romeo-v7 / README.md
JonusNattapong's picture
Upload README.md with huggingface_hub
51b8c9f verified
---
language: en
license: mit
library_name: sklearn
tags:
- trading
- finance
- gold
- xauusd
- forex
- algorithmic-trading
- smart-money-concepts
- smc
- xgboost
- lightgbm
- machine-learning
- backtesting
- technical-analysis
- multi-timeframe
- intraday-trading
- high-frequency-trading
- ensemble-model
- capital-preservation
- risk-management
- recovery-mechanisms
datasets:
- yahoo-finance-gc-f
metrics:
- accuracy
- precision
- recall
- f1
- sharpe
- max_drawdown
- cagr
- win_rate
- profit_factor
- capital_preservation_score
model-index:
- name: romeo-v7-15m
results:
- task:
type: binary-classification
name: 15-Minute Price Direction Prediction with Capital Preservation
dataset:
type: yahoo-finance-gc-f
name: Gold Futures (GC=F)
metrics:
- type: accuracy
value: 57.1
name: Win Rate
- type: profit_factor
value: 2.10
name: Profit Factor
- type: max_drawdown
value: 8.2
name: Max Drawdown
- type: capital_preservation_score
value: 28.4
name: Capital Preservation Score
---
# Romeo V7 — Capital Preservation & Recovery Trading Model
## Model Details
### Model Description
Romeo V7 is an enhanced version of Romeo V6 with advanced capital preservation strategies, recovery mechanisms, and consistent profitability features. It combines tree-based models (XGBoost and LightGBM) with sophisticated risk management to provide stable returns with lower drawdown.
- **Model Type**: Ensemble Classifier with Capital Preservation (XGBoost + LightGBM)
- **Asset**: XAUUSD (Gold Futures)
- **Strategy**: Smart Money Concepts (SMC) with capital preservation and recovery
- **Prediction Horizon**: 15-minute intraday (next bar direction)
- **Framework**: Scikit-learn, XGBoost, LightGBM
### Key Enhancements over V6
- **Dynamic Position Sizing**: Adjusts position sizes based on current capital and drawdown
- **Recovery Mechanisms**: Reduces risk during drawdown periods, increases confidence during profitable periods
- **Confidence-Based Filtering**: Only trades high-confidence signals with volume and volatility confirmation
- **Capital Preservation Rules**: Multiple safety checks to protect capital during adverse conditions
- **Volatility Adjustment**: Reduces position sizes during high volatility periods
### Model Architecture
- **Ensemble Components**:
- XGBoost Classifier: Gradient boosting with conservative parameters
- LightGBM Classifier: Efficient gradient boosting with risk-aware features
- **Enhanced Features**: 52 features including capital preservation indicators, recovery signals, and risk metrics
- **Capital Preservation Engine**: Dynamic position sizing, confidence filtering, recovery mode logic
- **Serialization**: Tree models saved in joblib format
### Intended Use
- **Primary Use**: Research, backtesting, and evaluation on historical XAUUSD data with capital preservation
- **Secondary Use**: Educational purposes for understanding risk-managed trading models
- **Out-of-Scope**: Not financial advice. Requires proper validation and risk controls for live trading
### Factors
- **Relevant Factors**: Market volatility, economic indicators, capital preservation requirements
- **Evaluation Factors**: Tested on unseen data with realistic slippage, commission, and risk management
### Metrics (Capital Preservation Mode)
- **Evaluation Data**: Unseen 15m intraday data (out-of-sample)
- **Risk Parameters**: 10% risk per trade, 2% stop loss, 5% take profit
- **Capital Preservation Settings**: 65% confidence threshold, dynamic sizing enabled
- **Metrics**:
- Initial Capital: 100
- Final Capital: 144.24
- Total Return: 44.24%
- Max Drawdown: 8.2%
- Total Trades: 133
- Win Rate: 57.1%
- Profit Factor: 2.10
- Sharpe Ratio: 4.37
- Capital Preservation Score: 28.4/100
- Recovery Effectiveness: 100%
- Risk-Adjusted Return: 5.38
- High Confidence Trades: 98/133 (74%)
- Recovery Mode Trades: 0/133 (0%)
### Capital Preservation Features
- **Dynamic Position Sizing**: Adjusts based on capital, drawdown, and volatility
- **Recovery Mode**: Activates when drawdown exceeds 85%, reduces risk by 50%
- **Confidence Filtering**: Minimum 65% confidence required for trades
- **Volatility Control**: Reduces position sizes during high volatility (>1.5% ATR)
- **Volume Confirmation**: Requires volume above 20-period average for entry
- **Safe Zone Trading**: Prefers entries within support/resistance levels
### Usage Instructions
```python
from v7.backtest_v7 import CapitalPreservationBacktester
# Initialize with capital preservation settings
backtester = CapitalPreservationBacktester({
'confidence_threshold': 0.65,
'max_risk_per_trade': 0.15,
'recovery_mode_threshold': 0.85,
'volatility_adjustment': True,
'dynamic_position_sizing': True
})
# Run backtest
results = backtester.backtest_capital_preservation(
risk_per_trade=0.10,
stop_loss=0.02,
take_profit=0.05
)
```
### Risk Management
- **Maximum Risk per Trade**: 15% of current capital
- **Recovery Mode Threshold**: 85% drawdown triggers reduced risk
- **Stop Trading Threshold**: 95% drawdown stops all trading
- **Profit Target Reset**: Returns to normal risk after 2% profit recovery
- **Volatility Filter**: Skips trades when volatility > 2%
### Performance Comparison vs V6
| Metric | Romeo V6 | Romeo V7 | Improvement |
|--------|----------|----------|-------------|
| Total Return | 10.79% | 44.24% | +33.45% |
| Max Drawdown | Higher | 8.2% | Lower |
| Win Rate | 49.28% | 57.1% | +7.82% |
| Profit Factor | ~1.5 | 2.10 | +0.6 |
| Sharpe Ratio | N/A | 4.37 | N/A |
| Capital Preservation | Basic | Advanced | Major |
### Training Data
- **Source**: Yahoo Finance GC=F (Gold Futures)
- **Timeframe**: 15-minute intraday data
- **Period**: Historical data with enhanced feature engineering
- **Augmentation**: Noise injection for robustness
- **Validation**: Out-of-sample testing with capital preservation metrics
### Ethical Considerations
- Designed for capital preservation and risk management
- Includes multiple safety mechanisms to prevent excessive losses
- Recovery mechanisms help maintain trading capital during adverse conditions
- All results are historical backtests, not guaranteed future performance
### Maintenance
- Retrain monthly with fresh data
- Monitor capital preservation metrics
- Adjust confidence thresholds based on market conditions
- Validate recovery mechanisms effectiveness
---
*Romeo V7 represents a significant advancement in algorithmic trading with a focus on capital preservation and consistent profitability.*