Forum Replies Created
-
Posts
-
I also just noticed if you look at the source code of Petko’s EA for August vs mine there are some key differences.
“Additionally, when I compare the source code of my old EAs to the new ones, I notice the same pattern. I’m beginning to wonder if the presence of the “void” is causing the new EAs to not execute trades.”
Thank you, everyone, for your invaluable advice. I currently own two PCs, both equipped with Intel i7 CPUs. To ensure the safety of my collections, I have connected both of them to battery backups. This precaution allows me enough time to save my work when the power inevitably goes out. I primarily use my personal PC for collecting strategists, and I store all my collections on Google Drive. I also make use of the validator every month. In terms of strategy, I am mostly following Petko’s 12 EA system. Additionally, I utilize a VPS to run all my MT4/5 platforms.
At the moment, I am simulating the operation of EA Studio to evaluate its profitability in finding strategies over the past five months. I am well aware of the importance of separating “train” and “test” data to determine if a strategy has potential in the live market. Here are some specific questions I would like answered:
What should be the duration of my train set? Should it cover one year, two years, or six months?
How long should my validation set be?
Which symbols does EA Studio identify as having the most profitable strategies?
Is a 10%, 20%, or 30% out-of-sample allocation more suitable?
Should I enforce the use of a take-profit in EA Studio?
Which time frame yields better long-term profitability?
I have a notion that strategies with lower time frames should be updated more frequently. For instance, a 4-hour strategy may not require monthly updates, but a 5-minute strategy might benefit from weekly revisions.How consistent is EA Studio? If I run it multiple times using the same data, does it consistently identify strategies that generate profits in out-of-sample scenarios?
These are just a few examples of the questions I seek answers to.Furthermore, I understand that there are diminishing returns when attempting to discover the best settings due to the market’s inherent randomness. However, I firmly believe that maintaining a diversified portfolio with various time frames, symbols, and strategies is crucial for consistent gains.
Through numerous tests, I have already obtained answers to some of my inquiries. For instance, I have found that Walk Forward Validation does not necessarily correlate with more profitable strategies in out-of-sample data.
While performing all these tests, Express Generator could prove useful. However, when it comes to discovering new strategies for the upcoming month, EA Studio has proven more than sufficient.
I am confident that many of you possess insights into some of the questions I have raised. I eagerly await your thoughts and feedback.
Are there good instructions on how to run express-generator?
I know there are youtube videos.
I am a web developer so I know a bit of JS.
Just wondering what instructional material you found the most helpful.I have a live account with Trading.com and they do NOT allow hedging and they have FIFO rules just like all the other brokers. But I do like them for their tight spreads and MT5.
My favorite broker is Forex.com because they offer 2x more data than any other US broker that I know about.I have tested IG, Oanda, Trading.com, and Forex.com, and my EAs made more money on Forex.com.
I plan to do another broker test to see if this is still the case.Hi David, I have been experiencing the same thing. The EAs that I created for June have been losing so far. But I think they will recover maybe next month or even this month.
I have another question about Trading Session.
What time zone are the Trading Sesion filters using?
Like are they using the time zone of your data? Or are they using UTC time?
Hi Petko,
I ran the new EAs through the validator in EA Studio and on GBPUSD I get these results.
Am I missing something here?
