Sensation! A profitable strategy for playing beagle has been found! - page 9

 

Whichever price series generation function you use (random numbers or market prices), you can see statistics on how many consecutive losses your system has had over the study period. Based on this data, you can calculate the initial lot and initial deposit to use Martin on your strategy.

You cannot win on an infinite series, but you can win on a finite series. It does not contradict mathematics.

Why not risk your capital with 10 years of statistics, with the prospect of super profits? No one is forcing you to trade all the time, withdrawals happen too. If you make a double, you get it, and then you don't have to worry. The probability that you have activated your system to make a double, just when the 10-year statistics have changed, is VERY low.

Then why not take up the cause instead of blathering on?

Never used martin myself, but it's earned fine. If tomorrow scientists prove that the market is pure eagle-eyed, then what? Give up trading because mathematically at infinity you will inevitably lose?

 
mql4com >> :

Whichever price series generation function you use (random numbers or market prices), you can see statistics on how many consecutive losses your system has had over the study period. From this data you can calculate the initial lot and the initial deposit to use Martin on your strategy.

The statistics you will be looking at are only one realisation of a random process. It is very risky to draw conclusions about the whole process based on this one realisation. It can certainly be done for ergodic processes, but who says that a price process (or price returns process) is ergodic?!

 

Ultimately, it is possible to take a risk. An infinite series is an unattainable concept. Every person is mortal, so his "series" is finite, therefore if one selects such a martin, which would be profitable with a high probability at the 40-60-year period, then one can take a risk.

Secondly, one can risk a relatively small amount while obtaining a profit exceeding the initial deposit by hundreds of times. I would risk $200 for the sake of $200 000, even if it were a martin. And by the way, there are real examples. For example Andrey Trefeev's PAMM.

 
Mathemat >> :

There is always a risk. Everyone defines their own risk tolerance in their own way.

 

to C-4

I've got it figured out, I just don't understand MQL... I just don't understand MQL, so I had to read the documentation.) But I still do not understand what has been changed by HideYourRichess. So, it really does not work properly? How awful, you should be stricter with them. By the way, how can I be so sure I'm using a standard Cysh generator and not some tcl-oas? I couldn't find anything in documentation, is it some kind of insider information?
 

Wow! What a "Mathsrand" there is!

void MathSrand( int seed)

The function sets the initial state to generate a series of pseudorandom integers. To re-initialise the generator (i.e. set the generator to the previous initial state), the value 1 must be used as the initialisation parameter. Any other value for the initial number sets the generator to a random starting point. MathRand returns consecutively generated pseudorandom numbers. Calling MathRand before any call to MathSrand generates the same sequence as calling MathSrand with parameter 1.


However, you know all about it, but I see it for the first time. How interesting it is, damn it 6o) It's not so familiar after maths and matcad.


PS:

How about trying it this way:


for(int i=2;i<100;i++ )
{
MathSrand(i);
MathRand();
}

return(0);
}


Will the result be the same?



....


Tried it, the result is always the same:

2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 16 value 90
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 15 value 87
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 14 value 84
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 13 value 81
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 12 value 77
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 11 value 74
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 10 value 71
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 9 value 68
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 8 value 64
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 7 value 61
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 6 value 58
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 5 value 54
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 4 value 51
2009.06.12 00:07:12 ddd EURUSD,M15: Param MathSrand(i) 3 value 48


The price goes up. Is it supposed to be like this? The value of 1, I think I did not use, the code is very simple, I even understand it.


 
grasn >> So, is it really not working properly? How awful, you should be more strict with them. By the way, how could you be so sure that it's a standard sysc generator and not some tcl one? I couldn't find anything in documentation, is it some kind of insider information?

Seryoga, I was interested in this question some time ago. Stringo himself has answered me, that MathRand() is wrapper of Cish rand(). May I help you to get the link?

And the problem, which I stumbled upon C-4, Yuri and I discussed on our mutual resource. First Yury, just like C-4, was generating Bernoulli scheme with p=0.5, and he too obtained that subsequences longer than 14-15 units (zeros) don't occur. I suggested a general solution, a particular case of which HideYourRichess pointed out here. It happens that subsequences with length up to 20 and more zeros (ones) appear in long sequences as well.

P.S. And MathSrand() only needs to be called once, say, in init().

 
Mathemat >> :

And the problem that C-4 encountered, Yury and I discussed on our common resource. At first Yury, just like C-4, generated Bernoulli scheme with p=0.5, and he, too, got that subsequences longer than 14-15 units (zeros) don't occur. I suggested a general solution, a particular case of which HideYourRichess pointed out here. In the long sequences, subsequences of up to 20 or more zeros (ones) also appeared.

What's the common solution? Common solution of what?

C-4 >> :

Z.U. Now I think, maybe write a letter to ANSI C, and outraged, what the hell is wrong? Why did your random generator make me lose respect? Are you guys eating pancakes? Fix it!!! :)))

As a consolation, you can take credit for finding a simple method of evaluating the quality of a cis generator.

mql4com >> :

You can't win with an infinite sequence, but you can with a finite one. It does not contradict mathematics.

Not really. On a finite series, with equal stakes and equal initial capitals and with an "equilibrium coin", you can "win" in about half the cases. This does not contradict the maths. The total winnings will be about 0.

grasn >> :

Here is this topic "... The problem is just this: when even/non-even, it gives amazing results, when in usual way - trivial results" was discussed not only here and with my answer I, of course, brought nothing new, except a small remark which I am not sure about (somewhere it was discussed on exbite and such problem was solved somehow). What is your problem?

I have no problem with that. It's very clear and straightforward to me in this situation. You can't use the "even/uncount" operation as a method of generating a series of random binary values with probability 0.5. At least for a Cish generator.

 

to Mathemat

Серега, я в свое время интересовался этим вопросом. stringo сам ответил мне, что MathRand() - обертка сишной rand(). Тебе найти ссылку?

And the problem C-4 has encountered, Yury and I discussed on our general resource. At first Yuri , just like C-4, created a Bernoulli scheme with p=0.5 and he too did not find subsequences longer than 14-15 ones (zeros). I offered a general solution, a particular case of which HideYourRichess pointed out here. It happens that subsequences with length up to 20 and more zeros (ones) appear in long sequences as well.

I see I'm getting deja vu here. I've seen it somewhere and I know it exactly! Thank you for clarifications. I didn't "catch" it at once, for the simple reason that I'm not using MQL so far and am not really interested in these problems. Assembled an astrolabe from VisSIM, MAthCAD and MT (which I wrote about earlier). It works, but unfortunately my 2 core computer with 4 GB RAM gets so loaded after 3-4 hours that it can barely crawl. Somewhere there is a mess. :о(((

P.S. And MathSrand() need only be called once, say, in init().

This I understand, but it seems that the example does not contradict the documentation and the numbers should be generated randomly. Or maybe I do not understand something?


to HideYourRichess

I have no problems. Everything is quite clear to me in this situation. You cannot use the "even/odd" operation as a method of generating a series of random binary values with the probability 0.5. At least for a C-generator.

Sincerely happy for you :o) And if you try dividing not by "2" but by "2.0" in parity condition:

MathMod(MathRand(), 2)==1)

MathMod(MathRand(), 2.0)==1)

 
grasn >> :


to HideYourRichess

Sincerely happy for you :o) And if you try dividing not by "2" but by "2.0" in the parity check condition:

MathMod(MathRand(), 2)==1)

MathMod(MathRand(), 2.0)==1)

A very fresh solution. But it won't save the patient. It's a medical fact.