Analysis of test results and optimisation in the MetaTrader 5 strategy tester - page 10

[Fernando Carreiro]

@Andrei Znamenskii #: just tried a complete overkill with no genetics.... same thing. Although all the parameters are highlighted.

Your original question was unclear, and it still is somewhat. So, let me try to clear it up for you ...

1. You can optimise over more than 16 parameters, without issue.
2. Your issue, is that you are not able to view more than 16 at the same time in the "Optimization Results" tab, so you will have to be selective and enable/disable according to your visual analysis. You can do that at any time during and after the optimisation is complete.
3. However, when you "Export to XML", you will have access to all of the parameters that were selected for optimisation (see point 1). So you view and process all the data as you require.

[Andrei Znamenskii]

Thanks for the pretty clear answer. Well, I'll have to look into it. I've never encountered this before.

I understood correctly that if I select all parameters (more than 16), they are all optimised, but I can see only 16. And if I want to see the others, I have to select them manually? But what I don't understand is this... Will I see the best results (e.g. optimisation by Balance) by the results of only 16 parameters or by all of them?

[Fernando Carreiro]

@Andrei Znamenskii #:

Thanks for the pretty clear answer. Well, I'll have to look into it. I've never encountered this before.

I understood correctly that if I select all parameters (more than 16), they are all optimised, but I can see only 16. And if I want to see the others, I have to select them manually? But what I don't understand is this... Will I see the best results (e.g. optimisation by Balance) by the results of only 16 parameters or by all of them?

It will optimise over all the parameters selected in the "Inputs" tab. You don't need to make any changes to your selections of the inputs and they can be more than 16. The optimisations will take all of the selected into account.

In the "Optimization Results" you can enable/disable which ones you see in the results for the columns or graphs, but it will not affect the optimisations at all, be it for Balance, or what ever criteria you selected. It will only affect the visual selection of which columns appear in the table, not the underlying results data. Think of it as just a visual filter that hides some of the data so it is easier to read and nothing else. So you can enable/disable on the fly without affecting the results.

[Andrei Znamenskii]

Fernando Carreiro #:

Optimisation will be performed for all parameters selected on the "Inputs" tab. You do not need to make any changes in the selection of input parameters, there can be more than 16 of them. The optimisation will take into account all selected parameters.

In the Optimisation Results section, you can enable or disable the display of results as columns or graphs, but this will have no effect on the optimisation, whether it is balance or any other criterion you have selected. It will only affect the visual selection of the columns displayed in the table, but not the underlying results data. Think of it as just a visual filter that hides some data to make it easier to read, and nothing more. So you can switch it on and off on the fly without affecting the results.

There, now all the nuances are clear. Thank you very much. I've been racking my brains about this....

[Ilya Malev]

It is written in the article https://www.mql5.com/ru/articles/2612:

Если же вам необходимо полностью достоверное воссоздание истории в тестере стратегий, то воспользуйтесь режимом "Все тики на основе реальных тиков". В этом режиме тестер самостоятельно выкачивает с торгового сервера брокера записанные реальные тики и строит развитие цены именно по ним. Для участков истории без реальных тиков тестер моделирует цену так же, как и в режиме "Все тики".  Таким образом, если у брокера есть вся записанная история по требуемым символам, вы можете провести тестирование подлинных исторических данных без искусственного моделирования. Правда, расплатой за такую потиковую точность будет существенное увеличение времени тестирования, как это показано в таблице с результатами сравнения трех режимов.

Where can you find out how, by what algorithm or principle, the tester "builds the price development just by them" (real ticks in the "Each tick based on real ticks" mode), and how the tester defines "history sections without real ticks", where "the tester models the price in the same way as in the "All ticks" mode" - what is the, formal definition, of "history section without real ticks". And how to make the tester pass strictly by those ticks that are given to it in the tick history of the (custom) instrument - give the Expert Advisor all the ticks that are in this history, and do not generate any ticks that are not in this history. Thanks

[fxsaber]

Ilya Malev #:

how to make the tester pass strictly by those ticks, which are given to it in the tick history of the (custom) tool - give the Expert Advisor all ticks, which are in this history, and do not generate any ticks, which are not in this history.

In the custom symbol, first write the tick history, and then write the self-generated M1-bar history from these ticks. Did it here.

[Ilya Malev]

fxsaber #:
In the custom symbol, first write the tick history, and then write the self-generated M1-bar history from these ticks. I did it here.

But if I don't want there to be minute bars, but for example I want to have 4-6 ticks per hour like OHLC, but not on M1, but for example on H1. And I want this OHLC to take into account not only high-bid and low-bid, but also high-ask and low-ask, in their real chronological sequence. Is it realistic to do this?

[fxsaber]

Ilya Malev #:

Is it really doable?

Yes. Exactly as I wrote earlier. In the Tester it will be a perfect match with what CopyTicks produces in the Terminal.

[Ilya Malev]

fxsaber #:
In a custom symbol, first record the history of ticks, and then - thehistory of M1-barsgenerated independently from these ticks. I have done it here.

What is the reason for the need to upload not only ticks, but also to generate minute bars independently from them? What will be done wrong if we let Metatrader generate minute bars etc. from downloaded ticks automatically?

[fxsaber]

Ilya Malev #:
What is the reason for the need to upload not only ticks, but also to generate minute bars from them? What will be done wrong if you let the metatrader generate minute bars etc. from uploaded ticks automatically?

I don't remember anymore.