An AI Crypto Trading Workflow with Performance-Based Risk Controls

Summary

This document describes an automated cryptocurrency trading system organized around two workflows: an hourly analysis and decision cycle, and a separate frequent monitor for take-profit and stop-loss conditions. The analysis cycle gathers multi-timeframe candles, technical indicators, funding data, account state, open positions, and per-asset trading history. A language model then returns structured signals, including direction, leverage, size, targets, confidence, and directional bias, for the execution module to process.

Risk controls include sizing based on risk amount and stop distance, limits tied to available account funds, and a per-coin freeze after consecutive losses. The decision rules adjust exposure using historical performance and directional results, while initially relying on technical analysis when a coin has no track record. A layered parser is intended to recover signals from imperfect model output. This is an architecture and rule description, not evidence that the system is profitable or reliable: no backtest results or live performance measurements are supplied. Its behavior also depends on prompt quality, data integrity, execution conditions, and the validity of the historical statistics used to adjust risk.

Key ideas

• The system separates periodic signal generation from frequent monitoring of open-position exits.
• Market inputs combine multiple timeframes, technical indicators, funding data, account state, and trade history.
• Risk sizing depends on stop distance and account constraints, with exposure adjusted using asset performance history.
• A consecutive-loss freeze rule suspends trading in affected coins for a defined period.
• The document explains system design but provides no measured trading results.

Tags

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