Strategy tester.
All in all yes.
No tester (like the previous one) ...
And as always, the developers know better than we do what we need (and it's sad :(()
I did not find anything in reports for example about accounting for commissions and spreads.
A lot of comments ... But I won't annoy the owners ...
There are some strange things that look like glitches.
Since my Expert Advisor categorically refused to be tested, I ran a MACD Sample from the distribution kit on USDCHF,D1 from 2003.07.18 up to now.
Model - All ticks, without optimisation.
The result is unclear.
1. the balance represents a perfect straight line going strictly upwards (grail, nothing else).
2. Results of trades are unclear
No tester (like the previous one) ...
And as always, the developers know better than we do what we need (and it's sad :(()
I did not find anything in reports for example about accounting for commissions and spreads.
A lot of comments ... But I won't annoy the owners ...
There are some strange things that look like glitches.
Since my Expert Advisor categorically refused to be tested, I ran a MACD Sample from the distribution kit on USDCHF,D1 from 2003.07.18 up to now.
Model - All ticks, without optimisation.
The result is unclear.
1. the balance represents a perfect straight line going strictly upwards (grail, nothing else).
2. Results of trades are unclear
1 2003.07.18 00:00 sell 1.00 1.3695 0.0000 1.3665 2 2003.07.18 00:00 t/p 1.00 1.3665 0.0000 1.3665 220.95 10220.95 3 2003.07.18 00:00 sell 2 1.00 1.3573 0.0000 1.3543 4 2003.07.21 04:22 t/p 2 1.00 1.3543 0.0000 1.3543 217.00 10437.95 5 2003.09.02 00:00 sell 3 1.00 1.4015 0.0000 1.3985 6 2003.09.05 10:55 t/p 3 1.00 1.3985 0.0000 1.3985 -2.88 10435.07 7 2004.02.19 00:00 buy 4 1.00 1.2474 0.0000 1.2504 8 2004.02.20 00:00 t/p 4 1.00 1.2504 0.0000 1.2504 603.79 11038.86 ....................................
Everywhere TakeProfit is triggered after 30 pips, but the trade results are very different for some reason (220.95, 217.00, -2.88, 603.79).
3. Further, for some reason the exact same deals come in bunches
.......................... 19 2004.04.08 00:00 sell 10 1.00 1.2767 0.0000 1.2737 20 2004.04.08:44 t/p 10 1.00 1.2737 0.0000 1.2737 140.55 12285.68 21 2004.04.0808 08:44 sell 11 1.00 1.2757 0.0000 1.2727 22 2004.04.08 08:45 t/p 11 1.00 1.2727 0.0000 1.2727 235.76 12521.44 23 2004.04.08:45 sell 12 1.00 1.2757 0.0000 1.2727 24 2004.04.04.08:46 t/p 12 1.00 1.2727 0.0000 1.2727 235.76 12757.20 25 2004.04.08:46 sell 13 1.00 1.2757 0.0000 1.2727 26 2004.04.08:47 t/p 13 1.00 1.2727 0.0000 1.2727 235.76 12992.96 27 2004.04.08 08:47 sell 14 1.00 1.2757 0.0000 1.2727 28 2004.04.08:48 t/p 14 1.00 1.2727 0.0000 1.2727 235.76 13228.72 .............................
And they go one after another, but their opening and closing price is the same.
Sell at 1.2725 and t/p at 1.2727...
At least you waited for the result :)) I have to wait for the evening :) And looking at what he draws and remembering how the expert traded in real time I have some doubts - what he is testing :)
And everything works for me, a test for a year for 15 min with a simulation of 1 min. Runs in 45 seconds. And is almost identical to MT3. The difference is insignificant, a few pips due to the fact that the minute candles are priced at (asc+bid)/2 . In this regard, we would like to be able to convert from (ask+bid)/2, to bid prices when importing history.
Gentlemen, so that you don't have any doubts, do a few things.
1. after generation of the sequence (just wait until the stop button becomes active) open the corresponding file offline - it is marked in the list with a blue icon with the letter G. and set the display to Japanese candlestick mode. this will inform you very clearly about the development of the bar
2. insert TOHLCV output in the beginning of the expert file for a more accurate analysis.
