Bad srategies

[Luigi]

I have been using the Builder now for several months. I have tried many combinations of parameters, different time periods and different instruments: I have not been able to create a good strategy in any possible way. Either I am missing something or it's something with the software. I am not talking about OOS periods or similar issues: I am talking about weak strategies which get big drawdowns, bad looking equity curves, etc. 

I am attaching a couple of my project files in the link. Please help

https://drive.google.com/drive/folders/1nXY8_x7hhhgMFd1nd_fIARkbtAJoGTlv?usp=sharing

Luigi

[MikeBryant]

There's no reason you shouldn't be able to develop good strategies with Builder. I took a look at your files. In file #1, I noticed you used the neural network indicator. You should keep in mind that this is basically a composite indicator. As such it adds an additional filter to your entry conditions, which will reduce the number of trades. That's probably why you're getting such a low number of trades. I would simply remove that and build again.

In file #2, the strategy you selected (#103) has a net profit of $103,000 with $18k drawdown, 67% winners, an average trade size of $1300, and better results in the OOS period than in training. How is that a bad strategy?

I agree that in file #3, there are no acceptable strategies. However, I can see that it only ran for 10 generations. While you used the build failure rules, it looks like you also had a termination rule after 10 generations, so there are no rebuilds; it simply stops after 10 generations. I would let it run for at least 50 generations, if not 100 or more. In cases where you're not getting good results after a few runs, it's usually just a matter of looking at the results, seeing what's not acceptable, and adjusting your build metrics accordingly. In some cases, it's as simple as just running another build. Because there's an inherent element of randomness to the process, you might find that good results are achieved on the next build, even with the same parameters. 

Also, I noticed that you didn't include Complexity as a metric. I would add the complexity metric with a small weight (e.g., 0.1 compared to a weight of 1.0 for the others) to your build objectives to make sure your strategies don't become overly complex.