October 28, 2019 at 15:38 #25228
Good day, Petko.
With the EA’s that you generate in your courses, you mentioned that you generate the average parameter values for different brokers that you used in generating the strategies so that the EA’s would have the similar performance for many brokers (although not all). Given that,
a. What are the things that we need to do inorder to check whether the EA’s would perform well with our broker? I’m assuming that we upload the EA’s onto the softwares (EAS/FSB), import the historical data of our broker, and see whether there is a (huge) difference or not in the performance. Am I right by saying this? Are there some other things that we need to consider aside from what I’ve mentioned?
b. Suppose the EA’s are not giving a similar performance with the historical data of our broker, what do we need to do inorder to arrive at a similar profit line relative to your backtest when you initially generated the strategies? Should we optimize the EA’s?
c. May I know which brokers you used in generating the strategies, and thereafter, coming up with an average value of the parameters found within each strategy?
Jay-ROctober 30, 2019 at 7:36 #25480Bart MeijrinkParticipant
I had the very same question to Petko during one of the 1:1 sessions with him( I did a couple and they were all very useful).
One of the ways is exactly what you said – import the data from the other broker and check if the equity line is similar or not. It you see totally different results, you can really do a lot. If you optimize it, there is a risk of over-optimization.
If you want to take average values for the parameters (what Petko did) you need to switch between the different data sources, optimize with round numbers and acceptance criteria, and take the average values.
Let’s say the SL with one of the brokers is 80, with the other is 70, and with the third is 40. You will take average SL of 66.
That is a lot of work that I tried as well, and I figured out that I really do not need to do it for myself because I use just 2 different brokers, and I generate separate EAs for each one.
Petko does that because he provides many EAs in his courses, and he aims at providing EAs that are suitable for everyone (even that is quite impossible according to him). However, I am very impressed that he does it anyway.October 30, 2019 at 13:41 #25496
I see. Yes, that’s what I guessed regarding optimizing the strategies—having the risk of over-optimization.
With regards to what you’ve mentioned that I can still do a lot more things if I get to see totally different results, would you mind expounding on that? I’m interested to hear what your thoughts on that.
Yes, what Petko is doing is very commendable when it comes generating the EA’s for the courses that could be potentially used for different brokers.
Kind regards,November 1, 2019 at 10:47 #25692Bart MeijrinkParticipant
My bad, I wanted to write you can NOT really do a lot. 🙂 Just missed the “not”.
Yes, if the results are different, it means that Historical data is different and the brokers have different quotes/prices at any moment. So you can not do a lot on that…
Kind regards,November 1, 2019 at 12:49 #25700
It’s alright Bart. At least, the good news is it’s lesser work to be done. Lol. I’m kidding. I love working =)
Cheers,November 2, 2019 at 19:34 #25838Petko AleksandrovKeymaster
Keep up the hard work, guys. That is the only way to success.
Many people think that Algo trading is lazy work, we put one EA in the account, and it makes us constantly profit. Well…that is not really the reality. But when the hard work is done, the results are coming.
You must be logged in to reply to this topic. Login here