own way of seeing randomness through connected past results. It's for sure that you have an expectation to see more HH than HT. That's because the inversions of HT don't count and the inversions of HH do count. Yes, I think bean counters have been gambling for centuries based on observation like the video above. You see a lot of people using matrix styled charts for instance. But I know this for a fact. I can see far more with my charting methods because of visual dexterity. I get to use overlaps. I also get to use coincidental overlaps from 6 different streams of data simultaneously. You can't see this in lineal data charts like "HHTTHTHHTTTHTHTHTHHTTHT."

Now look at hThhTTThThThThTTThhhTThhTTT. You can begin to see by visual dexterity.

 

Actually, the simplified bet selection I simulated didn't show any significant difference from betting on red only in terms of the number of marty busts. The BS was just follow the last outcome until two losses in a row, then change to follow the chop, and after the first loss in this mode change back to follow the last. If it's at all possible to reduce the dispersion I believe it's going to take a fairly complex selection process; a simple BS isn't going to cut it.

I'm going to investigate the Conway leading numbers. I also have some ideas which involve the Hamming distance between sequences.

 

True, a simple BS isn't going to cut it. True, it is going to take a fairly complex selection. It's there, keep looking, you'll find it eventually.

 

Oh, if you only knew what's going on behind the scene.

 

XOR will provide the hamming distances, just count the bits of the result A XOR B. Here is a table of hamming distances for tuples of length 3:



Note that each column sums to 12.

 

This also explains quite nicely why HT beats HH

 

Here is the gap methodology https://www.gamblingforums.com/threads/who-likes-to-win.23880/page-8