top of page


Headlines and Market Trends: A Sentiment Analysis Approach to Stock Prediction

Jem Guhit, Sarasi Jayasekara, Nawaz Sultani, Timothy Alland, Ogonnaya Romanus, Kenneth Anderson


Financial markets are often affected by sentiment conveyed in news headlines. As major news events can drive significant fluctuations in stock prices, understanding these sentiment trends can provide important insights into market movements. This proposed project aims would like to answer the question whether the sentiments extracted from financial news headlines can effectively predict stock movements. The proposed structure for this project that can be done in the span of this bootcamp would be,
Data sources:
- For financial news, news APIs like Bloomberg, CNBC, and Reuters (other suggestions welcome)
- For stocks, currently YFinance for historical stock information (other suggestions welcome)
Planned Approached:
- Baseline: Average sentiment model -- will use average daily sentiment score from financial news to predict either a downward or upward movement in the stock price
Initial models to check:
- Model 1: Logistic regression
- Model 2: Support Vector Machines
- Model 3: Random Forest Classifier
Other models of interest: Boosted Decision Trees/Arima/LSTMs
With time constraints, I initially thought that we could use sentiment scores as the primary features in the above models to predict stock movement to see the direct effect of sentiment scores. Then later as an extension we can include other relevant variables in the analysis.

Screen Shot 2022-06-03 at 11.31.35 AM.png
github URL
bottom of page