Is martin so bad? Or do you have to know how to cook it? - page 13

 
tol64:

I don't understand what you were counting.

And the result is just a beautiful fluke. The risks need to be much lower. By as much as 10 times. And the profit will also decrease. )) In general, you have to work on the system first of all. This one just has a fixed Take Profit and trailing.

I have counted the times of deposit increase in the testing period. Judging by your poor description random entries with fixed stops and profits should not work with such amount of trades. It would be nice to have a report. Look for a pattern.
 
ivandurak:
I counted how many times the deposit increased over the testing period. Judging by your poor description the random entries with fixed stops and profits should not work with such a number of deals. It would be nice to have a report. Look for a pattern.

Not 165 times, but only ~16 times. The initial deposit in the test was 100,000, not 10,000. And what I described works as shown in the second screenshot. With a martin you just happen to get the results as shown in the first screenshot.

So much for the "alchemy". )))

 
tol64:

Not 165 times, but only ~16 times. The initial deposit in the test was 100,000, not 10,000. And what I described works as shown in the second screenshot. With a martin you just happen to get such results as shown in the first screenshot.

So much for the "alchemy". )))

All seen now. Going to drown my toad.

If you are interested in the topic of martin. I can throw in an idea.

 
ivandurak:

I've seen it all now. I'm going to go drown my toad.

If you're interested in martin. I can float an idea.

I am interested in all ideas, even the rejected ones. All ideas I first try out in their pure form, and then the most interesting part starts - trying to improve and develop them.

And don't be in a hurry to drown the toad. If you manage to get such results steadily, as on the first screen, it == grail. )))

 
tol64:

I am interested in all ideas, even rejected ones. All ideas I first try out in their purest form, and then the most interesting part begins - trying to improve and develop them.

And don't be in a hurry to drown the toad. If you manage to get such results steadily, as on the first screen, it == grail. )))

We build a channel at high lots beginning from yesterday (being optimized) to today (being optimized). We set pending orders on the border of the channel, stops of pending orders in the middle of the channel, and the profit is equal to the channel width. Thus TP =2*CL. When one of the pending orders triggers, the second one is modified according to the opened position volume, according to the scheme 1 2 2 4 4 8 8 16 16 32 32, etc. If the price does not break through the channel, the pending orders are removed and we wait for tomorrow. If a volatility affecting country Germany France England Switzerland USA Japan is not trading today. The channel should meet the condition of being at least one pip wide and not more than one pip long. The general idea is to predict the volatility surge and it uses the Asia-Europe shift. The volatility is increasing and all the dancing around the channel is to find the targets.
 
ivandurak:
We build a channel on highs and lows starting from yesterday (being optimized) till today (being optimized). We set pending orders on the border of the channel, stops of pending orders in the middle of the channel, and the profit is equal to the channel width. Thus TP =2*CL. When one of the pending orders triggers, the second one is modified according to the volume of the open position, according to the scheme 1 2 2 4 4 8 8 16 16 32 32, etc. If the price does not break through the channel, the pending orders are removed and we wait for tomorrow. If a volatility affecting country Germany France England Switzerland USA Japan is not trading today. The channel should meet the condition of being at least one pip wide and not more than one pip long. The general idea is to predict the volatility surge and it uses the Asia-Europe shift. The volatility is increasing and all the dancing around the channel is to find the targets.

Quite. It is a good trading strategy even on its own. It makes sense to work on it. And volatility is my favourite subject in general. I even know people who use such systems in trading. In manual trading. But as I'm a robot :))), I want to make a full automatic. Or at least a wonderful semi-automatic machine, which can be controlled by light and not frequent mouse movements.

By random inputs above I meant roughly the same scheme, but clumsy (a test version for debugging in extreme situations). In other words, two pending orders are set, but indiscriminately, wherever the wind blows. But there is a control for losing trades. It is a kind of decision-making block when the algorithm knows what to do if something has gone wrong. I am still developing this idea because there are so many options to improve the results. I like to think about security. I think about it first and foremost. :)

 
tol64:

Quite. It is a good trading strategy by itself. It makes sense to work on it. And volatility is generally my favourite subject. I even know people who use such systems in trading. In manual trading. But as I'm a robot :))), I want to make a full automatic. Or at least a wonderful semi-automatic one, which can be controlled by light and not frequent mouse movements.

By random inputs above I meant roughly the same scheme, but clumsy (a test version for debugging in extreme situations). In other words, two pending orders are set, but indiscriminately, wherever the wind blows. But there is a control for losing trades. It is a kind of decision-making block when the algorithm knows what to do if something has gone wrong. I am still developing this idea because there are so many ways to improve the results. I like to think about security. I think about it first and foremost. :)

This is the indicator normalised to the 0 - 1 range of the ATP, which shows this picture. On the 15 minute chart.

If you find a use for it, whisper it to me.

It's also worth noting that the breakdown of volatility itself sucks a lot of false positives. But it allows reducing the number of reversals in a Martin.

Files:
NormATR.mq5  3 kb
 

As a variant of EA operation, without martin the standard drain after a period of optimisation.

 
ivandurak:

This is the indicator normalised to the 0 - 1 range of the ATP, which shows this picture. On the 15 minute chart.

The intraday volatility cyclicality can be seen quite well without normalisation. It just looks better with normalization.

ivandurak:

It is also worth noting that volatility breakout itself sucks a lot of false entries. But it allows to reduce the number of flips in the martin.

The point is that volatility breakdowns can be implemented in so many ways, so everything is as relative as ever.
ivandurak:

As a variant the work of the Expert Advisor without martin is a standard loss after a period of optimization.

Examine each drawdown in visualisation mode to find out what was actually going on there. A detailed analysis will give you ideas about how to reduce drawdowns. And you can also discover errors, after fixing them, the result can also improve. ))
 
tol64:

The intraday cyclicality of volatility can be seen quite well without normalisation. With normalisation it just looks more pronounced.

The point is that a breakdown in volatility can be realised in so many ways, so it's as relative as ever. Examine each drawdown in visualisation mode to find out what was really going on there. A detailed analysis will give you ideas about how to reduce drawdowns. And you can also discover errors, after fixing them, the result can also improve. ))
There is a thin gray vertical line on the equities chart which separates the optimization period. Everything else is forward. One can hardly expect stable results from a robot if optimization was performed under the tsar peas. Imho, at first I should find out how stable it is during overoptimization and what is the maximal drawdown. If the results are satisfactory, then you can think about a full automatic system with self-optimization. This perpetuum is not supposed to work.