Importance, Settings, and Procedures Regarding Forward Testing
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Hi,
I've been reading numerous threads, documentation, and resources regarding forward testing. I've also been using it for quite a while now, and I'm curious about your overall thoughts on this concept.
It's understandable; we all know what this is: a popular and somewhat reliable way of avoiding overfitting in any system. Many times, it can be underwhelming. You witness how impressive the backtest results are, feeling like you've discovered the wheel, only to find that with forward testing, reality hits you, often without any empathy and not caring about your feelings.
I wanted to create this thread to discuss it more in-depth, to learn about your opinions and experiences with it, and to share with other traders who might not be fully aware of how important (in my opinion) this concept is. But what actually motivated me to open it is to gather insights into the best procedures for applying it, specifically the proportion of out-of-sample (OOS) data and the actual optimization dataset.
In my opinion and experience, I wouldn't use more than 1/4 of the actual dataset for forward testing. For example, if we have 4 years of data, I would optimize the system with 3 years and use only the last one to test the results. However, I'm eager to read your opinions on this issue.