XAUUSD V03H SELL
Experimental machine-learning SELL model for XAUUSD / Gold.
Model information
- Model ID:
V03H_WF2025_RF_SELL - Strategy ID:
V03H_F002_SELL - Instrument:
XAUUSD - Direction:
SELL - Timeframe:
M15 - Model family:
Random Forest - Inference format:
ONNX - Required features:
300
Decision rule
The model produces a SELL probability. The runtime converts the score
into a historical percentile rank using score_distribution.parquet.
Current operational rule:
rank_pct >= 0.90
Original locked research threshold:
rank_pct >= 0.99
Trade contract
Direction: SELL
Timeframe: M15
Stop Loss: 1.0 ATR
Take Profit: 2.0 ATR
Maximum Hold: 32 M15 bars
Magic Number: 7300201
Included files
final_rf_sell_model.onnx
feature_order.json
score_distribution.parquet
policy_config.json
model_registry.json
Feature contract
Inputs must use exactly the features and ordering defined in:
feature_order.json
The live runtime validates:
required features = 300
missing features = 0
null features = 0
infinite features = 0
before inference.
Runtime flow
MT5 closed bars
-> feature engineering
-> 300-feature vector
-> V03H ONNX model
-> SELL probability
-> historical percentile rank
-> rank threshold
-> SELL signal or HOLD
Timestamp requirement
Live market timestamps must use consistent UTC semantics.
Incorrect broker-server time handling can alter session and time-derived features even when other indicators appear correct.
Intended use
This model is intended to operate as one independent strategy inside a multi-strategy XAUUSD portfolio framework.
Other BUY or SELL models do not need to agree before this strategy generates its own signal.
Risk notice
This repository contains experimental trading research software.
Historical, validation, or backtest performance does not guarantee future profitability. Leveraged trading can result in substantial losses.
Use a demo or paper environment first and independently validate model behavior, execution, spread, commission, slippage, and risk controls.