While it is a topic that deserves too much attention and innovation, I see absolutely no work on it.
Why can't we create an MM that fights with randomness for long sustaining through worst of it and ultimately winning when variance cools down to the level of the house edge alone or closer?If you analyse any bet, you will find that after a reasonable span of time, a bet gets smooth enough to be beaten. All the extremes of variance are temporary and gradually every bet confirms to the mathematical expectancy attached to it, including the house edge. We see all the progressions meant for few spins or a small session. Have anybody ever thought of a long term progression? By long term, I mean like a progression for an EC for 500 spins or 1000 spins for a dozen or may be until we get a net win.

                         Long term doesn't only mean a single long session of thousands of spins but could be for 10 or 20 sessions or more taken together as long run.

Hello dear Sumit,

When you talk about progression for long sessions, I can relate it to progression in the style of the "boom" technique for a single number; rising the base unit back and forth, consciously.

ie. after two missed tries player goes back to minimal unit while keeping track of current unit size. Later on right after a successful attack the player resumes further attacks back at the same tracked unit size level.

When dealing with measuring after those 200-some "short" sessions, -likewise- a boom-style money management for several numbers at a time (working cooperatively) can be worth a try.

With the added benefit of knowing there's a clear stop-loss “check-point”, with no chasing.

P.S. Another LONG progression coming to my mind is the progression in risk comprised by six (6) 36-unit parachutes, with strategy:

- Start one after the other at the even-chance level.

- On a hit you use the parachute covering the least (and hence paying more).

- On a loss, you use the parachute(s) covering the most (to try to get the much-needed hits to sustain the game in order to stay the longest).

This technique was widely studied at some spanish-speaking study groups. Over the English-speaking forums it was member GGasoft who advocated it.

  I believe that we can discard martingale, Labouchere and Fibonacci type super aggressive progressions that none can afford to use being realistic of the obvious failures that we can face any moment. They are of course not meant for playing for the long run.
                      Oscar's Grind or raising +1 on a win and betting the same on losses is one of the long run progressions suggested. It is still better than many other conventional progressions but still, any simulation of long run proves it as a failure because it fails to earn unless we get a good streak of wins after losses. A normal flow of wins and losses could create many layers that it fails to break.
                         

Hi Albalaha,
in your blog, you say...
===========================
Every bet selection reaches at its break even point minus house edge in long run


I do have my own set of trackers and bots. I have extensively tested all bet selections over millions of spins. In long run, say in 3700 spins (100 cycles of all numbers) or more, every bet selection gets to its break even point minus house edge. This applies very accurately to outside bets.
               This fact ensures profit for casinos unless a player get lucky streaks and he could capitalize that too.

So 3700 spins is the long run?

What happens when you have done 100 of these long runs and you now have 370000 spins?

Doesn't that make your 3700 spins look like a short run, or hit and run?

I'm confused..

Hi Albalaha,
you say,
------------------------------------
------------------------------------------
Virtual limit of persistent bad doing bets, in my observations(in european roulette):

straight up:only 4 hits in 1000 spins
split: only 15 hits in 1000 spins
street: only 40 hits in 1000 spins
corner: 60 hits in 1000 spins
line bet: 110 hits in 1000 spins
column/dozen: 250 hits in 1000 spins
EC bet: 405 hits in 1000 spins
double dozens/columns: 575 hits in 1000 spins

unquote.
----------------------

Thus, if we could have a progression, that beat, say,
EC bet: 405 hits in 1000 spins,

win=+405u,
lose=595u,
losses=-190...

then we need to have a progression,
that recoup -190losses,
by 405hits.
=====================================
or win, at least 1u, with only 405hits/1000spins!!!

that's a grail!!!

Greenguy,
3700, is  assumed,
that all bet may around BREAKEVEN.

If you bet flat, u may win, or lose,
accordingly to your ladyluck.

Say, if we see, that a single number sleeps for 500spins, then you may bet that single, continuously, to 3700th spin.
and hopefully, as, or if it breakeven earlier, then you win, even before the 3700th spin.

column/dozen: 250 hits in 1000 spins

then, if you can,

250x2=+500u
losses=1000-250=-750.
losses=750-500.=-250

if you can figure out, how to recoup,
-250losses, by 250hits,


or win, with only 250hits/1000spins...
then you got a way to win...constantly.

Quote from: greenguy on March 17, 2016, 07:14:13 AM
So 3700 spins is the long run?

What happens when you have done 100 of these long runs and you now have 370000 spins?

Doesn't that make your 3700 spins look like a short run, or hit and run?

