The miracles continue! - page 8
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
Hello Angela.
Which data type do you use to handle double or int ticks ? And if you convert it to integer type, how do you do it?
The point is that the terminal very often makes errors in the last digit during operations with double type.
If you compare two equal variables, for example, like this (the numbers don't have to be like this):
double a=1.5555;
double b=1.5555;
if (a-b>0) Print ("a>b");
else if (a-b<0) Print ("a<b");
else Print ("a=b");
then for some a and b equal to each other, the result can be a>b or a<b, though a=b should be.
Preliminary normalization doesn't give the right result.
Errors occur when comparing, subtracting, dividing and determining the remainder of a division. I didn't check the rest of them - the results I found are enough:)))) I can't say how these errors depend on concrete numbers (I was too lazy to find it out). There is a probability it is random, i.e. it may or may not happen with the same data. One thing I can tell you for sure: an error occurs at the last digit.
If your Expert Advisor uses operations of type double and there are many of them, the error is gradually accumulating.
This may be the reason.
PS: By the way, I found this error on Alpari terminal. I have not checked it on other broker terminals, but maybe it is there too.
I have faced with these problems, but in this case I have a different kind of problem, on one terminal it is stable, on the other it is not.
No, I haven't.
Look in my personal message.
There is also an option.
When connecting to the server, the terminal automatically downloads data on open charts and adds it to the history. If there is data in the history for the same period that it downloaded when connecting, the data in the history files is replaced with the new data.
Perhaps, the terminal MQ is downloading them from another place than the Alpari website.
You can find it out this way. Connect the Alpari terminal. Download the quote archive. Reboot the terminal. Launch the tester. When testing is over we need to block connection of the terminal to the server.
Go to proxy server settings (Service-Settings-Server-Proxy...) write a non-existing proxy (e.g. 192.168.100.100:10000) in "Server" field. Click on "OK". In tab "Server" tick "Enable proxy server". Restart terminal. After that terminal can't connect to server.
Similarly for MQ terminal go to proxy server settings (Service-Settings-Server-Proxy...) write non-existing proxy (for example 192.168.100.100:10000) in "Server" line. Click on "OK". In tab "Server" tick "Enable proxy". Close the MQ terminal.
Then once again start testing in Alpari and after its completion completely copy the "history" and "tester" folders from the Alpari terminal to the MQ terminal (MQ should be closed).
Now start the MQ terminal and test the Expert Advisor.
If after that there are no discrepancies, then we have downloaded the history from different places.
If there are still discrepancies, then we can clearly say that there is something wrong with the Alpari terminal.
By the way, the Alpari history is very different from other brokerage companies and it is not only about 5 digits and floating spread. The data starts to differ around April 2004.
I have an Expert Advisor that shows stable growth in the tester from 1999 to 2006. Since 2007 the Expert Advisor has also been steadily losing money. I have tested it on several brokerage companies. The situation is almost the same everywhere. Until October 2006 a stable growth followed by up to January 2007 the balance remained almost at the same level and then it stably went down the drain.
And in Alpari terminal the stable loss has started since April 2004.
A similar question was also asked here https://www.mql5.com/ru/forum/112852
I have an EA that has been showing stable growth in the tester from 1999 to 2006. Since 2007 the Expert Advisor has also been steadily losing money. I have tested it on several brokerage companies. The situation is almost the same everywhere. Until October 2006 stable growth, further till January 2007 balance remained on the same level, and then it stably drains.
There is such a thing. In my opinion, the reason is that robots have been on the market since 2006 and every year there are more of them. The result is that classic TA no longer works. I'm sure many would disagree, but I'm sure that everyone without exception has encountered a situation in the market where the price goes against all TA laws.
Was everyone happy before 2006 because the laws worked? Or were most of the "experts" just as successful before and after?
... since 2006 robots have been on the market ...
So robots were only born in 2006? :)))
Well, well...
And computers were probably born in '05?
.... As a result - classic TA doesn't work anymore. ...
You'd better be careful with such a harsh statement, because if it doesn't work for you it doesn't mean that it doesn't work at all.
For example, I struggled as much as I could, but I couldn't get any practical use out of neural networks. But it is not the basis for a conclusion that the NS do not work.
Batter's got it working and it's working pretty good. And I'm guessing he's not the only one.
Angela.
I had some thoughts on your (our!))) fuss. And then I thought, - it's all interesting, of course, but not worth bothering with. You seem to have come to the same conclusions.
The stability of TC is more interesting than this problem. Even so - it is very good that there is such a problem - you can check TC for lousiness. (Yes, you can find positives in everything.))
Z.U.
At first I thought I accidentally came to a wrong branch - deja vu, some people who have problems with TA. True, they assure me that it's TA who has problems.))). Typical loser logic.
You guys have your own thread. Emotions are getting the better of you? (For God's sake, don't answer - the question is rhetorical!))
Angela.
I had some thoughts on your (ours!))) fad. And then I thought, - it's all interesting, of course, but not worth bothering with. You seem to have come to the same conclusions.
The stability of TC is more interesting than this problem. Even so - it is very good that there is such a problem - you can check TC for lousiness. (Yes, you can find positives in everything.))
Z.U.
At first thought that accidentally got on a wrong branch - deja vu, some people who have problems with TA. True, they assure you that it is the TA problems.))). Typical loser logic.
You guys have your own thread. Emotions are getting the better of you? (For God's sake, don't answer - the question is rhetorical!))
Yes, I've actually said it all, and I don't do it any more. The problem has been narrowed down to the Alpari terminal (although there are probably other brokerage companies, where the problem will be just as acute). Once again, to clarify the last experiment, I work in the autonomous mode, combing the history on the Alpari terminal, the simulation is 90% correct, no chart mismatch errors, I transfer folders of quotes history from the Alpari terminal to the MQ terminal. On the MQ terminal I get the same and stable results as on the previous history, the modelling quality is also 90%. On the Alpari terminal it is still not stable and does not correspond to MQ. On MIG terminal it is fully compliant with MQ, and even better by all indications. Operation without connection to the server, neither demo accounts nor any other reasons may influence the results.
Conclusion: Different brokerage companies have different settings in terminals, and it greatly affects results for tick-sensitive strategies. Apparently this problem is common and is always present on all terminals, but it appears to a different degree depending on settings, TS sensitivity etc. In my case the transfer of TS from MQ terminal to Alpari terminal led to the fact that 60% of profit was lost, with a less sensitive strategy the losses may be 10%-20%, probably most people just did not do a thorough research in this direction and did not pay attention to small losses and differences, considering it simply related to market changes. Apparently often when someone buys a TS and it shows completely different results than advertised, this problem is also present, not all sellers are badasses that try to sell a wrong one, just another broker terminal, though everything in this world is relative.
I am not dealing with this problem any more, I use another strategy, they are born fast as kittens, I lost count of how many I have done in last 2 months alone. But for now satisfying me to 3 indicators (drawdown not more than 6%, profit - not less than 4, the number of transactions not less than 40 per month, and TS did not require periodic reoptimization), - have not received, one of the indicators out of the norm, but I hope I will.
I hope I will get it. I may consider this topic closed, anyway all secrets of black box will never be known to me.
We can consider the topic closed, all the secrets of the black box will never be revealed to us anyway.
>> Amen.