Systematic strategy research and portfolio optimization engine.
AlphaPilot is a quantitative research platform designed for testing rule-based trading strategies, analyzing historical performance, and studying portfolio construction under classical mean–variance theory.
The system provides a structured workflow for financial data analysis, strategy simulation, and portfolio optimization using historical market datasets. Rather than focusing on single trading rules, the project emphasizes systematic experimentation, performance evaluation, and risk–return modeling.
AlphaPilot implements a basic pipeline for systematic strategy research:
- financial time-series ingestion from CSV datasets
- return and volatility estimation
- rule-based trading signal generation
- historical strategy backtesting
- equity curve construction
- portfolio allocation via mean–variance optimization
- interactive visualization of strategy and portfolio behavior
The platform provides an exploratory environment for studying strategy performance and portfolio construction under simplified assumptions.
The system supports basic rule-based trading strategies.
A simple trend-following rule based on the relationship between short-term and long-term moving averages.
Signals are generated as Buy → short MA crosses above long MA Sell → short MA crosses below long MA
Momentum signals are generated using the Relative Strength Index.
Typical thresholds: RSI < 30 → Buy signal RSI > 70 → Sell signal
These strategies serve as baseline examples for studying systematic trading rules.
The backtesting module evaluates strategy performance using historical data.
Key outputs include:
- daily return series
- cumulative equity curve
- strategy profit and loss
- volatility estimation
- basic performance statistics
The goal is not to claim profitable strategies but to provide a framework for analyzing strategy behavior under historical market conditions.
AlphaPilot implements classical mean–variance portfolio optimization following the Markowitz framework.
Given asset return estimates: maximize wᵀ μ subject to wᵀ Σ w ≤ σ² ∑ w = 1
where
- μ represents expected returns
- Σ represents the covariance matrix
- w represents asset weights
The optimizer computes portfolio allocations that balance expected return and risk.
Typical analysis flow:
- Upload historical price dataset
- Compute daily returns
- Select trading strategy
- Generate buy/sell signals
- Run historical backtest
- Evaluate performance statistics
- Construct optimized portfolio allocation
Input datasets should contain at least one price column. DATE,CLOSE 2024-01-01,3200.5 2024-01-02,3215.3 2024-01-03,3198.7
Additional columns such as OPEN, HIGH, LOW, and VOLUME may also be included.
- Python
- NumPy
- Pandas
- Streamlit
- Plotly
- rule-based trading strategies
- historical backtesting
- mean–variance portfolio optimization
Clone repository git clone https://github.com/Aryan-core/AlphaPilot.git cd AlphaPilot
Install dependencies pip install -r requirements.txt
Run application streamlit run app.py
https://alphapilot.streamlit.app
AlphaPilot is part of a broader quantitative research toolkit:
- QuantCore AI — financial data ingestion and research infrastructure
- VaRGuard — Monte Carlo risk simulation and Value at Risk estimation
- AlphaPilot — strategy testing and portfolio construction
Together these tools form a simple research pipeline: Data Processing → Risk Analysis → Strategy Research
Current version focuses on:
- systematic strategy experimentation
- backtesting infrastructure
- basic portfolio optimization
The system is intended as a research environment rather than a production trading system.
Markowitz, H.
Portfolio Selection
Journal of Finance, 1952.
Hull, J.
Options, Futures, and Other Derivatives.
Glasserman, P.
Monte Carlo Methods in Financial Engineering.
Aryan Khan
Drexel University
MIT License