Discussion of article "Self-adapting algorithm (Part IV): Additional functionality and tests" - page 4
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
Here, I'll help you see the error of your ways.
I didn't read the article, don't even know what it's about :) but just looked at the graphs and stats for 10 minutes.
I want to ask a few questions first.
1. What server was used for testing. If it is on the MetaQuotes-Demo server, then there the quotes with a very low spread and no commission. Sometimes the spread is reduced to zero and even below zero. You can check it.
2. All ticks mode, do you mean "Every tick" mode ? If yes, these are simulated ticks, not real ones.
Try testing in "Every tick based on real ticks" mode .
But that's all right.
The main problem you have is the maximum drawdown. If you compare it with monthly growth, there is a big difference between them, and when you try to increase profitability, you will get Stop Out.
Usually, when brokerage companies or investors are looking for traders to manage their funds, one of the main conditions is this: for 3 months of trading, get a monthly gain of at least 5%, with a total maximum drawdown of no more than 10%.
This is of course not a standard for everyone. But roughly like this.
Look what you have going on.
Forex quotes from admiral real, appl from stock account, moex custom symbols downloaded from finam. I know how to test, I know how real ticks differ from synthetic ones. In the article, tests in ohlc mode, the results are not very different from ticks, I specially make algorithms so that the differences were minimal, inside the bar is done at a minimum, timeframe m1.
You don't want to hear criticism?
You need to spend only 5 minutes to test at another broker, on real ticks and show the results starting from 01.01.2020 and show right here.
And if you say that it works without optimisation on all symbols, you could show the test results (table) in the mode "On all symbols from Market Watch", selecting 40-50 pairs.
That would be interesting for everyone.
You don't want to hear criticism?
You only need to spend 5 minutes to test at another broker, on real ticks and show the results starting from 01.01.2020 and show it right here.
And if you say that it works without optimisation on all symbols, you could show the test results (table) in the mode "On all symbols from Market Watch", selecting 40-50 pairs.
That would be interesting for everyone.
I don't want to interfere, but since you are talking about modelling quality and have uploaded this information into the working areas of your brain, can I ask you what it means: Testing on real ticks and history quality 4% in the results?
I don't want to interfere but since you are talking about the quality of modelling and have uploaded this information to the working areas of the brain, can I ask you what it means: Testing on real ticks and in the results the quality of modelling is only 4%
You should know that the possibility to test on real ticks (on MT5) appeared about 4 years ago (I don't remember exactly). I am talking about the Every tick based on real ticks mode .
And if the tick history is not enough, simulated ticks are used. Of course, not all brokers have it, but mostly the quality of real ticks is 100%.
You don't want to hear criticism?
You only need to spend 5 minutes to test at another broker, on real ticks and show the results starting from 01.01.2020 and show it right here.
And if you say that it works without optimisation on all symbols, you could show the test results (table) in the mode "On all symbols from Market Watch", selecting 40-50 pairs.
That would be interesting for everyone.
Automated experts should always have their advantages over live traders. The approach is correct and it only remains to wish good luck in finding the parameters of the machine.
So why should I test with another when I have already tested on ticks from a real account? I can show a month test comparing ohlc and on ticks ha one instrument. And it's not 5 minutes. One run for 2 years on 28 instruments takes a month of real real time. I want to criticise, but not the yield charts, but the methods.
You show a lot of charts for several years.
If you want your TS to be evaluated, show at least the test results NOT on ticks (Every Tick), but on real ticks ( Every tick based on real ticks), at least for this year, for 3-4 pairs.
Is it also difficult ?
Here you are showing a lot of graphs, over several years.
If you want your TS to be evaluated, show at least the results of the test NOT on ticks (Every Tick), but on real ticks ( Every tick based on real ticks), at least for this year, for 3-4 pairs.
Is it also difficult ?
You're not reading what I'm writing correctly, it seems. I don't understand where you got the idea that I tested on simulated ticks. Naturally I tested to make sure that the results match on ohlc with the results on real ticks. Moreover, the robot is originally made so that everything coincides. But okay, I will show later the comparison of ohlc mode and real ticks as soon as I make the tests. I don't understand why though... I'm not proving anything, just showing the methodology and how it works. I don't sell anything to anyone, who cares how my implementation of the algorithm works.
Dear Maxim !
Keep on going. The topic of self-tuning Expert Advisors is very interesting to me. I tried to implement it myself, but I can't reach it. OOP is already hard for me. And do not listen to well-wishers. They would like to blame their ineptitude on someone else - this is not the way it is, that is not the way it is. They can't run it on their own history of ticks or in real life. They are experienced. Good luck!