Structuring AI Trading Signals with a Dispatch and Parsing Layer
Summary
The article presents an MQL5 architecture that turns AI analysis into structured signals an Expert Advisor can route, parse, display, and potentially execute. A central dispatcher maps stable action identifiers to separate workflows, each with its own data collection and prompt. The example console includes actions for recent chart data, bar direction, candlestick scans, daily bias, trendlines, key levels, and clearing drawings. AI responses use a line-based key-value format with explicit conventions, making them easier to parse than free-form prose.
The described execution layer sends some signal types as market orders and maps key-level support or resistance plus bounce or break bias to pending-order types. Stops and targets use a buffer based on recent bar range. The system also draws labeled signals on the chart and is intended for backtesting. The article mainly explains software structure and workflow; it does not establish predictive accuracy or profitable performance. Its trading rules include hardcoded mappings and AI-generated judgments, so results depend on prompt behavior, implementation details, and appropriate validation.
Key ideas
- A central dispatcher routes each user action to a dedicated data and prompt workflow.
- A line-based key-value response format makes AI signals easier to parse and log.
- The example distinguishes market-order signals from pending orders for key-level setups.
- Recent bar range provides a symbol-relative buffer for stop and target distances.
- Chart drawings and backtesting support inspection, but the article does not demonstrate profitability.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.