In today’s increasingly interconnected world, financial news can be found through a variety of sources, including social media platforms and/or more traditional media like well-known newspapers. Much research on the impact of Twitter on the stock market has shown its strong ability to influence the market as well as traditional media. Especially in recent years, Twitter has emerged as a major social media platform and generated great interest from sentiment analysis researchers. In this thesis project, Twitter is also used as a social media news source to be explored deeper. Indeed, the objective is to apply different machine learning models for Natural Language Processing (NLP) to both financial news and Twitter posts to get their sentiments. After that, sentiments are compared to explore potential correlations between news and tweets. The models applied are VADER and FinBERT. Understanding the connection between news and the US stock market during the Covid-19 pandemic is another crucial step that has been analyzed. Although news is generally negative and pessimistic during crises, 2020 turned out to be a good year with positive performance for the US stock market. The baseline trend considered for the US stock market is the Standard & Poor’s 500 index. Overall, FinBERT showed much stronger correlation results to the S&P 500 daily returns, compared to VADER. The results of this study could be extremely useful to gain an overview of the wider public opinion behind certain topics and the connection they might have on the market by considering the Standard and Poor’s 500 index. It could also be useful as a complementary tool for companies, traders, and any investors who aim to taking data-driven investment decisions in future pandemics.
Nel sempre più interconnesso mondo odierno, le notizie finanziarie possono essere trovate attraverso una varietà di fonti, comprese piattaforme di social media e/o media più tradizionali come i giornali. Molte ricerche sull’impatto di Twitter sul mercato azionario hanno dimostrato la sua forte capacità di influenzare il mercato così come i media tradizionali. Soprattutto negli ultimi anni, Twitter è emerso come una delle principali piattaforme di social media e ha generato un grande interesse da parte dei ricercatori di analisi del sentimento. In questo progetto di tesi, Twitter viene utilizzato anche come fonte di notizie sui social media da esplorare più nel dettaglio. Infatti, l’obiettivo è applicare diversi modelli di apprendimento automatico per l’elaborazione del linguaggio naturale (NLP) sia alle notizie finanziarie che ai post di Twitter per ottenere i loro sentimenti. Successivamente, i sentimenti vengono confrontati per esplorare potenziali correlazioni tra notizie e tweet. I modelli applicati sono VADER e FinBERT. Comprendere il collegamento tra le notizie e il mercato azionario statunitense durante la pandemia di Covid-19 è un altro passaggio cruciale che è stato analizzato. Nonostante le notizie siano generalmente negative e pessimistiche durante le crisi, il 2020 si è rivelato un buon anno con una performance positiva per il mercato azionario statunitense. Il trend di base considerato per il mercato azionario statunitense è l’indice Standard & Poor’s 500. Nel complesso, FinBERT ha mostrato risultati di correlazione molto più forti con i rendimenti giornalieri dell’ S&P 500, rispetto a VADER. I risultati di questo studio potrebbero essere estremamente utili per avere una panoramica dell’opinione pubblica più ampia dietro determinati argomenti e del collegamento che potrebbero avere con il mercato, considerando l’indice Standard and Poor’s 500. Potrebbe anche essere utile come strumento complementare per aziende, trader e tutti gli investitori che mirano a prendere decisioni di investimento basate sui dati in future pandemie.
Sentiment analysis of financial news released during Covid-19 pandemic
Bressanelli, Giulia
2021/2022
Abstract
In today’s increasingly interconnected world, financial news can be found through a variety of sources, including social media platforms and/or more traditional media like well-known newspapers. Much research on the impact of Twitter on the stock market has shown its strong ability to influence the market as well as traditional media. Especially in recent years, Twitter has emerged as a major social media platform and generated great interest from sentiment analysis researchers. In this thesis project, Twitter is also used as a social media news source to be explored deeper. Indeed, the objective is to apply different machine learning models for Natural Language Processing (NLP) to both financial news and Twitter posts to get their sentiments. After that, sentiments are compared to explore potential correlations between news and tweets. The models applied are VADER and FinBERT. Understanding the connection between news and the US stock market during the Covid-19 pandemic is another crucial step that has been analyzed. Although news is generally negative and pessimistic during crises, 2020 turned out to be a good year with positive performance for the US stock market. The baseline trend considered for the US stock market is the Standard & Poor’s 500 index. Overall, FinBERT showed much stronger correlation results to the S&P 500 daily returns, compared to VADER. The results of this study could be extremely useful to gain an overview of the wider public opinion behind certain topics and the connection they might have on the market by considering the Standard and Poor’s 500 index. It could also be useful as a complementary tool for companies, traders, and any investors who aim to taking data-driven investment decisions in future pandemics.| File | Dimensione | Formato | |
|---|---|---|---|
|
Msc_Thesis_POLIMI_Giulia_Bressanelli.pdf
accessibile in internet solo dagli utenti autorizzati
Dimensione
1.46 MB
Formato
Adobe PDF
|
1.46 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/195050