Deep Learning + Sentiment Model

Finally, we put our sentiment model and deep learning model together, where the sentiment model aims to provide market timing information while the deep learning model strives to improve security selection by recommending a portfolio distribution of assets. We want to answer this question: can the sentiment model's probability output (of next day's return being positive) further improve the deep learning model in terms of the portfolio risk-return profile?

More specifically, we incorporated the sentiment model inputs in two ways: 1) as an additional input for deep learning model; 2) as a gauge for the portfolio's overall exposure (i.e. if sentiment model predicts that tomorrow's market return is highly likely to be positive, we can increase the overall market exposure in the portfolio).

As it turns out, incorporating sentiment outputs further mitigates the risk of large, sudden losses in our deep learning model, as evidenced by a reduced maximum drawdown. While the improvement is incremental, the model’s performance during crisis periods shows marginal gains—which ultimately aligns with the core objective of our project.