How to learn Algo Trading | Have You Made One of These Three Algo Trading Screwups?

[@rotrades]

I have a few questions. I always forward test, so I'd imagine there's a smaller chance of something like that happening to me. I try to not run any iterations though because I fear that very thing; curve-fitting. I'll test a few variables by hand, but that's about it. So on 3 years I'd test xyz and play with the params until it works well enough. And then I'd forward on another 3 years to see if those exact params work. Then another 3 years to see the same thing. The whole time I won't change any settings. I have a few questions: 1) if it doesn't work on the 1st out of sample 3 year test, what do i do then? Scrap the whole idea? 2) it has worked on 2 out of sample 3 year tests before, but then on this most recent 2024 data, it doesn't work, leading me to believe it was still somehow curve fitted. So what's that about, and what would I do then? 3) is it better to be trained on a much larger time period, then forward tested on a small one, or is my time period part of my development process okay? 4) I feel like my whole process of not running iterations because I fear curve fitting is inefficient. Is there a better way I should be going about things? Thanks Kevin

[@AlgoTradingWithKevinDavey]

Thanks for the comment. It is really hard for me to comment on what you are doing - if it is good or bad - because I have never done anything like what you are doing. The ultimate judge of your method is to run it on 100+ strategies, being consistent in the process, and see if your process leads to profitable real time performance. That is the ultimate judge.