Trading System Project

Modular backtesting environment that follows the four parts outlined in the project brief: data acquisition and preparation, exchange simulation, strategy backtesting, and an Alpaca paper-trading handoff. Everything runs locally in pure Python.

---

Part 1 – Data Download and Preparation

Step 1: Download Intraday Market Data

• data_pipeline.download_equity_data() pulls equity candles via yfinance and stores them in raw_data_stock/ as CSV files with columns Datetime, Open, High, Low, Close, Volume.
• data_pipeline.download_crypto_data() fetches crypto candles from the Binance REST API and stores them in raw_data_crypto/ with the same schema.

from data_pipeline import download_equity_data, download_crypto_data

raw_equity_path = download_equity_data("AAPL", period="7d", interval="1m")
raw_crypto_path = download_crypto_data("BTCUSDT", interval="1m", limit=1000)

Step 2: Clean and Organize Data

• data_pipeline.clean_market_data() removes missing/duplicate rows, enforces a chronological Datetime index, and adds derived features such as returns, rolling volatility, and momentum.
• The cleaned datasets are saved to clean_data_stock/ or clean_data_crypto/ and are ready for ingestion by the gateway.

from data_pipeline import clean_market_data

clean_path = clean_market_data(raw_equity_path)
print("Cleaned data:", clean_path)

Step 3: Create a Trading Strategy

• strategy_base.Strategy defines the interface, while strategy_base.MovingAverageStrategy implements a moving average crossover with clear entry/exit rules and position sizing logic.

from strategy_base import MovingAverageStrategy

strategy = MovingAverageStrategy(short_window=20, long_window=60, position_size=15)

---

Part 2 – Backtester Framework

Market Data Gateway (gateway.py)

Reads cleaned CSV files and streams candles row-by-row through an iterator or the stream() generator. Mimics a live feed by optionally adding delays.

Order Book (order_book.py)

Stores bids and asks in heaps to enforce price-time priority. Supports order addition, modification (by cancel + re-add), cancellation, and matching.

Order Manager & Logging (order_manager.py)

Performs capital sufficiency checks, position-risk checks (long + short limits), and order rate limiting. OrderLoggingGateway writes a JSONL audit trail of submitted, rejected, and executed orders.

Matching Engine (matching_engine.py)

Randomly decides whether matched orders are filled, partially filled, or cancelled, and returns execution reports for the backtester.

---

Part 3 – Strategy Backtesting

Strategy_Backtesting.Backtester ties together the gateway, strategy, order book, order manager, matching engine, and logger to simulate execution. The loop:

1. Pulls a new candle from the gateway.
2. Runs the strategy to update indicators and generate a signal.
3. Builds an order (with configurable position size) when signals fire.
4. Validates via OrderManager and inserts orders into the OrderBook.
5. Uses the MatchingEngine to simulate fills/partials/cancellations.
6. Tracks portfolio cash, positions, equity, and trade-level PnL.

Strategy_Backtesting.PerformanceAnalyzer computes PnL, Sharpe ratio, max drawdown, and win rate. plot_equity() visualizes the equity curve.

Quick start

python test_system.py

This generates sample data (if needed), runs the backtest, and prints summary statistics.

---

Part 4 – Alpaca Trading Challenge

1. Create an Alpaca account at alpaca.markets and complete identity verification (paper trading only).
2. Configure paper trading via the Alpaca dashboard (Paper Overview → Reset virtual funds as needed).
3. Obtain API keys from the dashboard and keep them secure. Use only paper trading endpoints.
4. Retrieve market data with the Alpaca SDK:

   import alpaca_trade_api as tradeapi
   
   api = tradeapi.REST(API_KEY, API_SECRET, "https://paper-api.alpaca.markets")
   data = api.get_bars("AAPL", "1Min", limit=100).df
   data.to_csv("alpaca_data/AAPL_1m.csv")

5. Save market data in flat files (CSV/Parquet) or a database. Organize by asset and timeframe, and handle timezones carefully.
6. Use your Part 1 strategy with Alpaca by feeding Alpaca candles through the same classes (MarketDataGateway, Backtester, etc.). No new strategy code is required; you can re-use and tune the existing implementation.

---

Project Structure

clean_data_crypto/
clean_data_stock/
raw_data_crypto/
raw_data_stock/
data_pipeline.py
gateway.py
order_book.py
order_manager.py
matching_engine.py
strategy_base.py
Strategy_Backtesting.py
test_system.py
Readme.md

---

Requirements

Create a virtual environment and install dependencies:

pip install -r requirements.txt

Key libraries: pandas, numpy, matplotlib, yfinance, requests, alpaca-trade-api (for Part 4), and standard Python tooling.