A high-throughput quantitative trading backtesting framework and automated live execution engine built in Python. The system features multi-indicator signal filtration, Numba JIT-accelerated vector computation, Optuna Bayesian hyperparameter optimization, and exchange connectivity via CCXT.
The architecture is divided into three functional layers:
-
JIT-Accelerated Backtesting Core (
parameter_training.py):- Compiles core iteration logic down to native machine instructions using Numba (
@njit). - Achieves orders-of-magnitude speedups over pure-Python backtesting loops, evaluating thousands of historical bars within milliseconds.
- Interfaces with Optuna for automated Bayesian optimization across multidimensional parameter spaces.
- Compiles core iteration logic down to native machine instructions using Numba (
-
Signal Filtration & Strategy Execution (
main_funcion.py):- Multi-indicator statistical validation combining trend strength, market efficiency, volatility expansion, and volume anomalies.
- Enforces strict risk-managed position transitions.
-
Live Market Connectivity (CCXT Integration):
- Interfaces directly with cryptocurrency exchange endpoints (e.g., Binance, Bybit) for real-time order placement, position tracking, and balance synchronization.
The signal generator evaluates market regimes using a synchronized indicator stack:
- Kaufman's Efficiency Ratio (ER): Quantifies market trend efficiency versus noise, avoiding whipsaw trades in range-bound chop.
-
Linear Regression (
$R^2$ ) Determination: Confirms the statistical linearity and strength of the ongoing price trend. - Average Directional Index (ADX): Filters out weak, low-momentum setups when trend strength falls below predefined thresholds.
- Bollinger Band Width (BBW): Identifies volatility squeeze and expansion phases to time explosive directional breakouts.
- Volume Climax Multiplier: Detects localized institutional volume surges relative to rolling baseline averages.
- Candlestick Solid Body Ratio: Ensures price action momentum confirms directional sentiment before triggering order fills.
The strategy logic has been implemented and backtested on TradingView using Pine Script across various asset classes and timeframes:
- Bitcoin (BTC/USDT) - 5-Minute Timeframe: View Pine Script & Performance Report
- Semiconductor ETF (SOXX) - Daily Timeframe: View Pine Script & Performance Report
Terminal execution logs showing real-time market data retrieval, state evaluation, and order fulfillment:
Signal triggers, entry levels, and profit-target executions visualised on interactive chart layouts:
Parameter sweep and objective convergence graphs evaluated via Optuna:
- Single Position Constraint: Strictly maintains at most one open long or short position at any given timestamp, preventing margin overexposure.
- Dynamic Risk Sizing: Configured with a default risk allocation (e.g., $1,500 trade quantity on a $300 capital base with predefined leverage controls).
-
Hard Stop-Loss (SL) & Take-Profit (TP): Automatic order cancellation and exit triggers enforced by percentage thresholds and absolute price movement bounds (
MAX_LOSS_PTS,MIN_MOVE_PTS). - Friction & Fee Modeling: Accounts for bilateral exchange commissions (default: 0.04% per side) to provide realistic net-of-fee performance metrics.
- Core Runtime: Python 3.9+
- Performance & Optimization:
numba(Just-In-Time compilation to native machine code)optuna(Bayesian optimization of indicator thresholds)
- Data & Quantitative Analysis:
pandas,numpy(Vectorized tabular computations)pandas_ta(Technical analysis library)
- Exchange Integration:
ccxt(Unified cryptocurrency exchange API)
Clone the repository and install required packages:
git clone https://github.com/Kylechen0815/High-Performance-Quantitative-Trading.git
cd High-Performance-Quantitative-Trading
pip install ccxt pandas pandas_ta numpy colorama optuna numba