Sentiment Analysis in the Norwegian Stock Market: Predicting Stock Price Movements Using Media Sentiment

Abstract

A historical belief in financial economics states that stock prices react immediately to available information, making it impossible to predict changes in stock prices. However, research in the last few decades contradicts this theory to a degree. Some research has been done on whether the sentiment or “tone” of financial media can be used to predict stock price movements, although it has often been focused on U.S. newspapers and markets. This thesis will analyse Norwegian news articles from the online newspaper Dagens Næringsliv and stock prices from companies listed on the Oslo Stock Exchange, or Oslo Børs, to explore whether a relationship can be found between the sentiment of news articles and stock price changes. Sentiment analysis will be used to identify the level of positivity or negativity in the news articles and four statistical methods will be performed to attempt to predict stock prices using media sentiment. The methods are logistic regression, K-nearest neighbors, gradient boosted trees and support vector classifier. The accuracy of using media sentiment to predict stock price changes varies from 53.69% to 57.38% using the four methods. These results are in line with the accuracy that previous research on U.S. newspapers and companies has found, suggesting that there is a weak but significant relationship between the sentiment of Norwegian news articles and the stock price movements of Oslo Børs-listed companies.