How do you achieve a qualitative leap in market analysis? There is an option:

 

With the advent of History Center, a great many people are coming up with and then optimising their strategies for these quotes. I have the following situation:

1. I ran my strategy on HC history: Profit = 1000 pips
2. I tried my strategy on my brokerage company`s history: Profit = -1000 points.
3. I run it through my brokerage company's history performing trades at the same time as in 1st point: Profit = 800 points.

I.e. it turns out that the system displays signals on the normalized quotes of HC, rather than on the filtered quotes of DC.

The question is how to get the real-time quotes that are closest to the History Center? In order to receive signals based on these, even if they are purely indicative, and at the same time, trade at the real quotes of brokerage companies.

The problem of binding two Expert Advisors, launched in two different terminals (one - the real quotes of brokerage companies, the other - indicative or normalized quotes), is solved easily.

But how to obtain in the second terminal indicative or normalized quotes - have not yet decided.

There are two options:
1. Find a source of indicative quotes and send them to the terminal (several methods, how to do it). I haven't found such source yet.

2. Take all brokerage companies, that use MT4. And evaluating the current quotes of all brokerage companies, make the current normalized quote and direct it to the 2nd terminal.

I think that the second variant is the closest to the History Center. But how to do it without being a clumsy - I have not worked out yet.

I understand that there is a single format of quotes receiving to the MT4 terminal from brokerage companies' servers. Is there a description of it? In order to correctly implement the idea of getting normalized quotes in real-time.

I am sure the possibility for everyone to receive such real-time "objective" quotes - is a significant shift towards profitable strategies writing, based on History Center's analysis, and not to individually filtered quotes of each brokerage company, where the system works, as long as the appropriate filter works.

The easiest way to solve this problem would be directly to MT4 developers, but as always, in the end we have to rely on ourselves and others like me, i.e. YOU.

I hope I didn't do it in vain.

 
Try running your system on quotes from Alpari (rumoured to be close to NS?).

1. If everything works, you can open an account there as well :)
2. If you want to work with another broker, run the system on the Alpari demo, and execute orders wherever you want.
3. Try to filter the ticks yourself. Options:
a) exponential sliding on ticks with short period.
b) median on the last 3 ticks (more is possible, but probably worse)
c) some combination of a and b (i.e. first b then a)
 
This is not a problem. This is a pseudo-problem.

Some time ago there was published a remarkable article of professor Sinitsyn in the journal "Chemistry and Life".
In particular, there the question on formulation is considered.

For example, it sounds very interesting: "Influence of electromagnetic vibrations with wavelength from 3800 up to 7400A on crystal lattice of steel 25G2S". It's very scientific, almost majestic.
In reality, in plain language, this nonsense means the following: the effect of moonlight on rails.

I.e. it turns out that the system produces correct signals on the normalized quotes of HC, rather than on the filtered quotes of DC.
The question is how to obtain real-time quotes that are closest to the History Center? In order to receive signals based on these, even if they are purely indicative, but at the same time the trading was carried out at the real quotes of brokerage companies.

In medieval times, if people thought the Earth was standing on three elephants, it does not mean that it really is standing on them. It's not worth developing a big trough from which to feed those elephants. They simply do not exist.

 
Mak писал (а):
Try running your system on quotes from Alpari (rumoured to be close to NS?).

1. If everything works, you can open an account there as well :)
2. If you want to work with another broker, run the system on the Alpari demo, and execute orders wherever you want.
3. Try to filter the ticks yourself. Options:
a) exponential moving average on ticks with a short period.
b) median by last 3 ticks (more is possible, but probably worse)
c) some combination of a and b (i.e. b first then a)


" Don't vat on the kirsa. We'll poison each other for good!" (M. Zhvanetsky)
 
SK. 23.11.2006 22:44
In due time in a magazine "Chemistry and Life" the remarkable article of professor Sinitsyn was published.
In particular, there the question on formulation is considered.

For example, it sounds wonderful: "Influence of electromagnetic vibrations of wave length from 3800 to 7400A on a crystal lattice of steel 25G2S". Straightforward, scientific, almost majestic.
In reality, in plain language, this nonsense means the following: the effect of moonlight on rails
.

Cool! ))))) and right on!
If only they could pick up on the wind outside the window or the rustle of fallen leaves and start making up algorithms for them and complain about "fluffiness".
 

Scepticism is a good thing. It is a shame that it is often based on a belief in the correctness of one's logic and reasoning.

