Learning Embedded Representation of Stock Correlation Matrix Using Graph Machine Learning

Explore an innovative approach to understanding stock market correlations through graph machine learning in this 34-minute conference talk from the Toronto Machine Learning Series. Dive into Bhaskarjit Sarmah's presentation on using Node2Vec algorithm to create embedded representations of stock correlation matrices. Learn how this method compresses complex financial networks into lower-dimensional spaces, allowing for more nuanced analysis of stock relationships. Discover the parallels between this approach and techniques used in Natural Language Processing, and examine both quantitative and qualitative metrics for evaluating the resulting embeddings. Follow along as Sarmah demonstrates the application of this method to S&P 500 stock data, discussing its potential impact on risk management, portfolio construction, and trading strategies. Gain insights into the future research directions and practical use cases of this cutting-edge technique in financial analysis.

Intro
AGENDA
PROBLEM STATEMENT
INTRODUCTION TO WORD EMBEDDINGS
DATA PREPARATION
METHODOLOGY
CREATING SENTENCE LIKE STRUCTURES FROM NETWORK
CREATING CONTEXT AND TARGET OF STOCKS
Training the embeddings
Quantitative Evaluation of Embeddings
USE CASES
FUTURE RESEARCH AND REFERENCES