Skip to content

We applied #networkscience methods to #algorithmictrading in order to identify asset groups, which were implemented as a part of a #quantitativetrading strategy.

Notifications You must be signed in to change notification settings

lukablagoje/algo-trading-group-trends-graphs

Repository files navigation

Algorithmic Trading Using Group Trends in Graph Representation

This project is part of a competition at WUTIS - Academic Trading And Investment Society, focusing on leveraging graph representations to understand and capitalize on group trends within stock market data. Our approach, recognized through a first-place victory in an Algorithmic Trading pitch competition, involves creating a dynamic graph representation based on the cross-correlation of stock price time-series data, identifying coherent and deviating group trends among stocks. More details are in the presentation file, with sample slides below:

Finding group trends in the training period:

Finding assets that deviate from group trends:

Technical Project Overview

Our project is split into three sections:

1. data_collection_graph_analysis.ipynb - Collecting the data, constructing the representation graph and analyzing potential group parameters.

2. group_trends_trading_strategy.ipynb - Formulating a trading strategy around the stocks that deviate and potentially returning to group trends.

3. parameter_optimization_strategy.ipynb - Backtesting and ranking based on the historical data to find optimal parameters.

Data

Data is downloaded from the yahoo finance Python library - https://pypi.org/project/yfinance/.

Python file where all used functions are stored - trading_functions.py

About

We applied #networkscience methods to #algorithmictrading in order to identify asset groups, which were implemented as a part of a #quantitativetrading strategy.

Resources

Stars

Watchers

Forks