Last week, Elvys Linhares Pontes, an NLP Engineer from Trading Central’s innovation unit, TC Labs, took the stage at the CORIA-TALN Conference in Marseille, France, to share the team’s latest research on the role of sentiment analysis in trading strategy development.
The presented study, titled "Backtesting Sentiment Signals for Trading: Evaluating the Viability of Alpha Generation from Sentiment Analysis", is a collaboration of TC Labs data scientists—Elvys Linhares Pontes, Mohamed Ben Jannet, and Yuxuan Zhao—with researchers from University of Tours, University of La Rochelle, and University of Toulouse.
In the study, the team backtested sentiment signals extracted from financial news articles on Dow Jones 30 stocks, comparing the performance of sentiment-based strategies against a benchmark Buy & Hold approach. Their findings revealed that sentiment analysis—especially when using regression models that capture sentiment on a continuous scale, such as Trading Central’s proprietary sentiment model—can significantly enhance trading performance, outperforming traditional strategies. The research also points to further potential by integrating sentiment with macroeconomic and technical indicators.
The CORIA-TALN Conference (Conférence en Recherche d’Information et Applications et sur le Traitement Automatique des Langues Naturelles) is a leading event in the field of Natural Language Processing (NLP), bringing together researchers, practitioners, and students to discuss the latest developments and innovations in various aspects of NLP.
Read the full research article by TC Labs here.
Powered by proprietary AI and natural language processing developed by TCLabs, TC Market Buzz® and TC Crowd Insight® analyze over 100,000 professional news articles, social media posts and blogs to uncover market-moving developments, gauge crowd sentiment and deliver actionable insights across more than 50,000 instruments. These tools help investors cut through the noise and make confident, informed decisions.
