Help make the robot profitable - page 5

 
forexman77:
Just wondering how stable the system has proven to be over time.

I've recently tested some EAs and realised that they need to improve, at least find new settings (old ones were from early 2013).
Or to upgrade somehow, on/off martingale, trading on certain market conditions (AbsultreStrength indicator) and so on.

But the signal systems(opening of buy or sellpositions "by the arrows on the chart" using indicator-filter for false signals) - is an eternal theme, that is - will never cease to be relevant. The red arrow down is a sell, the blue arrow up is a buy... AscTrend for example :)
 
Евгений Панин:

Clear sideways, but ADX does not help to filter it out.

What sideways filter options can you suggest?

If you want to enter the market when it is trending, then first fix the trend and then enter.

I`m in a trend when the price crosses МА 96-176 - you may count bars or points ...

I recommend to use the SCO channel - enter only if price has broken through it and exit the same day on the way back to the channel.

You should use the channel for the last n days - from the start date of the day to the end date, excluding the current day.

 
-Aleks-:

If you want to enter the market when it is trending, first fix the fact that there is a trend and then enter.

I fix the trend when the price crosses the MA 96-176 - you can count bars or pips...

I recommend to use the SCO channel - enter only if price has broken through it and exit the same day on the way back to the channel.

You should use the channel for the last n days - from the start date of the day to the end date, excluding the current day.


Alex, thank you! I will try
 
Евгений Панин:
And how to interpret them correctly?

for the purpose :-) look at the formulas to begin with...

You can tell that for example the SMA does show the trend, but the way it was half a period ago. It is not because it is so mean, it was made that way. We may consider that SMA shows a certain price equilibrium for a period. But it has smooth derivatives.

By the way, the thoughtless addition of indicators/filters/signals, not coordinated with other components, kills the whole idea of the strategy and makes it 100% unprofitable.

I agree with @newdigital, strategies are made in a very long and painful way. And coding is such a last stage that it sometimes doesn't even come to that - quite everything works by hand and personal discipline.

PS/ as a small boast, a strategy developed six months, only on MA, in the alpha stage of running it by hand and catching up with the right stats (entering the market today from 9.30):

 
Petros Shatakhtsyan:

If anyone thinks that what they do with their hands will be able to program them, they are deeply mistaken.

So far, no one knows how to programmatically identify things like current support/resistance levels and the current trend or channel.

Without them it is impossible to make a consistently profitable trading robot.

I don't quite understand why it is impossible to define support/resistance levels or a trend-following channel.

Of course, everything that someone does with his hands cannot be programmed - because hands sometimes make completely unpredictable, in fact, random actions.

But, if the actions are clearly defined - why can't they be programmed?

Take support and resistance levels - what is so "unprogrammable" there? You take a fractal extremum and that is the level.

Or a channel - the same thing, take the maximum and the minimum for a certain period - here you have a channel. What cannot be programmed?

The only thing that cannot be programmed is "according to the left heel". When the trader takes a number from the ceiling and says "this is the level". Everything else is easier or more complicated, but programmable.

 
Maxim Kuznetsov:

we can say that for example the SMA does show a trend, but the way it was half a period ago. It is not because it is so mean, it is just because it was made that way. We may consider the SMA shows a certain price equilibrium for a period. But it has smooth derivatives.

Where do they come from, the smooth ones? SimpleMA? It also has a tail. If it's simple, then it should be WMA. And by changing the scales, you can get the response you want without involving complicated maths.
 
Yuriy Asaulenko:
Where do they come from, the smooth ones? SimpleMA? It also beats the tail. If you want simple, you'd better use WMA. And by changing the scales, you can get the response you want without involving complex maths.

For two or even three years I have been trying to make the point on various forums (including this one) that you are confusing tangents and secants and derivatives with increments. All of you have been trying to persuade me that you are confusing tangents and secants with increments.
I'm not going to change your mind.

But SMA does have unexpectedly smooth derivatives.

By the way, LWMA is the limit of the average of SMA from 1 to period:-)

 
Maxim Kuznetsov:

I've been trying for two or three years on different forums (including this one) to convey the idea that you confuse tangents and secants and derivatives with increments. Everyone has unanimously resisted, preferring to remain within the framework of the 8th grade secondary school.
I will not change your mind.

But the SMA does have unexpectedly smooth derivatives.

By the way, LWMA is the limit of the average of SMA from 1 to period :-)

First, subtract the two SMAs with a shift of 1 - get the sum of the differences of the two extreme terms - is that what you consider a derivative? Smooth - the weights are small, so it's smooth. But it's anything but derivative - isn't that too primitive? Of course it's a derivative of the SMA, but it has nothing to do with the signal anymore. Roughly speaking, we will get something like momentum difference divided by the coefficient.

It's not about the efficiency of SMA in a particular strategy and not about its application. It's out of the question.

 
George Merts:

I don't quite understand why it's impossible to programmatically define support-resistance levels or a trend channel.

Of course, everything someone does with their hands cannot be programmed - because hands sometimes make completely unpredictable, in fact, random actions.

But, if the actions are clearly defined - why can't they be programmed?

Take support and resistance levels - what is so "unprogrammable" there? You take a fractal extremum and that is the level.

Or a channel - the same thing, take the maximum and the minimum for a certain period - here you have a channel. What cannot be programmed?

The only thing that cannot be programmed is "according to the left heel". When the trader takes a number from the ceiling and says "this is the level". Everything else is easier or more complicated, but programmable.


You have to be a programmer to do this and try to program them. That's when it will be clear why no one knows how to program the current trend.
 
Petros Shatakhtsyan:

You have to be a programmer to do that and try to program them. That's when it will be clear why no one knows how to program the current trend.

Kind of a programmer.

Seems to have plenty of experience.

It seems to be a cakewalk to set the level by the fractal. The level is easily calculated and may be used.

Setting the channel by maximum/minimum seems to be no problem. The channel is easy to set, you may use it.

What's the problem? It's all easily programmable. How do these levels or channels differ from those created manually according to the fractal or highs/minimums? Their functionality is the same whether they are calculated manually or automatically. If a level rebound happens, it will not depend on whether it is set manually or by the Expert Advisor.