Open-source EAs on GitHub and the MQL5 code base are a good learning resource. Taking one apart and finding out why it fails (or works) teaches more than most courses. A thread for these autopsies could be useful, with a few ground rules so it stays clean.
Suggested format for each autopsy post:
- Source: repository or code base name, licence, last update. Link only to the code itself, not to sales pages, signal services or referral offers.
- What it claims to do, in one paragraph.
- What it actually does, after reading the code. These often differ.
- Test setup: pair, period, tick data source, spread and commission assumptions.
- Results with and without realistic costs.
- The failure mode, if there is one: a missing stop, martingale buried in a function, a hard-coded server time, an indicator that repaints.
Reading the source is the point. Many free EAs that look like trend scalpers have a hidden averaging-down routine, and you only see it in the code.
A word of caution: compile and run anything downloaded on a demo account in a separate terminal. Some "free" EAs phone home or include DLL calls that have no business in a trading robot.
Who has a candidate for the first autopsy?
Open-source scalp EA autopsies: how to post links carefully
-
FXS London Scalper
- Posts: 806
- Joined: Sun Oct 18, 2026 12:25 pm
Open-source scalp EA autopsies: how to post links carefully
FXS Official Team | London Session Desk