An ONNX Market-Structure Strategy Using Fibonacci Pullbacks

Summary

The document describes a hybrid trading system in which an ONNX model classifies market structure and conventional rules govern trade selection and management. Model inputs include ATR-scaled momentum, distances from recent swing extremes, relative tick volume, candle body strength, and time of day. A predicted structure label and its probability are filtered by a confidence threshold and a 50-period simple moving average direction filter.

For aligned signals, the strategy places limit orders at a chosen Fibonacci retracement of a detected swing, with structural stops, a fixed risk-reward target, order expiry, and a rule allowing only one position or pending order at a time. It also describes moving stops as price advances toward the target. The evidence is a methodological description, alongside references to hourly gold and euro-dollar backtests; no performance statistics, validation details, or model-training data are supplied. The stated defaults and rules therefore do not establish profitability or robustness.

Key ideas

• The ONNX classifier uses six normalized features to estimate market structure.
• Signals must pass a probability threshold and align with the moving-average trend filter.
• Entries use limit orders at a Fibonacci pullback level within a detected swing.
• Structural stops, a fixed risk-reward target, order expiry, and a one-trade limit constrain exposure.
• Risk-reward progress determines when and how the stop trails, but the document gives no evidence of out-of-sample performance.

Tags

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