Testing Turbo's "1 unit per visit" + added simple moving average.

[mr.ore]

I was testing some Turbo's systems which can be found on his page, and one of his systems is "1 unit per visit system", which works like this:

You are betting a number which had least appearances during last spins with a progression 37x1, 37x2, 37x4, 37x8, 37x16.

Turbo's system caught my interest, because I had some nice sessions when playing in RX to have some fun, so I decided to code some into my testing framework. Of course it did not work so nice, because RNG I used for test is very good, ie. very nasty.

I tried to add SMA shifted little down do give progression a Little space, and switch to virtual play when bankroll is bellow it, and continue play for real when it is above. With that, I got some interesting results, but it fails often - it can prevent big loss after a winning streak, but also can prevent from winning when numbers are hitting less than expected and progression would work here.

Simple moving average (SMA) is an average of n last values, n is number which affects how quickly will it react to changes. If big, it will react slowly, if small, you can measure short time trends with it. I am trying to use SMA as a kind of stop loss.
Also if you have two SMA's and one is slow and second fast, then you can use it as an indicator to start playing, if the quick one crosses the slow one and is above it. The idea of SMA is taken from some forex tutorials, which I have been looking on lately. Actually, why bother with roulette? Forex seems much better, there are trends for sure, which are not just random fluctuations like in roulette. And instead of sports betting the broker does not want you to loss like bookie, because in the long term he has got better profit from successfull trader than from loser. But to learn how to "play forex" may take at least a year of virtual play in a demo account, if you want to be successfull. On the contrary - to be successful at roulette, it will take whole eternity

Well, I will post some results to this thread, if someone have any idea how to use moving averages in roulette, post it here. Actually - all what is needed is a system, which can enforce presence of trends, and bet unit is not increasing too fast, so that you can stop while in profit. I need a system, where always is a winning streak or losing streak, and there are no "flat" places, which are the enemy for SMA(many traders do not trade when the graph is flat, waiting for trend). Losing streak can be avoided with moving average, winning can be backed with stoploss. If we could for sure catch trends, money management over sessions can help to be in profit. It is only my opinion.

Explanation of the graph:
1. red is balance from real play.
2. green is SMA, copying virtual balance.
3. blue is SMA shifted down a little to give those peaks from progression little space. If virtual balance crosses this blue line down, we stop betting until it crosses it up.

I'm thinking if after wake up from virtual play we are still down, it will play more aggressively until back. I will try it later and post the results then.

[mr.ore]

To explain idea of SMA based stoploss, there is simple example, I adder virtual bankroll to the graph.

In the first image, SMA based stoploss helped to avoid some big loss, when balance dropped under average of bankroll in previous spins.

In the second image it prevented terrible drawdown after a winning streak.

[mr.ore]

Some more examples of what it can do, this time I tried to make SMA based stoplosss(blue) very slow and put it really down to avoid only the biggest drawdonws.

[mr.ore]

Using SMA badly can change a nice winning session to a losing one. The problem is, how to set it, what I do not know yet, where the optimum is, and if any exists.

It is noticeable that Turbo's one unit per visit can give nice results, like there, where we are in profit after 100000 spins, at least in virtual play which was not tweaked with SMA.

[mr.ore]

Another experiment, "surfing" on very very slow moving average with n = 37*12*4, which serves as a second stoploss. The fact it avoided drawdown is caused by not betting so much units, but house edge has not been beaten even slightly.

[mr.ore]

Another idea - what would happened if we did the same thing with a graph, we get by virtual play some parts of a game? We add another level to this, and add another SMA level to the graph we got by playing by SMA stoploss, and play only if also this SMA indicates we should play. It is recursvie idea, we can do it many times, and each time we might get some other trends.

This time it made profit over same spins which were used in previous example. My aim is to learn, how to use averages to make profit on trends where there are some. When I saw those 30000 spins where are many obvious trends, I realized that if it was given to a professional trader to make profit on, without telling him it is roulette graph, he should be able to do so, because all what is needed is to know when to start, and when stop.

And this is the result with same data:

[mr.ore]

And it failed badly in the end. Now I will try to add another SMA to avoid this drawdown on such a big scale as 100000 spins. Now I made my own challenge to beat as many spins which RNG will generate from this RNG state as possible. How far will I get?

[mr.ore]

Back to low level, I added another SMA to the cascade. This is example of how can be this idea used to push negative trend upwards. Even under such bad conditions, it was possible to get from violet graph to red graph. Now even less spins were played, obviously. Why to play if it is not doing well? As long as trends are present, SMA will help to filter them.

[VLSroulette]

mr.ore

Thanks for taking the time to post your filters.

Perhaps this type of filters is what strategic players refer to when claiming while a system can lose, an strategic player can "make it work for him" 

But it is done using the human brain computer...

Very interesting concept your SMA.

[mr.ore]

Test of 66k spins from playtech I found in this thread:

nolinks://vlsroulette.com/actuals-permanences/21-400-playtech-spins-(mansion-casino)/

[mr.ore]

Another test:
20k playtech spins failed in the end, and 10k playtech spins is an example of perfect loser - cascade of death pushing it all down  . To find optimal values for SMA is necessary, and maybe they should be adapted with the game - when there is a positive trend but graph is too choppy, it causes too many false stops, and causes drawdown. Or add another SMA to disable game for thousands of spins? Why when it is profitable in virtual play? This is difficult to find optimal parameters which would work well. I am almost sure, that no matter how well the parameters would be set, there would always exist a run from hell where those little peaks will slightly cross the barrier, disable game, then win, enable game, then do this again, and loses are being cumulated this way. This is the problem of moving averages, and stoplosses in general.

[mr.ore]

-