The current study is a preliminary exploration of the implication of the introduction of advanced   machine learning sentiment analysis into financial forecasting feature engineering, leveraging financial news, social media (Twitter and Reddit), and corporate earnings call transcripts. The purpose is to test the effect of such sentiment measures on trading volumes in capital markets. It is interesting to note that the bulk of literature on sentiment related financial market phenomena focuses on raw price data. As Such, sentiment driven tend to be vulnerable in making biased decisions because they treat the existence of a tweet or news event as a trigger to buy or sell stock without considering how the implication of sentiment content of that tweet or news event may affect overall market activity. The work employs a wide variety of predictive models, including classical methods like Random Forest, XGBoost, or deep learning architectures like LSTM and transformers based on BERT to discover the most impactful sentiment sources. This suggests sentiment signals (e.g., from Reddit and earnings calls) can provide strong improvements in performance prediction. Future studies on hybrid sentiment-based financial prediction systems can be built upon the findings of this study.