C++ Probabilistic Trading Backtester

This repository is a small C++ backtester for exploring a Simple Moving Average (SMA) crossover trading strategy as a probabilistic experiment. It is intended for research and learning — not for live trading.

The engine simulates trades on historical candlestick data and reports per-trade statistics such as expected value, variance, and win probability.

Notes about the current codebase

• The CMake target is trading_bot (see CMakeLists.txt).
• By default src/main.cpp loads data/NVDA.csv (NVIDIA historical prices). You can replace the CSV in data/ or edit main.cpp to point to a different file (for example data/BTC-USD.csv).
• Two Python downloader scripts are included to fetch sample CSVs:
  • NVDA_Downloader.py — downloads data/NVDA.csv using yfinance.
  • BTC_USD_Downloader.py — downloads data/BTC-USD.csv using yfinance.

Project layout

• include/ — public headers (backtester.hpp, strategy.hpp, etc.)
• src/ — implementation (main.cpp, backtester.cpp, strategy.cpp)
• data/ — sample CSVs (committed or generated by the downloaders)
• BTC_USD_Downloader.py, NVDA_Downloader.py — convenience scripts to fetch CSVs using Python

Prerequisites

• A C++17-compatible compiler (Clang/GCC on macOS/Linux, MSVC on Windows)
• CMake (>= 3.15)
• Internet access for CMake's FetchContent (it downloads cpr and nlohmann_json during configure)
• Optional (for dataset downloading): Python 3 with pandas and yfinance if you want to use the included downloader scripts

Install the Python deps if you plan to use the scripts:

python3 -m pip install --user pandas yfinance

Build (recommended)

Generate build files and build the executable using CMake:

cmake -S . -B build
cmake --build build

The resulting executable is build/trading_bot (or build\\trading_bot.exe on Windows).

Run

By default main.cpp is configured to load data/NVDA.csv. To run the backtester:

./build/trading_bot

If you want to run the bot on Bitcoin data, either:

1. Use the provided downloader to create data/BTC-USD.csv and edit src/main.cpp to point to data/BTC-USD.csv, or
2. Replace data/NVDA.csv with your CSV named NVDA.csv (same format), or
3. Modify main.cpp to accept a command-line path and recompile (recommended if you plan to run different datasets often).

CSV format expected by the loader

• Header line (ignored by the loader)
• Each subsequent line: Timestamp,Open,High,Low,Close

Example downloader usage (creates CSVs under data/):

python3 NVDA_Downloader.py
python3 BTC_USD_Downloader.py

Dependencies managed by CMake

The CMakeLists.txt uses FetchContent to download and make available:

• cpr — a C++ HTTP client library (used if you add network code)
• nlohmann_json — JSON handling library

These are fetched automatically during the CMake configure step.

License

See the repository root for licensing information.