1) Post full test results, pictures won't tell anyone anything.
2) Describe the expert, what he uses and on what TFs he works (better post the expert too).
1) I have posted it.
2) I built it on the basis of my indicator, it does not work with history. Its working principle: it accumulates ticks in a reverse array, compresses them and then filters. The logical unit that generates trading signals is controlled based on parameters obtained from different filters. TS is on M5, but it is only for convenience of trade, because the principle of ticks processing on a zero bar.
My main question was if someone had encountered such a problem, as for help, I understand that here the inadequacy of the terminals themselves, how can you help me with this?
coaster писал(а)>> And where did the first two buys on the day of knowledge get lost? The rule of position opening must be the same? "Who" didn't want to (or couldn't) open according to the rules?
That's the question, where to? TC I just copied from one terminal to the other, so the logic of operation must have been exactly the same.
I say again, the difference in the experiment is only in the terminals, one is loaded with Alpari, the other with MQ, but the story for it is also with Alpari, naturally, and the TS is the same. And it has nothing to do with the market, it's a question for DCs.
The test is performed on very fresh data, if the terminal accesses the server, it is possible that the latest history is re-synchronized with the server.
Plus, the different quality of the test clearly shows a mismatch of quotes, the terminal has nothing to do...
If you want to share your quotes history, it's important to understand that your knowledge of the subject is not very good, so why not start with the stove? How do you download the Alpari history, how did you recalculate all the TFs, how did you copy etc.
I say again, the difference in the experiment is only in the terminals, one is loaded with Alpari, the other with MQ, but the story for it is also with Alpari, naturally, and the TS is the same. And it has nothing to do with the market, it's a matter for the DC.
Too cool. Are you sure it's not one of the variations below?
Murphy's Law If any mishap can happen, it will happen.
Consequences
Things are not as easy as they seem.
All work takes more time than you think.
Of all the troubles, the one whose damage is greater will happen.
If four causes of possible trouble can be eliminated beforehand, there will always be a fifth.
Events that are left to themselves tend to go from bad to worse.
As soon as you take on some work, there is another that has to be done even sooner.
Every decision breeds new problems.
Callaghan's comment on Murphy's Law Murphy was an optimist!
Renard's observation. There are times when everything succeeds. Don't be horrified, it will pass.
Laws of Cheesholm
Everything that can go bad, goes bad. Corollary:
Anything that can't go wrong goes wrong too.
When things go well, something must happen in the very near future. Consequence:
When things are going badly, they will go even worse in the very near future.
If you think things are getting better, then you have missed something.
Any suggestions people understand differently than the person making them. Implication:
Even if your explanation is so clear that it rules out any misinterpretation, there will still be someone who misinterprets you.
If you're sure what you're doing is universally praised, someone isn't going to like it.
Scott's Law It doesn't matter if something goes wrong. Perhaps it looks good.
Finagle's Laws
If the experiment is successful, something is amiss...
In any set of inputs, the most reliable value that requires no checking is wrong.
If the work fails, any attempt to salvage it will only make matters worse.
Ginsberg's theorem You can't win. You can't stay on your own. You can't even get out of the game.
Ehrman's comments
Before it gets better, the situation gets worse.
Who says it will improve?
Murphy's Law of Thermodynamics. Things get worse under pressure.
Everitt's Second Law of Thermodynamics The confusion in society is constantly increasing. Only by very hard work can it be reduced somewhat. However, this very attempt will lead to an increase in cumulative confusion.
Pudder's Laws
Everything that starts well, ends badly.
Everything that starts bad, ends worse.
Stockmire's theorem If it looks easy to do a job, it will certainly be difficult. If it looks hard, it is absolutely impossible to do it.
Zimmerga's Law of Systems Dynamics If you have already opened a can of worms, the only way to seal them again is to use a bigger can.
Emerson's law of infinite fall Beneath every abyss reveals another, even deeper one.
Mark Twain's law of stardom Once in a lifetime fortune knocks at every man's door, but in many cases the man is sitting in a nearby joint at the time and does not hear her knock.
Angela>>: ticks are accumulated into a reverse array, compressed and then filtered. Based on the parameters received from different filters, the logical block that generates trading signals is controlled. The TS stands on M5, but it is only for the convenience of trading, because the principle of ticks processing is on a zero bar.
In the tester, ticks are simulated.
The Expert Advisor test will only be adequate on a real account.
..principle of operation: ticks are accumulated in a reverse array, compressed and then filtered. The parameters received from the different filters are used to control the logic block, which generates the trading signals.
So, I understand that the inadequacy of the terminals themselves is there, how can you help me with this?
I will join the previous comrades in their opinion.
