A Live Trading Workflow for Market Data, AI Analysis, and Order Routing

Summary

This example outlines a manually triggered workflow that collects account positions, recent candlesticks, and a long-short account ratio, then combines the data for language-model analysis. The analysis prompt asks for price action, volume, technical indicators, and current position risk, followed by a choice among five actions: enter or exit a long or short position, or take no action. A classification node routes the selected category to an order executor or a no-action notification; other nodes format and log the analysis and send app notifications.

The document is an implementation example, not a tested trading method: it provides no performance results, risk controls, or validation of the model's decisions. The workflow is inactive and uses a manual trigger, while the depicted trade nodes specify a fixed order amount. It demonstrates how data collection and AI-generated analysis might be connected to execution, but does not establish that the process is safe or profitable. The distinction between analysis output and a structured trading decision is handled by a separate classification step.

Key ideas

• The workflow gathers account position, candlestick, and long-short ratio data for analysis.
• A language-model prompt requests technical and position-risk assessment and one of five trade decisions.
• A classifier routes decisions to entry, exit, or no-action nodes, followed by logging or notifications.
• The example is manually triggered and inactive, with a fixed amount shown for entry orders.
• No backtest, decision validation, or performance evidence is supplied.

Tags

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