Mercury

Event Driven Backtesting Framework

Visualizing a mean reversion.

Building

This is a cmake project. Make sure you have some sort of c++ compiler that is detected by cmake. Steps :-

1. git clone https://github.com/Stoozy/mercury.git
2. cd mercury
3. mkdir build
4. cd build
5. Run cmake ... Makefile will be created in build directory.
6. Run make. Now, the executable build/mercury should be created.

Running

Mercury requires a config.json file to run. Here is a sample one:

{
  "historical_data": "AAPL.csv",
  "strategy": {
    "name": "MEAN_REVERSION",
    "moving_average": "EMA",
    "days": {
      "short_term": 10,
      "long_term": 30
    },
    "smoothing": 2,
    "output": "output.csv"
  },
  "capital": 10000,
  "positions": [
    {
      "symbol": "AAPL",
      "quantity": 0,
      "price": 0
    }
  ],
  "slippage": 0.005,
  "transaction_cost": 0.001,
  "start_date": "2018-10-12",
  "end_date": "2023-10-11"
}

1. Download historical data from internet. You can use Yahoo Finance.
2. Put the csv file in the same directory as config.json. In config.json, change the following fields
   • historical_data to the file-name.csv
   • start_date to the start date in your file
   • end_date to the end date in your file.
3. Go to the parent directory. Run build/mercury -c config.json. The file output.csv will be created.
4. Now, run python visualize.py to view plot of the output graphs. Note that you need pandas and matplotlib installed in your system.