Tagged: Common Acceptance Criteria
April 16, 2020 at 19:15 #45343vietcuong_gfMember
Hello Petko and trading friends,
After some time of working around with EA Studio, I’m impressive with its capacity of creating nearly infinite numbers of Expert Advisors. However, the key point is we need to have a good combination of common acceptance criteria so that it can generate enough good Expert Advisors in not too long time. That’s the point where Petko’s experience comes into play to help us. I suppose many of you are having the same questions below and I hope Petko will help us with it.
1a) What are your frequently used Common Acceptance Criteria and their frequently used values?
(I know we may have many combinations of common acceptance criteria such as minimum count of trades, minimum profit factor and it depends on the taste of each trader. The more acceptance criteria, the longer it may take to find a strategy. However, I believe after a long time of using EA Studio you find a good balance point, where we can set common acceptance criteria to generate good strategy but don’t cost much time for weeks or months to generate.
1b) In a course, you have set the following Common Acceptance Criteria: 1. Minimum system quality number: 0,98; 2. Minimum r-squared: 60; 3. Minimum profit factor: 1.2; 4. Minimum count of trades: 350. Are these criteria often used by you and in accordance to your experience of trial and error, are these criteria good enough for EA Studio to generate decent strategies?
1c) Do you use any other combination of Common Acceptance Criteria in addition to the above frequently used ones (mentioned in 1a, 1b)? If yes, may you please share your experience with us?
2. Usually how many Calculated Strategies (I mean the total number Reactor has generated, not the strategies put into Collection) have you had each time?
(Initially, I wanted to ask how long do you run Reactor each time, however, I think this question is useless since it depends on the power of each one’s computer. If your computer is so strong, you only need one night to generate totally 3 million strategies and that big number of strategies could provide decent final good strategy to put into Collections. So I’d better ask how many Calculated Strategies you have for each run so that I can calculate the needed time to run).
I have some more questions, but I think it will make my post too long. Furthermore, I want to have more posts so that Petko will give me discount for his course as promised. So let me post another question on another post.
Thank you Petko and trading friends in advance for answering me.
April 20, 2020 at 8:33 #45403Petko AleksandrovKeymaster
Hello, and welcome to the Forum!
Glad to hear you have many questions. This means that you are getting into the details.
1. It is important to find the balance of acceptance criteria for your Historical data. The number of bar matters here. If you have a small number of bars and you set strict acceptance criteria, you will not see any strategies in the collection.
Yes, these are criteria where I follow and do some changes if it is needed.
There is a Topic in the Forum about the Acceptance criteria, and many traders shared their way.
Simply, start with any acceptance criteria that you find logical, and you will see if you need to stricter or not.
It is better to have more strategies in the collection from where you can filter again.
2. When I generate strategies, I want to have at least 10 strategies that passed all the criteria and robustness tools.
Depends on the asset and what the plan is for the trading, sometimes I go for above 30 or 40. Just leave it longer or loosen the criteria.
- You must be logged in to reply to this topic.