From theory to practice - page 1211

 
multiplicator:

But these results only work for the optimization period. If you optimize for a different period, you get different results.

That's what I'm talking about.)

You have to agree that it's useless to use such a method.

What's the reason why a tested sample of 50 samples, for example, will still work in the future?

yes no because it's extremely simple - the market is constantly changing)

 
Martin Cheguevara:

Renat, do you remember the topic about variance and the line which is displayed so that the price variance around this line is about constant?

can you tell me where to look)

I think I already wrote you about it, maybe in private

have a look - is it wrong?

 

this is what was:

this is what became:

the test was conducted from 09.2014 on EURUSD to 03.2015.

For those who don't understand why I'm showing the quote here:

to put it simply, one open trade to the wrong place at the wrong time and you have -200 bachinskys from the minimum lot of just one order not to mention grids and increasing lots...

And here are examples of finding extrema by my analytical system:

I think no one else has to explain the veracity of my statements above. Unless you have to.

I repeat for all those who want to get something out of the market: unadapted sampling = failure, sooner or later.

And yes, the strata works on rollback in case anyone doesn't understand...
 
Martin Cheguevara:

That's what I'm talking about.)

You must agree that it is useless to use such a method.

What's the reason why a tested sample of 50 samples, for example, will still work in the future?

no, because it's extremely simple - the market is constantly changing)

There may be two variants:

1) if we optimize the strategy on different years, and each time we get different parameters, and the parameters of a particular year do not work on other years, then it is a dumb fit to the history. we fit the parameters to a particular piece of history. then the strategy is rubbish.

2) The market is volatile and parameter selection should be dynamic.

But how can these parameters be selected dynamically? for the last smallest period! If parameters selected for the last smallest period do not work, then check point 1 (strategy is rubbish).
 
Renat Akhtyamov:

I think I've written to you about this before, maybe in person.

look it up - is it wrong?

Thanks new-rena))

my robot has a rich toolkit for analyzing quote data - I've already done everything (I meant the question I asked an hour and a half ago))

hope that you do too)
 
multiplicator:
There may be two options here:

1) if we optimize the strategy on different years, and every time we get different parameters, and parameters of a particular year do not work in other years, then it's a dumb fit to the history. we adjust parameters to a specific piece of history. it means the strategy is rubbish.

2) The market is volatile and parameter selection should be dynamic.

But how can these parameters be selected dynamically? for the last smallest period! If parameters selected for the last smallest period do not work, then check point 1 (strategy is rubbish).

I've been struggling with this question for a couple of years.

So it's not that simple.

And it's not available on the internet, and it's obvious why.

So, the research is kind of over.


Good luck in our very difficult business!

 
Martin Cheguevara:

Good luck to everyone in our very difficult business!

I had to bump orders like that too, because you're right, it goes further at times, even if an extremum seems to have been found... But in the end I gave up, it was a hassle.

The more orders, the lower the profitability.

And the funny thing is - grandmas tends to go from plus to minus, from one pair to another, from one account to another,

i.e. anything as long as the equity does not go higher than the initial deposit. .......

so it's too early to say goodbye ;)

 
I looked it up, the channels were handled here on the forum by Nikolai Semko.
here's the topic:
https://www.mql5.com/ru/forum/28700
Индикаторы: Индикатор - Каналы
Индикаторы: Индикатор - Каналы
  • 2012.12.27
  • www.mql5.com
Статьи и техническая библиотека по автоматическому трейдингу: Индикаторы: Индикатор - Каналы
 
about some of the problems with the system being discussed here... our market model is a constant channel. 1) But even if the market were a constant channel - we still wouldn't be able to "take it". Because our sma indicator is not perfect. 2) The market is not a flat.
In short: you should look for a better indicator and learn to avoid unfavourable periods.
 
Martin Cheguevara:

I've been struggling with this question for a couple of years.

So it's not that simple.

And it's not available on the internet, and it's obvious why.

So, the research is kind of over.


Good luck to everyone in our very difficult business!

after optimising in the optimiser, we get a graph of results like this:
like an inverted parabola.

If you optimise for another month, the parabola shifts to the left or right.
something similar was once shown by Joker. he tried to predict how this parabola would move.

He even launched his own signal. but his signal "ended badly". hence, we can conclude that he never completed his theme.