The goal of the master’s thesis was to investigate and evaluate the possibility of developing an active trading strategy based on the sentiment analysis of financial news, and to verify whether this strategy can surpass the return of the passive buy-and-hold approach.
News data were collected for the 20 companies with the largest market capitalization in the S&P 500 index during the 2021–2023 period. Data were sourced from six different financial web portals. To calculate the daily sentiment, the news texts were preprocessed, and the TextBlob library's lexical method was used to determine polarity. The obtained sentiment then served as the foundation for developing and simulating the active trading strategy.
The main focus of the research was the identification of web sources with the greatest predictive potential. The research confirmed a clear connection between news sentiment and stock price fluctuation. The most important finding is the exceptional performance of the strategy using data from the Nasdaq source, which exceeded the return of the passive buy-and-hold strategy across all market periods. Furthermore, the sentiment strategy, utilizing sources like CNBC, BBC, and CNN, proved to be extremely effective during the bear market period.
The master's thesis concludes that an active trading strategy based on sentiment signals from selected sources can match or even surpass the returns of a passive strategy. However, its advantage quickly diminishes if transaction costs cannot be effectively minimized. The work provides a strong foundation for further research on optimizing strategies based on specific sources.
|
