Downloads · 30 days
0
xauusd-trading-predictor
Bhushanz16/xauusd-trading-predictor
xauusd-trading-predictor is a machine learning model from Bhushanz16. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
A machine learning model for predicting XAUUSD (Gold/USD) directional moves for swing trading with a 3-day prediction horizon.
Access
Public
Updated May 8, 2026
Repo size
—
Likes
0
Public
Repo makeup
Click a slice to open those files.
.json809 KB · 94%
At a glance
- Access
- Public
- Created
- May 8, 2026
- Updated
- May 8, 2026
- SHA
- 846bc51f
Model card
From the Hugging Face model README
XAUUSD Trading Prediction Model
A machine learning model for predicting XAUUSD (Gold/USD) directional moves for swing trading with a 3-day prediction horizon.
Model Overview
| Property | Value |
|---|---|
| Model Type | XGBoost Classifier |
| Task | Binary Classification (Up vs Down) |
| Prediction Horizon | 3 trading days |
| Lookback Window | 20 days |
| Data Period | 10 years (2016-2026) |
| Move Threshold | 0.3% minimum move |
| Confidence Threshold | 58% |
Architecture
The model uses a multi-asset feature engineering pipeline that combines:
Primary Asset Features (XAUUSD)
- Returns at multiple lags (1, 2, 3, 5, 10, 20 days)
- Moving average distances and slopes (5, 10, 20, 50 days)
- Volatility measures (5, 10, 20-day rolling)
- RSI, MACD, Bollinger Bands position
- Rate of change, momentum, trend strength
Cross-Asset Features
- DXY (US Dollar Index) - inverse correlation with gold
- Silver (SI=F) - highly correlated with gold
- S&P 500 (SPY) - risk sentiment proxy
Sequence Features
Features are flattened over a 20-day lookback window, creating ~2,340 input dimensions per sample.
Performance Metrics
Classification
| Metric | Value |
|---|---|
| Test Accuracy | 53.25% |
| Test AUC | 55.77% |
| Down Recall | 47% |
| Up Recall | 57% |
Trading Simulation
| Threshold | Trades | Win Rate | Total Return | Sharpe |
|---|---|---|---|---|
| 55% | High | Variable | Variable | Variable |
| 58% | Balanced | Best risk/reward | Optimized | Optimized |
| 60%+ | Low | High precision | Conservative | Conservative |
Benchmark (Buy & Hold during test period): +95.91%
Usage
Installation
pip install yfinance xgboost pandas numpy scikit-learn
Inference
from xauusd_inference import main
result = main()
# Returns: {'signal': '🟢 LONG (Buy)', 'prob_up': 0.62, ...}
Command Line
python xauusd_inference.py
Signal Interpretation
| Signal | Probability Range | Action |
|---|---|---|
| 🟢 LONG | Prob Up > 58% | Consider buying XAUUSD |
| 🔴 SHORT | Prob Up < 42% | Consider selling XAUUSD |
| 🟡 NEUTRAL | 42% ≤ Prob Up ≤ 58% | Wait for clearer signal |
Data Sources
All data fetched live from Yahoo Finance:
GC=F- Gold Futures (XAUUSD proxy)DX-Y.NYB- US Dollar IndexSI=F- Silver FuturesSPY- S&P 500 ETF
Disclaimer
⚠️ This model is for educational and research purposes only.
Financial markets are inherently unpredictable. Past performance does not guarantee future results. Trading financial instruments carries significant risk of loss. Always do your own research and consider consulting a financial advisor before making investment decisions.
Files
| File | Description |
|---|---|
xauusd_trading_model.json | Trained XGBoost model |
xauusd_scaler.pkl | Feature scaler (StandardScaler) |
xauusd_feature_names.json | Feature name mapping |
xauusd_model_config.json | Model configuration |
xauusd_inference.py | Real-time inference pipeline |
xauusd_final_model.py | Training script |
Research Foundation
This model is based on research in financial time series forecasting:
- Wavelet denoising preprocessing (PyWavelets, db4)
- Multi-asset feature correlation analysis
- Time-series cross-validation for non-stationary data
- XGBoost with class balancing for imbalanced market regimes
- Created
- May 8, 2026
- Updated
- May 8, 2026