Looking for patterns - page 5

 
I have found a 100 per cent pattern in the macd indicator. Naturally, I won't tell anyone.
 
The field is such that knowledgeable people are reluctant to share their knowledge, because it is not profitable to educate their competitors).
But okay, here's another pattern. If we take the market as a source of entropy, and two people trade against each other, the probability of winning is greater for the one who has more money. For example, two people are playing, one has $100, the other $10000. Then the other one will come up with $1000, and he will also divest the one who has more money. So the one who has more money will win. Hence, the abstract strategy: if you find a specific participant and trade against him, then you can earn more money, even if the market is random. By the way, those who have access to certain information do not disdain this algorithm.
 
In general, it would be more useful to start with a definition of what a trend is. Because as much as I have written here, for some reason people have avoided this question. And it is fundamental.
 
Vladimir Baskakov:
I have found a 100 per cent pattern in the macd indicator. Naturally, I won't tell anyone.
I know this 100% pattern)
 
The problem is that in today's world people can't test their patterns.
No one will teach you and no one will tell you.
Manual trading will always have psychological and irrational risk.
Manual trading will never be faster than automatic (EA) trading.
It is physically impossible to follow the market 24/7.
We cannot predict the future.

But we do have the ability to use real market quotes and Spreads.

https://www.mql5.com/ru/blogs/post/73260

Any of your patterns can translate into statistics.
Statistics determines the possibilities of your patterns.
 
Maxim Romanov:
It's an area where knowledgeable people are reluctant to share knowledge because it's not profitable to educate their competitors)
But okay, here's another pattern. If we take the market as a source of entropy, and two people trade against each other, the probability of winning is greater for the one who has more money. For example, two people are playing, one has $100, the other $10000. Then the other one will come up with $1000, and he will also divest the one who has more money. So the one who has more money will win. Hence, the abstract strategy: if you find a specific participant and trade against him, then you can earn more money, even if the market is random. By the way, those who have access to certain information do not disdain this algorithm at the exchange.

How have you checked the implementation of this pattern?

I remember a case when I was a child: we were playing with badges using the knock-out method. And I, in general, played as well as everyone else. One day a friend came to visit me with three badges, while I had a few dozen in the box. And so we played with him and I lost absolutely everything.

The other point is that actually where we are, we don't play against each other. Just because I opened 0.01 lot on a particular pair at a particular moment does not mean that someone necessarily did the same in the opposite direction. And the fact that I closed an order with a profit of USD 1 does not mean that someone will close the same order with the same minus. Objectively, if you look at it objectively, accounts with $10000 will sink just as nicely as accounts with $100. And they are not draining to each other but to somewhere else.

Another point is that if someone has access to certain information, they have an advantage with any amount in the account.

That is, completely different initial conditions and you can't tie it all into one.

 
Maxim Romanov:
And it is fundamental.

Maxim, I didn't want to get into a pips/points discussion here. I'll try to tell you how I see it. A trend is a directional price movement that periodically updates the extremum towards which it is moving. If the extremum is not refreshed for a long time, it means that there is no trend in that direction.

If you have a different view, tell me. I will be glad to listen to you, but I will not argue and I will immediately agree.

 

Trend is an English term. And in my opinion the most correct meaning in this field is aspiration. Aspiration has sources and boundaries. But there are no parameters such as length and intermediate points concerning distance. The difficulties in this regard are enormous. It is practically unrealistic to determine in advance the end of one aspiration and the beginning of the opposite one. There are levels, but they are always seen as probabilities rather than clear reversals. Given the influence of fundamental data, force majeure and other events, many movements become clear after they are fully completed.

I believe it is these uncertainties that have led to the debate about the presence of a trend at all. I am closer to a position expressed as "the trend is not my friend" than to a denial of this phenomenon in general.

That is, there is some trending (as a process) in one of two directions. It starts and ends equally suddenly and at any moment. It is influenced by an immeasurable quantity of versatile and unrelated data. In this case technical analysis is effective only in situations of relative flat, and almost completely (or almost) loses its power under the conditions of a strong pressure from any of a huge number of parties affecting the quotes.

 

There is a question that has been bothering me for a long time, but there is no one to ask. And I haven't had the chance to ask it. I think this is a good time. I do not know about experts and how they are written. But I've read about self-training robots long ago. And so I imagined such a few lines of robot that are self-learning and couldn't. I apologize in advance if my imaginings are ridiculous and may cause mental injury to anyone. But can someone explain in general terms the structure of trading robots? If they work purely by a static algorithm, the probability of losses is inevitable. It's a matter of time and chance. But if we consider large-scale databases based on detailed history and dimensionless numbers of variants of events and reactions distributed over a multitude of parameters, then the question of the so-called grail could have been solved long ago.

And here is the question: are there any automatic trading systems continuously connected with external sources that contain such fine details and variants of events-reactions? Or does the whole range of such automated systems consist of a few kilobytes algorithm so far?

Once again I apologize if my question touches someone's professional dignity.


Alexey, as a developer could you give an answer to this question?

 

Vitaly, a broad question.

It is impossible to create a completely profitable EA, or to trade manually in the same way. The ideal EA, which everyone strives for, makes losing trades. But the total profit is greater than the total loss and this condition should be met at any time. There is enough history now to find the patterns, so I think there is no point in creating a self-training robot. Another thing is that the Expert Advisor should have several strategies for working in different parts of the market. Therefore, we need a mechanism for recognizing these sections to switch between strategies, because the nature of price movement changes from time to time.

And continuous self-training smells like adjusting of parameters to a small part of history. With a disappointing result in reality. My personal opinion, without claiming to be objective.