Vectorized Algorithmic Trading Engine

Dynamic Moving Average Backtester

A robust, object-oriented backtesting engine capable of simulating technical analysis strategies on historical financial data. It features a vectorized execution model for high-performance testing (utilizing Pandas/NumPy) and includes a parameter optimizer (Grid Search) to identify alpha-generating signal combinations.

Project Architecture

algo-trading-engine/
|-- backtester.py     # Encapsulates data logic, signal generation, and performance metrics (OOP).
|-- optimizer.py      # Implements Grid Search to solve for optimal Lookback Windows.
|-- main.py           # Entry point for strategy execution and visualization.
|-- requirements.txt  # Python dependencies.
|-- .gitignore        # Git configuration.
|-- README.md         # Project documentation.

Strategy: Golden Cross (Momentum)

The strategy utilizes a "Golden Cross" methodology to capture medium-term momentum. It employs two Simple Moving Averages (SMA):

• Fast Line (SMA_25): Reacts quickly to price changes; represents the short-term trend.
• Slow Line (SMA_55): Reacts slowly; filters out noise and represents the medium-term trend.

The Math

The Simple Moving Average (SMA) is calculated as the arithmetic mean of the closing prices over the last $N$ days: $$SMA_n = \frac{1}{n} \sum_{i=0}^{n-1} P_{t-i}$$

Trading Logic

• The system generates signals based on the crossover of these two indicators:
• BUY SIGNAL (Long): When the Fast Line (25) crosses above the Slow Line (55). This indicates that short-term momentum is outpacing the long-term average.
• SELL SIGNAL (Neutral/Cash): When the Fast Line (25) crosses below the Slow Line (55). This indicates momentum is failing, triggering a shift to a risk-off position.

Performance Analysis

The engine was backtested on Apple (AAPL) data from 2024-2025.

Key Metrics

• Total Return: 70.13% (vs Benchmark ~30%)
• Max Drawdown: -14.05% (vs Benchmark -30%)
• Sharpe Ratio: 1.82

Key Insight: Notably, during the Q1 2025 market correction, the algorithm successfully identified the trend reversal and shifted to a cash position (neutral). This risk management behavior prevented the ~30% drawdown that the "Buy & Hold" benchmark suffered, allowing the strategy to significantly outperform the underlying asset by year-end.

How to Run

1. Clone the repository:

    https://github.com/SamaKool/Algo-Trading-Engine.git

2. Install dependencies:

    Bashpip install -r requirements.txt

3. Run the strategy:

    Bashpython main.py

Future Roadmap

Planned improvements to enhance the engine's capability:

• Short Selling: Implementing logic to profit from downtrends rather than just moving to cash.
• Risk Metrics: Calculating Sharpe Ratio, Sortino Ratio, and Max Drawdown automatically.
• Walk-Forward Optimization: Implementing rolling-window training to validate parameters and prevent overfitting to historical data.