Using Genetic Algorithms for Ongoing Expert Advisor Reoptimization

Summary

The article proposes moving parameter reoptimization into an Expert Advisor so it can adapt without stopping for repeated manual runs in the Strategy Tester. A genetic algorithm searches for inputs, initially for a selected strategy and later potentially for the best strategy from a collection. The EA is described as trading completed bars on one instrument, with no position additions or partial closes in the basic simulation. Its fitness function can simulate open prices only, while a more detailed every-tick simulation is suggested for stop and take-profit handling.

The discussion includes practical components such as detecting new bars, opening and closing positions, calculating available volume, and tracking relative account drawdown. Strategies mentioned include moving-average crossovers and a neural-network approach, with an example of trade history included. The examples do not establish that continuous optimization avoids overfitting or improves out-of-sample returns. Results depend on the fitness simulation, market data, and risk settings; the text also frames selection among different strategies as unfinished work.

Key ideas

• The Expert Advisor can be designed to decide when to reoptimize and run a genetic algorithm while operating.
• The basic fitness simulation trades completed bars and omits additions and partial closes.
• A more detailed tick simulation is needed to model stop and take-profit behavior accurately.
• Position sizing and relative drawdown tracking are included as supporting mechanisms.
• The article raises strategy selection from a system bank as a further challenge rather than a completed method.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.