Year: 2024
Type: dissertation
Abstract: Given that the stock market for football clubs is heavily influenced by public sentiment, the aim of this thesis is to investigate the predictability of Google Trends data in the context of the Portuguese football clubs, using it as an indicator of public interest. Specifically, it examines the potential relationship between Google search trends and the stock market price. This research involves the development of three models with distinct features, per football club and machine learning model: one utilizing traditional features related to the stock market, another incorporating both these traditional features and Google Trends data specific to each football club, and lastly, one incorporating these two and Google Trends data associated to financial variables. The stock market prices, volume traded, oil price, gold price and euro to dollar exchange rate were extracted from a python library called ‘yfinance’, the Google Trends data was imported from ‘trends.google.com’ and the rest was feature engineered. By comparing the performance of these models, this study seeks to establish whether the inclusion of Google Trends data enhances their predictive accuracy, and which model has the best performance. We will also make the comparison between each football club. The results obtained suggest that the inclusion of Google Trends data regarding football clubs is an improvement to football clubs stock market price predictive models. The inclusion of Google Trends data of economic keywords adds too much complexity to the model, not improving the predictions. The best results, measured with a RMSE metric, were achieved using a Multilayer Perceptron. The findings of this research could provide valuable insights for investors, analysts, and policymakers regarding the usefulness of alternative data sources, such as Google Trends, in predicting stock market movements.
Author: Miranda, João Miguel Dias
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W6995847294