1. Strategies with bar opening controls work stably in the tester, in contrast to postics. The developers have repeatedly warned about the undesirability of tick strategies due to their increased sensitivity to history data.
2. A sophisticated Expert Advisor of an average author can show a swine flu growth chart instead of the balance.
3. With the process you have described, taking into account that ticks are generated by the tester from minute data, the slightest difference in the history can lead to a snowball effect. A slight change in the minute history leads to significant changes in the tick stream, and the specific methods of processing and filtering sensitively catch those differences and produce inconsistent data for the logic unit. Roughly speaking, the Expert Advisor works on noise.
4. I do not want to offend you, but when fighting with the terminal I always had to admit my own inadequacy in the first place. :))
Try to run the Expert Advisor again on both terminals, will the results be the same?
The wonders continue!
I ran the tests again, of course I didn't change anything from the previous ones, I only pressed the start buttons, the only difference was that now the testers worked with the terminals connected to the server and not in autonomous mode.
In the terminal with MQ, which was used for development and debugging of TC, the test was absolutely the same as the previous one:
№
Time
Type
Order
Volume
Price
S / L
T / P
Profit
Balance
1
2009.09.02 10:55
buy
1
0.10
1.61527
0.00000
0.00000
2
2009.09.02 12:40
close
1
0.10
1.61966
0.00000
0.00000
43.90
1043.90
3
2009.09.03 05:25
buy
2
0.10
1.62826
0.00000
0.00000
4
2009.09.03 08:10
close
2
0.10
1.63116
0.00000
0.00000
29.00
1072.90
5
2009.09.03 11:05
buy
3
0.10
1.63618
0.00000
0.00000
6
2009.09.03 12:20
close
3
0.10
1.63842
0.00000
0.00000
22.40
1095.30
7
2009.09.04 16:35
buy
4
0.10
1.63453
0.00000
0.00000
8
2009.09.04 18:10
close
4
0.10
1.63968
0.00000
0.00000
51.50
1146.80
9
2009.09.08 10:30
buy
5
0.10
1.64892
0.00000
0.00000
10
2009.09.08 12:10
close
5
0.10
1.65647
0.00000
0.00000
75.50
1222.30
11
2009.09.15 13:05
buy
6
0.10
1.64922
0.00000
0.00000
12
2009.09.15 14:39
close
6
0.10
1.64473
0.00000
0.00000
-44.90
1177.40
13
2009.09.15 18:10
buy
7
0.10
1.64386
0.00000
0.00000
14
2009.09.15 19:05
close
7
0.10
1.64628
0.00000
0.00000
24.20
1201.60
15
2009.09.16 17:10
buy
8
0.10
1.64978
0.00000
0.00000
16
2009.09.16 20:45
close
8
0.10
1.65010
0.00000
0.00000
3.20
1204.80
17
2009.09.18 11:00
buy
9
0.10
1.63481
0.00000
0.00000
18
2009.09.18 14:51
close
9
0.10
1.63479
0.00000
0.00000
-0.20
1204.60
19
2009.09.18 19:15
buy
10
0.10
1.62649
0.00000
0.00000
20
2009.09.18 22:57
close
10
0.10
1.62693
0.00000
0.00000
4.40
1209.00
21
2009.09.21 17:40
buy
11
0.10
1.62314
0.00000
0.00000
22
2009.09.22 08:40
close
11
0.10
1.62798
0.00000
0.00000
48.35
1257.35
23
2009.09.23 23:20
buy
12
0.10
1.63480
0.00000
0.00000
24
2009.09.24 02:15
close
12
0.10
1.63579
0.00000
0.00000
9.75
1267.10
25
2009.09.25 03:30
buy
13
0.10
1.59311
0.00000
0.00000
26
2009.09.25 06:45
close
13
0.10
1.60030
0.00000
0.00000
71.90
1339.00
On the terminal with Alpari the test is different, with the simulation quality, after connecting to the server, also became n/a, and the number of simulated ticks has changed:
Allow the use of cookies to log in to the MQL5.com website.
Please enable the necessary setting in your browser, otherwise you will not be able to log in.
You are signed in. Two-factor authentication has been disabled. Set it up again and scan a new QR code, or continue with two-factor authentication disabled.
If you want help, not communication:
1) Post full test results, pictures won't tell anyone anything.
2) Describe the expert, what he uses and on what TFs he works (better post the expert too).
1) I have posted it.
2) I built it on the basis of my indicator, it does not work with history. Its working principle: it accumulates ticks in a reverse array, compresses them and then filters. The logical unit that generates trading signals is controlled based on parameters obtained from different filters. TS is on M5, but it is only for convenience of trade, because the principle of ticks processing on a zero bar.