Engaging in polemics is "the last thing to do". There are certain facts:

1. I don't know what the quotes in History Center are.
2. The developers say they are normalised quotes from many sources.
3. My system is not profitable on DC quotes in the tester.
4. The algorithm is not pipsing.
5. Optimization function on History Center changes smoothly from user input parameters.
6. The results of the previous paragraph, applied to History Center, also give a smooth change.
7. Applying the system, as written above, I gain profit on brokerage companies with maximal drawdown up to 200 points.
8. I have made several hundreds of deals, and it is more than two years.
9. The system works only with one symbol.
10. When I change currency pairs (tried only major pairs) I get profit.
11. Only one position is always opened + a pending order, which will be triggered only after closing the current position.

Reasoning from these facts, I did not give up the system at once, reasoning "never, because never". I want to find out where the error is, if there is one. The probability is 99.999999%, maybe higher.

Here is my opinion.

 
getch 23.11.2006 23:33

Scepticism is a good thing. It's a shame that it is often based on confidence in the correctness of one's logic and reasoning.

It's not scepticism but hard truth, and it hits very hard.

1. I don't know what the quotes in History Center are
2. The developers say they are normalised quotes from many sources.

What good are smooth and (convenient for Expert Advisors tested on history), but in reality all such experts lose
(it's a known fact) they even have a name for them - "Grails".
I think it is more important, if in "hard quotations history" EA will earn at least three quid - chances are this EA will work on a real account.

3. My system does not make profit on GC quotes in the tester.
4. Algorithm is not pipsing.
5. Optimization function on History Center changes smoothly from user input parameters.
6. Results of the previous paragraph applied to History Center also give smooth changes.
7. Applying the system, as described above, I make profit on brokerage companies with maximal drawdown up to 200 points.
8. There are several hundreds of deals, my period is more than two years.
9. The system works only with one symbol.
10. When I change currency pairs (I tried only basic ones) I get profit.
11. Always open only one position + pending order, that will work only after closing the current position.

You are unfortunately not alone in this, many others have similar problems (by the way I am not an exception).
Based on the statements of one of the leaders of the Championship (in my opinion, they should be listened to) there are no simple and, moreover, profitable systems.

On the basis of these facts I did not give up on the system at once, reasoning "never, because never". I want to find out where the error is, if there is one. And there is a 99.999999% chance of that, maybe higher.


I totally agree.

And the "awkward quotes" is just one of the problems that need to be solved.
I, for instance, am trying to solve this problem by installing an appropriate filter.

By the way, no offense meant - sorry.






 
Obvious and profitable systems may not exist. The simplicity of an idea and its obviousness are different things. For example, neural networks are simple, but not obvious. Trading on indicators - complicated, even though it is the first thing that comes to mind. There are simple ideas, very original and non-standard. And in spite of their simplicity they are not obvious at all. Automated trading brings me profit for more than one and half year, I have not participated in writing this MT4 system, only programming languages and mathematical packages. Unfortunately, I was not able to realize this system in MQL4 to be as effective and stable in real trading. Because unlike on MT4, in terminals of western brokers there are much more types of orders and their execution is much more strict. I am talking specifically about REAL. For detailed analysis the Matrix and in-house developed ones are indispensable. But I still want to hear something realistic: real-time History Center. I am not familiar with hacking programs, I think there is a civilized way to find out how quotes from brokerage company's server come to the terminal.
 
Why haven't the reports and source code been published?

It is likely to turn out that the expert is too sensitive to quotes.
 
I think a lot of people would be interested in assessing their expert if the results on one story were overlaid on another. The idea came up, I just wanted to look at it. It came out of the blue. I will not publish the source code for reasons of principle. I can post the tester report when I get to the Expert Advisor. But what is the point of all this? Why should I believe or not? I want to hear something about the implementation of the real-time History Center, and not to discuss the characteristics of this and other systems.
 

Renat, is it possible to solve the mystery of the quote normalisation algorithm just a little bit? I'm not asking you to give me the sources of the quotes. It's just that I still don't understand what normalisation is.

If this normalization leads to quotes that show different results in testing than those obtained using non-normalized history, then what is the point of it?

Here is an example: On October 2, there was a giant spike on gold at 01:20 at night (on Hi-Low watches = 23.5 figures!), and up to the minutes it is viewed as a single quote without any movement. Alpari, according to several people's feedback, compensated for the losses caused by this spike. Apparently he removed it from the real as not marketable, however it remained on the demo. I was told that the same spike was observed in other brokerage companies. What to consider it as - market or non-market? And what happens to it when it is normalised?

2 getch: what is the expectation of your trade when you test it? I don't think it is 2-3 pips, but even if it is less than 15 pips (for some major), it is still a pips, which obviously will be sensitive to quotes...

By the way, here's the calculation: 1000 pips on a few hundred trades is well under 10 pips...