EcoSta 2022: Start Registration
View Submission - EcoSta2022
A0409
Title: Extracting stock predictive information in fund managers decisions through machine learning with hypergraph Authors:  Chu-Lan Kao - National Yang Ming Chiao Tung University (Taiwan) [presenting]
Vincent Tseng - National Yang Ming Chiao Tung University (Taiwan)
You-Sin Chen - National Yang Ming Chiao Tung University (Taiwan)
Abstract: A machine learning framework is proposed that incorporates mutual fund managers' portfolio decisions to predict stock price movements. It employs a weighted hypergraph to connect the information of the funds' portfolio weight changing to the corresponding stocks in the portfolio, which is further aggregated with the price information extracted through classical machine learning methods. The framework then predicts the stock price movements as a classification problem. We put this framework to the test through the Taiwanese stock and mutual fund data, discovering that the managers' decisions do provide stock predictive information in addition to the classical technical data. Moreover, we determined that the level of information provided by the managers' decisions is asymmetric under different market conditions, which further establishes the existing findings in the literature on the mutual fund behavior under different market conditions. Further discussions on incorporating other expert major trading behaviors in Taiwan and model comparisons are provided.