Leveraging FOSS for Finance

Social media platforms such as Twitter and Reddit have become important venues for expressing and disseminating public opinion. The vast amount of user-generated content on these platforms provides a rich data source that can potentially offer valuable insights into market sentiment. Sentiment analysis, which involves analyzing textual data to determine the emotional tone behind words, has emerged as a powerful tool for gauging public opinion and predicting market trends. Despite the recognized importance of sentiment analysis, its integration into predictive models for cryptocurrency prices is still in its nascent stages. The surge in cryptocurrency investments has heightened the need for precise sentiment analysis tools that can effectively gauge market sentiment and predict price fluctuations. This talk presents a comparative analysis of advanced FOSS sentiment analysis tools, specifically Crypto-BERT, FinBERT, VADER, and SenticNet, to evaluate their effectiveness in deducing optimal cryptocurrency investments. The talk will explore the correlation between sentiment and cryptocurrency prices using a pre-trained transformer model, DistilBERT, which is fine-tuned to discern market trends and sentiment dynamics. The findings offer valuable insights for investors, analysts, and researchers in the cryptocurrency domain, ultimately contributing to more informed and strategic trading practices.

It is based on my research, "An Empirical Study of Financial BERT Models for Sentiment Analysis and Cryptocurrency Price Correlation", which was published, at the 2024 9th International Conference for Convergence in Technology (I2CT), one of the most premium conferences in Asia Pacific held in Pune, whose Technical Co-Sponsor is the IEEE Bombay Section.