You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
The Euro Pound was in a clear bullish trend at the end of 2007, and then on January 15 it immediately turned into a bearish trend, with a drop of 200 pips in 2 days.
Hence - adjusting the EA for the last 2-3 months will lead to unpredictable consequences.
Or am I wrong about something?
Well, if we are counting on a continuing bullish trend, then we don't need an EA - just open to Buy and go!
There was a clear bullish trend on the Euro Pound pair at the end of 2007. And on January 15 it instantly changed to bearish, with a drop of 200 pips in 2 days.
Hence - adjusting the EA for the last 2-3 months will lead to unpredictable consequences.
Or am I wrong about something?
Well, if we expect a bullish trend to continue, we don't need an Expert Advisor, just open a Buy position and go!
Well, I'm not going this way with you guys.
And the lag of indicators in such situations - does it not happen?
Or do they look ahead by a couple of days?
And in general, I cannot understand how an EA can be optimised in the last 2-3 months if the situation changes drastically over the course of a day in the current market. Unless it's before a short-term championship, and that's it.
I have an Expert Advisor with a built-in flexible optimization system that can optimize itself based on the latest historical data: when the behavior of the symbol changes, depending on the status of open positions, simply by the time period, etc. And the optimization period it tries to vary according to a number of criteria.
For a long time I was using it in the tester and on the demo - the result was the following: yes, there are periods when tuning to the latest history data may bring profit, but such a period will be necessarily followed by the period (I call it "atypical" for me) when this profit will be lost and result in 0 :) Although, it does not sink much... Less stable, something happens on TFs from 4H and higher, the global trends are not so volatile. But even there the profitability is so bad....
And in general, I cannot understand how an EA can be optimised in the last 2-3 months if the situation changes drastically over the course of a day in the current market. Unless it's before a short-term championship, and that's it.
I have an Expert Advisor with a built-in flexible optimization system that can optimize itself based on the latest historical data: when the behavior of the symbol changes, depending on the status of open positions, simply by the time period, etc. And the optimization period it tries to vary according to a number of criteria.
For a long time I was using it in the tester and on the demo - the result was the following: yes, there are periods when tuning to the latest history data may bring profit, but such a period will be necessarily followed by the period (I call it "atypical" for me) when this profit will be lost and result in 0 :) Although, it does not sink much... Less stable, something happens on TFs from 4H and higher, the global trends are not so volatile. But profitability is not so good there either....
So let's suggest that fellow "programmers for a price" bolt auto-optimisation to their EA and run it on at least 3 years of history.
So let's suggest that the fellow "programmers for the price" bolt the auto-optimization to their EA and run it on at least 3 years of history.
So let's suggest that fellow "programmers for a price" bolt auto-optimisation to their EA and run it on at least a 3 year history.
Figar0 This remark is not in your direction.
I mean the specific EA - which YuraZ offers us for discussion.
I would like to run it for 3 years on history.
I have repeatedly encountered the rake of optimization, because very often the selected parameters in the tester are just a fitting for the history.
And I have come to a conclusion - the longer the history the Expert Advisor can go without correction - the more stable it will behave in the future, because it has gone through more price behavior patterns. I'm not talking about perseptrons - that is a separate topic.
Well, let's suggest the fellow "programmers for the price" to screw the auto-optimization to their EA and run it on at least 3 years of history.
Figar0 This remark is not in your direction.
I mean the specific EA - which YuraZ offers us for discussion.
I would like to run it for 3 years on history.
I have repeatedly encountered the rake of optimization, because very often the selected parameters in the tester are just a fitting for the history.
And I have come to a conclusion - the longer the history the Expert Advisor can go without correction - the more stable it will behave in the future, because it has gone through more price behavior patterns. I'm not talking about perseptrons - that is a separate topic.
It's not all so fluffy.
It is also possible to open by hand, especially if the criteria are known, close the trawl and leave it to the Expert Advisor.
it is only a question of partial adaptivity. it is clear from the improved sitewide that it mainly works in the fifth wave, i.e. on corrections
and only along the trend
the trawl is adaptive
The parameters need to be optimized and that is the most unpleasant thing in this EA, but it works and that's good
It's clear that with some settings it works stable in the UP trend ... at other settings in DOWN
it has no sense to test it in three or fifteen years!
That is why I think it is impossible to show steady over a long period!
1 - during a trend change
2 - at a loss
Hence - optimization!
I do it the way I do it, probably many people do it the same way... i may be wrong.
the principles are
1 - do not take boundary values
i.e. we have 1 row with maximal profit - i won't take it
2-in an optimized sample I select for example the set that gives or the same profit
I have 2000-3000 thousand rows after optimization in the optimization results window
let's assume that 70 lines of the sample give the same profit and the same drawdown and the next 50 lines of the sample have a slightly worse profit and drawdown.
3- so we need to look for parameters in these samples
parameters are close enough in value, i.e. their small range has no effect
4 - I choose the average.
5 - the runs are in several stages
then I try to test one or two parameters in a sample, excluding the others, and so on with several resets
until I find something optimal
6 - after a loss, I think I need to reoptimize
7 - at a reversal - after the loss re-optimization is needed
(for example from last Tuesday 22.01.2007 the main trade was BAY) - hence re-optimization!
i may be wrong i didn't kill much time on the optimization - well, maybe 2 weeks while i was writing my 6th place winner
things started to go bad after the fall !
it's just that after the turn of 22.01 i just told him not to sell
I think he's good as an assistant, these are the signals of the one who stood on the Championship
2-in an optimized sample I select, for example, a set that gives or the same profit let's say
I have 2000-3000 thousand rows after optimization in the optimization results window
let's assume we have 70 lines of sample giving the same profit and the same drawdown the next sample of 50 lines slightly worse profit and drawdown
For example, SL or TP exceeding a certain size simply will not be reached.
In my opinion, it is not correct to take parameters from this area. It depends on the parameters, though, and on the strategy.
YuraZ, stupid poll - where's the code. i couldn't find it?
That question has already been answered: