Discussing the article: "Brute force approach to patterns search (Part VI): Cyclic optimization" - page 2

 
fxsaber number of trades from below always gives a beautiful curve on Sample. In this case such a TS can be considered as something rational, even if there is a complete loss on OOS. But it seems doubtful to search for beauty by Sharpe, R^2 or criteria in this article. Perhaps I am wrong.

Approaches are good, but with this you throw out most of the good variants, which may not be so good on the profit factor, but you should understand that in addition to this indicator there are a lot of curve parameters, which can give you an even line. All these are statistical characteristics that are not considered in classical approaches, but they can be generalised, for example, by applying my criterion. By the way, here is a link to an article where I proved many interesting things mathematically. It is based on the example of the buy algorithm, but in fact it is the analysis of many fundamental things. There is an effect that if you raise the lower boundary on trades, you automatically increase the probability of finding a flat line. This is mostly due to mathematics. If we start to understand what is a rational or good system, we will go into such maze )). At least a quantitative indicator should first be invented to determine the suitability).

Алгоритм докупки: симуляция мультивалютной торговли
Алгоритм докупки: симуляция мультивалютной торговли
  • www.mql5.com
В данной статье мы создадим математическую модель для симуляции мультивалютного ценообразования и завершим исследование принципа диверсификации в рамках поиска механизмов увеличения эффективности торговли, которое я начал в предыдущей статье с теоретических выкладок.
 
Evgeniy Ilin #:

there are a bunch of other curve parameters that can give you an even line.

It is not needed on Sample, but on OOS.
 
fxsaber #:
You don't need it on Sample, you need it on OOS.

Sample is a sample, OOS is an unknown abbreviation.

 
Evgeniy Ilin #:

Sample - sample, OOS is an abbreviation unknown to me.


Out Of Sample
 

Author, if it's not a secret, what kind of fin.res do you get out of it?

Are you still using the method?

 
Kristian Kafarov #:

Author, and if it's no secret, what kind of financial res you get out of it?

Are you still using the method?

https://www.mql5.com/ru/blogs/post/756379

Greetings, the system has been upgraded to a product as a result of the research. This is a guide to it. All the links if anything are there.

Neon Trade & Neon Shadow (руководство по установке и эксплуатации)
Neon Trade & Neon Shadow (руководство по установке и эксплуатации)
  • www.mql5.com
Русский     English     中国的     Español      Português      한국의     Deutsch      日本の     Français
 
Evgeniy Ilin #:

https://www.mql5.com/ru/blogs/post/756379

Greetings, the system has been upgraded to a product based on the results of the research. This is a guide to it. All links if anything are there.

Instead of constant re-generation of strategies, have you tried to select a stable set of strategies that go up and down in a sine wave on a large history (with a clear general trend, i.e. profitable) and enter them on drawdowns?

PS: the project is certainly powerful, but very complex. And I have little faith in robustness of complex models on financial markets at the moment (subjective judgement).
 
Kristian Kafarov #:

Instead of constant re-generation of strategies, have you tried to select a stable set of strategies that go up and down in a sine wave (with a clear general trend, i.e. profitable) and enter them on drawdowns?

PS: the project is certainly powerful, but very complex. And I have little faith in the robustness of complex models on financial markets at the moment (subjective judgement).

Well, I do so from time to time, there is one of the portfolios, which I assemble manually, using settings from the channel. That's how it works out in the end. Any strategy will need readjustment after a while. My setup generation system does just that, saving me the manual search - readjustment. About drawdowns it is another topic, yes it works and moreover it can be fractalised (just a hint, who knows will understand). It is too much of an imbovy topic, I did not dare to write such an article here, but I have developed this method a long time ago. This is already a hard departure into virtualisation and nesting of levels. For 200 quid, it's a waste to post such an article here.... There is an Expert Advisor that uses this method in full force, but it is only developed, now we will test it with a partner. After some time there will be a version for mql5, but I can't say about the terms yet, but the prototype is ready for testing.

 
Evgeniy Ilin #:

Well, I do that from time to time, there is one of the portfolios that I put together myself manually, using the settings from the channel. That's how it works out in the end. Any strategy will need readjustment after a while. My setup generation system does just that, saving me the manual search - readjustment. About drawdowns it is another topic, yes it works and moreover it can be fractalised (just a hint, who knows will understand). It is too much of an imbovy topic, I did not dare to write such an article here, but I have developed this method a long time ago. This is already a hard departure into virtualisation and nesting of levels. For 200 quid, it's a waste to put such an article here.... There is an Expert Advisor that uses this method in full force, but it is only developed, now we will test it with a partner. After some time there will be a version for mql5, but I can't say about the terms yet, but the prototype is ready for testing.

Here, a person is trying to apply a very similar approach, on crypto and stock markets.

Конвейер производства граалей
  • smart-lab.ru
Привет всем! Создал оглавление своего блога, а кому интересно — гляньте. Ну а тут напишу что-то суммаризирующее мои изыскания. Как
 
Kristian Kafarov #:

Here a person is trying to take a very similar approach, on crypto and stock markets.

Just general information, something similar. He uses indicators, I could even tell him how to improve them. In the bare form of indicators are certainly not very effective, with the exception of oscillators, but even then it is better to apply to relative values. In short, summarising, any approach has the right to exist with sufficient elaboration.