From the idea to the actual bill. - page 2

 

Good results. What is your risk, i.e. the % of your balance that you use to open trades? Hint at least what area your Expert Advisor is from: based on built-in indicators, your own indicator, a neural network or some mathematical model of the market?

 
gpwr писал (а):

Good results. What is your risk, i.e. the % of your balance that you use to open trades? Hint at least what area your Expert Advisor is from: based on built-in indicators, your own indicator, a neural network or some mathematical model of the market?


Tests were done with constant lot, if you test with % risk of balance, the tester does not have enough digits. Let's say if the account is not limited to 1000 lots.

I can use it to test with risk, but it's just for show.
I have not got a single error in my code, not a single error in the logs, I just think who is trying to cheat who - myself or the tester.
That's why I am looking for different testing methods.

I have tried it with all parameters, both with open prices and checkpoints.

A number of indicators of my own.

Attached detailed reports.
Files:
mts.zip  20 kb
 

If the size of the trades was constant, the results are really amazing. In that case, the balance grows arithmetically over time. When using % of balance in trades, the growth will be geometric. That is, instead of Balance*(1+N*a) you will obtain Balance*(1+a)^N, where Balance=$10,000 - the initial balance (deposit), a - the averaged profit of each trade in % of the initial balance and N - the number of trades. Judging by your results: Balance*(1+N*a)=$110,000 in 2 years with the number of trades approximately N=1000. We obtain a=0.01. Then switching to trade size as % of current balance, your balance will increase to Balance*(1+a)^N=10,000*1.01^1000=$209 million in two years. Congratulations! Start trading on the real. Why ask people what they think of your results unless of course this is an advertisement.

 
That looks impressive! It would be great if the results of demo tests could be posted somewhere on the Internet for everyone to see in real time. People would be interested to see how such a unique EA works.

If it is not a secret, how much is the average profit per transaction in pips for each of the currencies? And what is the average number of trades per day?
 
solandr писал (а):
What, if not a secret, is the average profit per trade in pips for each of the currencies? And what is the average number of trades per day?

This question has not been investigated thoroughly yet. We should collect the history of deals in a demo or real account to be more precise.
Of course, we can use the Strategy Tester to calculate the above values, but we do not have time yet.

I have found the right market entry and now I am looking for the right exit.

I will write only once for everybody who is interested in the e-mail: hidden@hidden-system.ru and all questions about what and how it is organized to write there.
Do not breed irrelevant posts in the branch.

Answer the branch who knows whether to believe such impressive results from the tester. As I wrote above I put all my efforts into quality testing.
I just do not want to deceive myself. Of course I understand that without the source code it is difficult to say if I can trust the results or not, but perhaps there are some other limitations or specific testing approaches that are not described on forums. I understand that with the arrival of the 200 build and changed structure of tick generation in the tester it became much harder to write "grails", but that's how the case is.

Why are respected forum people like Rosh, Komposter, KimIV, administration and terminal developers keeping silent? I'm sure you know a lot of ways how else to test an EA, share it with me and the public. Many people make expert systems and all of them test them on the MT4 tester.
 
HIDDEN писал (а):
Reply in the thread who knows if such impressive reasonable results from the tester can be trusted. As I wrote above I have put all my efforts into quality testing.
I just do not want to deceive myself. Of course I understand that it is difficult to say if you can be trusted without the source code, but perhaps there are some other methods of limiting or specific approaches to testing that are not described on the forums yet.
Quite right, without the code you cannot tell what the catch is. There are a lot of ways of self-deception in the story test, and only one real judge is the actual work. If there are no trivialities in the code such as no stop-loss or peek-a-boo, everything looks very promising. Bet on the real and don't blow your brains out, either for yourself or for others. Why waste time coming up with different testing systems when the absolute perfect super tester is right there. "Practice is the criterion of truth".
 
HIDDEN писал (а):
solandr wrote (a):
What, if not a secret, is the average profit per trade in pips for each of the currencies? And what is the average number of trades per day?

It's a question that hasn't been investigated thoroughly yet. We should collect the history of deals in a demo or real account to be more precise.
Of course, we can use the Strategy Tester to calculate the above values, but we do not have time yet.

I have found the right market entry and now I am looking for the right exit.

I will write only once for everybody who is interested in the e-mail: hidden@hidden-system.ru and all questions about what and how it is organized to write there.
Do not breed irrelevant posts in the branch.

Answer the branch who knows whether to believe such impressive results from the tester. As I wrote above I put all my efforts into quality testing.
I just do not want to deceive myself. Of course I understand that without the source code it is difficult to say if I can trust the results or not, but perhaps there are some other limitations or specific testing approaches that are not described on forums. I understand that with the arrival of the 200 build and changed structure of tick generation in the tester it became much harder to write "grails", but that's how the case is.

Why are respected forum people like Rosh, Komposter, KimIV, administration and terminal developers keeping silent? I'm sure you know a lot of ways how else to test an EA, share it with me and the public. Many people make expert systems and all of them test them on the MT4 tester.
 
HIDDEN:


Why are respected forum people like Rosh, Komposter, KimIV, administration and terminal developers keeping quiet, you probably know a lot of ways to test an EA, share it with me and the public. Many people make expert systems and all of them test them on the MT4 tester.
If your Expert Advisor uses self-made indicators, you could lay out the Expert Advisor code without these indicators. Perhaps, the glitch in the order management system? But it is impossible to say anything without a blind eye. Looking at the tests, we can see that even if a stop loss is set, it obviously is not fixed or there is a market exit before the loss and there is nothing else to be understood.

For example, in the old builds, there was a tester glitch when Take Profit and Stop Loss overlapped on the same bar and the tester closed the deal at Take. Which build of MT4 was used for testing?
 

Vpryntsipa realistic, 50% per month a fixed lot it is not so fantastic. Confusing only that the last 3 months on which the optimized Expert Advisor could take into account more than a year and a half of the past, and did not affect the roof balance. Is the opening price test the same as the test for all ticks?

 
HIDDEN писал (а):
solandr wrote (a):
How much, if not a secret, is the average profit per trade in pips for each of the currencies? And what is the average number of trades per day?

I haven't checked it thoroughly yet. We should collect the history of deals in a demo or real account to be more precise.
Of course, we can use the Strategy Tester to calculate the above values, but we do not have time yet.

I have found the right market entry and now I am looking for the right exit.

I will write only once for everybody who is interested my E-mail: hidden@hidden-system.ru and all questions about what and how it is organized to write there.
Do not breed irrelevant posts in the branch.

Answer the branch who knows whether to believe such impressive results from the tester. As I wrote above I put all my efforts into quality testing.
I just do not want to deceive myself. Of course I understand that without the source code it is difficult to say if I can trust the results or not, but perhaps there are some other limitations or specific testing approaches that are not described on forums. I understand that with the arrival of the 200 build and changed structure of tick generation in the tester it became much harder to write "grails", but that's how the case is.

Why are respected forum people like Rosh, Komposter, KimIV, administration and terminal developers keeping silent? I'm sure you know a lot of ways how else to test an EA, share it with me and the public. Many people make expert systems and all of them test them in MT4 tester.
You can also run it through a visual tester.
Reason: