Create and Test Forex Strategies
You are not logged in. Please login or register.
Forex Software → Express Generator → Randomness in Strategy Generation
If two Strategy Collection Generators are run with identical settings and 6-hour processing time, one would assume the results against the performance metrics would be similar, however, the results are very different. How long does the Strategy Collection Generator need to be run to neutralize the randomness of the results?
Each instance of the generator when executed tries to find strategies with randomization of logical indicators and different periods, even if you run 5000 instances with the same generation configuration, they will always be randomized, but each file, Jason, will save its own collection to obtain linearly equal strategies, I do the following I import all collections in ea studio through the robot validator a unique test automatically the correlated strategies are turned off I could do this in the express itself but I still don't know how to do it
Even if one were to fix the stop loss, take profit (e.g. max and min the same value), and restrict the available indicators to a single indicator, the results sometimes differ dramatically. Maybe it is just me, but the complete randomness of the strategy generation begs the question, could a more disciplined method of backtesting reveal even better strategies?
If you are passing the same Studio parameters to the Express generator and then doing the back test you will see that the Express generator parameters will always be different because even with the fixed utp stop the Express generator closes the operations without touching the stop or the fixed TP because It doesn't have this function, is that what you're talking about or not?
Hey dlouisbriggs and gabdecsters,
Just chiming in here with my two cents. As gabdecsters already pointed out, the vast parameter space that the Express Generator navigates through is indeed colossal. Here's why the uniqueness in each run is not just expected, but also a feature:
Randomness & Complexity: With such a huge array of possible combinations, even 5000 instances can yield vastly different results. This is because the parameter space isn't just large; it's practically infinite when you consider all the variables and their potential values.
Genetic Algorithms & Local Maximums: While genetic evolution is an alternative method used by other platforms, it's not a silver bullet. Scientifically speaking, a local maximum is a point in the search space where the solution is better than the surrounding points, but it's not the best possible solution overall. Genetic algorithms can sometimes get stuck in these local maximums, failing to find the absolute best strategy.
Personal Experience: Since I started in 2008, I've tested a multitude of platforms, and I can confidently say that EA Studio and Express Generator have consistently delivered the best strategies for me. The Express Generator, being essentially a turbocharged version of EA Studio, offers the same quality results but at a much faster pace.
In conclusion, embrace the diversity of outcomes as a testament to the power and sophistication of the Express Generator and EA Studio. Keep experimenting, and remember that the search for the perfect strategy is a journey, not a destination.
Good luck, and may the odds be ever in your favor!
Forex Software → Express Generator → Randomness in Strategy Generation
Powered by PunBB, supported by Informer Technologies, Inc.