My main question was if someone had encountered such a problem, as for help, I understand that here the inadequacy of the terminals themselves, how can you help me with this?
And where did the first two buys on the day of knowledge get lost? The rule of position opening must be the same? "Who" didn't want to (or couldn't) open according to the rules?
That's the question, where to? TC I just copied from one terminal to the other, so the logic of operation must have been exactly the same.
I say again, the difference in the experiment is only in the terminals, one is loaded with Alpari, the other with MQ, but the story for it is also with Alpari, naturally, and the TS is the same. And it has nothing to do with the market, it's a question for DCs.
The test is performed on very fresh data, if the terminal accesses the server, it is possible that the latest history is re-synchronized with the server.
Plus, the different quality of the test clearly shows a mismatch of quotes, the terminal has nothing to do...
If you want to share your quotes history, it's important to understand that your knowledge of the subject is not very good, so why not start with the stove? How do you download the Alpari history, how did you recalculate all the TFs, how did you copy etc.
I say again, the difference in the experiment is only in the terminals, one is loaded with Alpari, the other with MQ, but the story for it is also with Alpari, naturally, and the TS is the same. And it has nothing to do with the market, it's a matter for the DC.
Too cool. Are you sure it's not one of the variations below?
Murphy's Law
If any mishap can happen, it will happen.
Consequences
Callaghan's comment on Murphy's Law
Murphy was an optimist!
Renard's observation.
There are times when everything succeeds. Don't be horrified, it will pass.
Laws of Cheesholm
Corollary:
Consequence:
Implication:
Scott's Law
It doesn't matter if something goes wrong. Perhaps it looks good.
Finagle's Laws
Ginsberg's theorem
You can't win. You can't stay on your own. You can't even get out of the game.
Ehrman's comments
Murphy's Law of Thermodynamics.
Things get worse under pressure.
Everitt's Second Law of Thermodynamics
The confusion in society is constantly increasing. Only by very hard work can it be reduced somewhat. However, this very attempt will lead to an increase in cumulative confusion.
Pudder's Laws
Stockmire's theorem
If it looks easy to do a job, it will certainly be difficult. If it looks hard, it is absolutely impossible to do it.
Zimmerga's Law of Systems Dynamics
If you have already opened a can of worms, the only way to seal them again is to use a bigger can.
Emerson's law of infinite fall
Beneath every abyss reveals another, even deeper one.
Mark Twain's law of stardom
Once in a lifetime fortune knocks at every man's door, but in many cases the man is sitting in a nearby joint at the time and does not hear her knock.
That's the question, where to? TC I just copied from one terminal to another, so the logic should have been exactly the same.
There's only one logic to it. "Pharaoh said so." :)
And so, don't be lazy, and compare the bars. At least in the report. There is such a point: bars in the story.
ticks are being accumulated in the reverse array
Oops... Ticks ?)
Hopefully at least a TF of 1 min? It's simple, post screenshots of visual tests of 2 of these passes, where there is divergence in trades...
Try to re-run Expert Advisor on both terminals, will the results be the same?
Try to re-run the Expert Advisor on both terminals, will the results be the same?
In the tester, ticks are simulated.
The Expert Advisor test will only be adequate on a real account.
Angela писал(а) >>
..principle of operation: ticks are accumulated in a reverse array, compressed and then filtered. The parameters received from the different filters are used to control the logic block, which generates the trading signals.
So, I understand that the inadequacy of the terminals themselves is there, how can you help me with this?
I will join the previous comrades in their opinion.
1. Strategies with bar opening controls work stably in the tester, in contrast to postics. The developers have repeatedly warned about the undesirability of tick strategies due to their increased sensitivity to history data.
2. A sophisticated Expert Advisor of an average author can show a swine flu growth chart instead of the balance.
3. With the process you have described, taking into account that ticks are generated by the tester from minute data, the slightest difference in the history can lead to a snowball effect. A slight change in the minute history leads to significant changes in the tick stream, and the specific methods of processing and filtering sensitively catch those differences and produce inconsistent data for the logic unit. Roughly speaking, the Expert Advisor works on noise.
4. I do not want to offend you, but when fighting with the terminal I always had to admit my own inadequacy in the first place. :))
Try to run the Expert Advisor again on both terminals, will the results be the same?
The wonders continue!
I ran the tests again, of course I didn't change anything from the previous ones, I only pressed the start buttons, the only difference was that now the testers worked with the terminals connected to the server and not in autonomous mode.
In the terminal with MQ, which was used for development and debugging of TC, the test was absolutely the same as the previous one:
On the terminal with Alpari the test is different, with the simulation quality, after connecting to the server, also became n/a, and the number of simulated ticks has changed:
This is getting more and more interesting!