A Staged Workflow for Screening Expert Advisor Optimization Results
Summary
The article presents a practical process for turning Expert Advisor optimization output into a smaller set of candidates for further testing. It recommends broad optimization first, with caution about genetic search concentrating around early promising parameter sets. Possible mitigations include repeated runs with different target criteria or fewer parameter combinations. Results are exported to a spreadsheet and screened using trader-selected columns.
Surviving parameter sets are batch-tested on separate historical intervals, then compared with their optimization results using profit per day, trades per day, and maximum drawdown. Sets whose measures diverge beyond a chosen tolerance are removed; repeating the process across history segments is intended to narrow the candidates before individual review and set-file generation. The article also describes automating data transfer, batch runs, and file creation. This is a practical workflow rather than a formal optimization theory, and its criteria are heuristic. Visual chart review remains part of final selection, and the author explicitly cautions that historical balance curves cannot guarantee future performance.
Key ideas
- Genetic optimization may focus its search near early profitable parameter sets and miss other candidates.
- Export optimization results and screen them using criteria chosen by the trader.
- Batch-test surviving parameter sets on separate historical intervals.
- Compare profit per day, trades per day, and maximum drawdown across optimization and test periods.
- Repeat screening across history segments, then inspect a small group of candidates individually.
- Historical chart quality and parameter stability cannot guarantee future results.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.