3. output the results of checking for entry and exit signals at least to the journal.
4. when deinitializing the EA print the entire history and look at commissions, swaps and profits. calculate the results. compare it with what the tester says. think about it.
and then make a complaint.
1. after generation of the sequence (just wait until the stop button becomes active) open the corresponding file offline - it is marked in the list with a blue icon with the letter G. and set the display to Japanese candlestick mode. this will inform you very clearly about the development of the bar
2. insert TOHLCV output in the beginning of the expert file for a more accurate analysis.
3. output the results of checking for entry and exit signals at least to the journal.
4. when deinitializing the EA print the entire history and look at commissions, swaps and profits. calculate the results. compare it with what the tester says. think about it.
and then make a complaint.
Does the optimisation work or not?
You set the minimum, maximum value and step. It seems to optimise something. Where are the results?
At maximum profitability and in general what result was chosen ?
You set the minimum, maximum value and step. It seems to optimise something. Where are the results?
At maximum profitability and in general what result was chosen ?
Gentlemen, so that you don't have any doubts, do a few things. <br / translate="no"> .......
Compare the results with what the tester said.
And then make a claim.
Compare the results with what the tester said.
And then make a claim.
The claim was about the speed. There were only vague doubts about the correctness of the test :))
As for the speed:
I have removed from the Expert Advisor.
1. installation and removal of graphic objects.
2. sleep(1000)
3. Working with global variables.
It turned out that one of them was responsible for the lags.
Everything seems to work fine now.
However, I still have not managed to find out how to optimize it. I will make one pass and that will be all.
Or
Optimization stopped
There were 1 passes done during optimization, 1 results have been discarded as insignificant
Or
Optimization stopped
test3 stopped : test limit 'consecutive loss=1000' reached
sorry! duplicate entrances at the same price! and their times are different
As for speed:<br / translate="no"> I have removed from the expert -
1. installing and removing graphical objects.
2. sleep(1000)
1. installing and removing graphical objects.
2. sleep(1000)
test3 stopped : test limit 'consecutive loss=1000' reached
3. Working with global variables.
It turned out that one of them was responsible for the lag.
Everything started to work quite smart now.
However, I still cannot figure out how to optimize it. I will make one pass and that will be all.
Or
Optimization stopped
There were 1 passes done during optimization, 1 results have been discarded as insignificant
Or
Optimization stopped
test3 stopped : test limit 'consecutive loss=1000' reached
It turned out that one of them was responsible for the lag.
Everything started to work quite smart now.
However, I still cannot figure out how to optimize it. I will make one pass and that will be all.
Or
Optimization stopped
There were 1 passes done during optimization, 1 results have been discarded as insignificant
Or
Optimization stopped
test3 stopped : test limit 'consecutive loss=1000' reached
Read sometimes our english forum as well. the questions are similar to "Expert Advisor optimization".
You have to set the start and end values and the step to change the parameters to be optimised.
The number of optimization steps depends on the number of parameters to be changed and the number of change steps.
What do you do not understand from these messages?
I don't understand why the optimization is stopped in this case
Optimization stopped
There were 1 passes done during optimization, 1 results have been discarded as insignificant
Parameters are set, steps and end values are of course the same.
Optimization stopped
There were 1 passes done during optimization, 1 results have been discarded as insignificant
Parameters are set, steps and end values are of course the same.
I don't understand why optimization stopped in this case<br/ translate="no"> Optimization stopped
There were 1 passes done during optimization, 1 results have been discarded as insignificant
Parameters are set, steps and end values are of course the same.
There were 1 passes done during optimization, 1 results have been discarded as insignificant
Parameters are set, steps and end values are of course the same.
Right-click in Optimization Results and uncheck "Skip Useless Results". In this way you will see all the results obtained during the next passes without discarding the completely failed ones.
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
And speed tests for different languages were they checked in the tester? I think the difference in speed tens of thousands of times.
For example, the tester I wrote does a half year history test on minutes in about 10 SECONDS. The same tester is already counting 3 HOURS and is about to finish counting the FIRST MONTH! Of course this is using fast method :) Well, if you need to optimise triple parameters, then .... - go to in a couple of years. Possibility to save intermediate results is not provided there.
As a whole there is a beautiful but useless toy :(