I'm confused..

3700 spins could be considered long run for an EC bet but I have defined it far better in this topic:
http://albalaha.lefora.com/topic/19400636/Long-run-in-gambling#.Vup3uNJ97IU

Quotethen we need to have a progression,
that recoup -195losses,
by 405hits.
=====================================
or win, at least 1u, with only 405hits/1000spins!!!

that's a grail!!!

Absolutely.
            You won't find any such progression that can beat 405/1000 wins of an EC in all patterns they can appear.

Quote from: Albalaha on September 27, 2014, 08:54:53 AM
                     While it is a topic that deserves too much attention and innovation, I see absolutely no work on it.
Why can't we create an MM that fights with randomness for long sustaining through worst of it and ultimately winning when variance cools down to the level of the house edge alone or closer?If you analyse any bet, you will find that after a reasonable span of time, a bet gets smooth enough to be beaten. All the extremes of variance are temporary and gradually every bet confirms to the mathematical expectancy attached to it, including the house edge. We see all the progressions meant for few spins or a small session. Have anybody ever thought of a long term progression? By long term, I mean like a progression for an EC for 500 spins or 1000 spins for a dozen or may be until we get a net win.

                         Long term doesn't only mean a single long session of thousands of spins but could be for 10 or 20 sessions or more taken together as long run.

It seems to me that the problem lies in the impossible task of measuring variance.  We can only guess at it and perhaps calculate the outer limits of SD.

Labby, Fib, Marty, etc. out?  Yes.  D'Alembert?  Not so fast.  Sorry to be a one horse show but this endeavor always brings me back to S. Bailey contemplating the flaw of D'Alembert and refining it to adjust to variance with various mechanical or semi-mechanical progressions.  A favorite of mine (and his) is -X, +X; Flat.  Whereby we go up 1 on a loss and if win stay at that level until a loss then go up 1 again.  Of course adjustments can be made to taste and not always have the same unit size for both sides.

Variance comes in as a matter of perception or SD if one could calculate risk of ruin for a win/loss "likely" sequence.  So how does one accomplish this to gain a net profit?

It's mostly guessing of course but there are limitations that one can count on.  Baccarat shoes:  I know that in the 5000 + shoes I've logged the lowest I've ever gone is a 32% hit rate within about 60 or 70 "spins" and that is rare.  The overwhelming majority of closed spin sessions are going to be between 47% and 53% I'd guess.

So when you begin to go south you got to increase but only to a point.  But when you go too far south you got to decrease.  At some point when a north trend begins you must increase.  Maybe incorporating rules such as Seth uses with "Turnaround" or his latest stuff to help soften the curve.  Half back, Cut back, dump and pump . . . ??  Big bankroll required.

This whole concept reminds me of "Price Cost Averaging" (not Dollar Cost Averaging) whereby algorithm used to buy as stocks drop and sell as they rise.  Works wonderfully for cyclical movements such as we are dealing with.

J

D'alembert or its other variants do work upto a particular stage and is not meant to work well in long run thereby. When there is a very bad stretch going like 25 wins in 100 spins or anything alike, it goes so bad that later it never gets to recover, in any manner. After extreme bad sessions, very good ones can not be expected always and in expectation of getting "corrective" wins, we get "not so good" time and "below average" times that makes it failure.

Quote from: Albalaha on March 17, 2016, 05:30:59 PM
D'alembert or its other variants do work upto a particular stage and is not meant to work well in long run thereby. When there is a very bad stretch going like 25 wins in 100 spins or anything alike, it goes so bad that later it never gets to recover, in any manner. After extreme bad sessions, very good ones can not be expected always and in expectation of getting "corrective" wins, we get "not so good" time and "below average" times that makes it failure.

I guess you weren't paying attention.  The D'Alembert like ALL progressions are failures without being able to take a measurement of losing trends and make adjustments.  My example above is just to demonstrate such an adjustment.  When one uses the -X, +X; FLAT or a variant offset it ceases to be a D'Alembert.

You can have the last word.  I just wanted to add my 2 cents about the progressions.

J

Going by Albalaha's finding,
then,
in an oversimplified example,

if we see, say,
an EC sleep for 10spins,
then there will at least 405/990spins.

thus a 'worst' table below.

sleep for...

0 spin=405hits/1000spins.[40.5%]
10spins=405hits/990spins.[40.9%]
20spins=405hits/980spins.[41%]
30spins=405hits/970spins.[41.7%]

to...

190spins=405/810spins=breakeven?!

=====================