Sentiment scores.xlsx

Investigation of the short-term impact of financial news sentiment on stock price movements using Apple Inc. (AAPL) as a case study. Stock data from 2020 to 2025 was cleaned using S&P 500 base index adjustment and trend-based nor- malised to isolate company-specific effects. Sentiment analysis was performed using two transformer-based models: DistilBERT fine-tuned on casual language (DBERTsst) and FinBERT fine-tuned on financial language. The study measures the correlation between sentiment scores and adjusted stock price changes over short time frames (3 and 7 days), finding a significant relationship at 3-day intervals, with diminishing results when increasing interval size. Linear regression models us- ing sentiment as independent variable, achieved R2 values above 0.63, explaining over 60% of price variance in the short term. Results suggest significant connec- tion between sentiment and market movement. However, this connection is greatly impacted by noise, limiting the usefulness for meaningful predictions.