Investor - Fund Manager Behavioral Discrepancy: How much does it affect portfolio performance ?

Investors-fund manager live streams are common in modern media supported finance platforms..The abundance of data on these live streams allows detailed analysis of behavioral disposition of both sellers and buyers. A projection or embedding of behavioral components like vocal arousal, visual expression etcetera, allows a deep multimodal analysis of fund sales. Using Canonical Correlation Analysis, one identifies dimensions highly correlated with sentiment, leading to the creation of the Consumer Seller Behavioral Discrepancy (CSBD) metric. This metric quantifies differences in behavior and sentiment towards funds between consumers and sellers. Empirical analysis reveals a significant positive correlation between CSBD and future returns that becomes negative beyond a threshold, particularly noticeable in stocks compared to bonds. This trend strengthens over 1, 5, 10, and 20-day periods, suggesting a propensity for consumers to sell off funds when overly optimistic, especially in the short term. Utilizing these findings, we analyze a ”short long” trading strategy that capitalizes on extreme CSBD values in stock funds, resulting in profitable outcomes. Additionally, we integrated CSBD with other variables into a neural network, and found that CSBD significantly impacts the training and testing of this neural network: it accelerates convergence and reduces the final loss.