5MWh Battery using 1.4 charge/discharge cycles per day - using a genetic algorithm to maximize profits trading on 15-min timeframes on the energy market.
Clone the git repo To install all required packages, do the following:
Create and activate a virtual enviroment, eg.
python3 -m venv venv
source venv/bin/activate
Install required packages
pip install -r requirements.txt
Use extractCSVColumn.py to extract the appropriate column of the file and output it to the prices.csv file.
Input all the recorded prices #.# \n. The prices will be split into groups of 96 representing one day.
Run the main.py file.
Depending on computational power and time available, adjust the POPULATION_SIZE and GENERATIONS variable accordingly.
The resulting trading strategy will be in the output/tradingStrategy.csv file which describes the chosen trading strategy.
loadShift
├── .gitignore
├── README.md
├── requirements.txt
├── input/
│ └── prices.csv
│ └── balancingEnergy2023.csv
│ └── dayAhead2023.csv
├── output/
│ └── tradingStrategy.csv
└── src/
└── main.py
└── extractCSVColumn.py
Historic price-data is input here via csv-files eg. 2023 day-ahead trading prices and 2023 balancing-energy prices.
The appropriate column of the input .csv file can be converted into the prices.csv file format via the extractCSVColumn.py function
Input all the recorded prices. The prices will be split into groups of 96 representing one day {24*4 = 96 15-min timeframes per day}.
Output of the trading choices made for each timeframe (Buy/Sell/Nothing) 1.4 Cycles equals to a total of 14 Buys and 14 Sells per day. This number of trades per day has to be reached.
This project uses simulated strategies and historical data, which may be incomplete or inaccurate. No guarantees of future performance or real-world applicability. Use at your own risk.