My recent journey down the path of testing and rapidly prototyping ideas has led me to find some pretty advanced software(s) that I have incorporated into my development process. I own and use licenses to both Strategyquant and Build Alpha and want to take the time to write a quick review of my experience with both and how they have helped me inch ever closer to my goal of full-time trading! I first came across Strategyquant and thought it was one of a kind. It is software that uses a genetic algorithm to create strategies without the user doing any programming and it also generates exported code for Tradestation (my trading platform mentioned in previous post) and a few other platforms. SQ has some great advantages and some very advanced features. I personally like searching through the strategies it creates and easily selecting good ones with combinations of signals I would have eventually found (and some not) on my own. This has greatly sped up my testing and development time. I should also note I have had more than a few unpleasant difficulties with SQ. It seems to produce strategies that are great in the platform but when I export the code to my platform they seem to fall apart and are nothing like the backtests I created. The other issues I have had are the strategies are often curve fit or overfit and fail miserably in live trading. I know it is difficult to find good strategies but far too many slip through the cracks and negate a lot of the positive performance I have seen using the software.
PROS: Fast way to build strategies and generate code
CONS: Tough to reproduce results and lots of overfitting
I then found Build Alpha a few months later and originally fell in love with it because I could actually import strategies I built with Strategyquant right into the Build Alpha software. Build Alpha has a very advanced set of tools to identify overfitting and determine if a strategy is robust (will do well in live trading or not) compared to Strategyquant. Most of the robustness tests I have never heard of actually but are now a part of my process and extremely useful. “You don’t know what you don’t know” I suppose. However, the fact that Build Alpha can also build strategies was just a bonus at the time of purchase and it turns out it runs much faster than Strategyquant and I have also not had any trouble reproducing the results with the Build Alpha generated code as I do at times with Strategyquant. The one downfall of Build Alpha is the built in data is for over 100 markets but does not contain any stocks data. It is primarily futures, FX, and ETFs. I primarily trade stocks and futures so it is a bit annoying importing stock data whenever I want to test but Dave (creator of Build Alpha) has created a custom script to make it as easy as possible to import but it still would be better if stock data was just included. If you use the software tell him to include stock data so he prioritizes it 🙂
PROS: Fast way to build strategies and generate code, much better robustness testing environment
CONS: Data import for stocks
Overall they are both wonderful pieces of software and have both made my life much easier than if I were to do all of what I do now on my own. I actually am trading strategies from both platforms and find myself reliant on the Build Alpha tests even for strategies I generate from Strategyquant. Both software licenses actually came with a video course and both substantially improved my knowledge of the software and rules based trading. I highly recommend watching both. Both softwares also have a forum although I do not participate as I am too much of a loner but a lot of smart traders I have learned quite a bit from. If you are thinking of investing in either I guarantee you will learn something from both that is worth the cost in the long run. If you use either let me know if you want to share any best practices or notes you have found.