Parallelizing Strategy Optimization with the MetaTrader 5 Cloud Network
Summary
The article explains how MetaTrader 5 distributes independent Strategy Tester optimization passes across local agents, a local network farm, and the MQL5 Cloud Network. It describes configuring agents, preparing an optimization, and distributing the Expert Advisor, required indicators and data, test conditions, and parameter sets. Cloud nodes assign task packages to available agents; historical market data is synchronized when needed. The article also notes deployment constraints, including that DLLs are not allowed in cloud tests, and recommends running a local test first so required history is available.
A benchmark using a moving-average Expert Advisor reports 14,040 passes taking just over an hour on eight local agents, under sixteen minutes with twenty additional network agents, and 26 seconds on the cloud network. These figures illustrate the speedup reported in that specific setup, rather than a general guarantee: hardware, agent availability, network conditions, and workload can affect results. The article concerns computation and tester infrastructure, not the validity of optimized parameters or the trading strategy’s out-of-sample performance.
Key ideas
- Strategy Tester optimization runs independent parameter combinations that can be split among multiple agents.
- The article describes local agents, local network farms, and cloud agents as ways to increase parallel capacity.
- Cloud nodes distribute task packages, while required market history is synchronized for agents.
- The reported benchmark shows a large speedup in one setup, but runtime depends on available resources and workload.
- Faster optimization does not establish that selected parameters will perform well on unseen data.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.