Tuesday 7 October 10:50 a.m. - 12:30 p.m. (Europe/Amsterdam)
Abstract
The growing influence of social media on financial markets is becoming increasingly clear, as seen in the impact Reddit discussions have had on the stock prices of companies like GameStop. This trend highlights the strong effect social networks can have on stock markets. However, analysing and predicting these effects is difficult due to the complexity and sparsity of these social networks. In response, recent advances have utilised dynamic graph neural networks to enhance analytical and predictive abilities. We further improved the performance of existing dynamic graph neural network models by incorporating a more efficient negative sampling strategy, which increases accuracy in modelling and forecasting user communication dynamics. These enhanced models allow us to better understand the connections between social network interactions and stock market fluctuations, capturing the social forces that drive market movements. This approach not only provides new tools for analysis but also helps identify potential "meme" stocks, giving investors valuable insights to take advantage of emerging market trends and make more informed decisions.
