EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
715. Impact of Sentiment analysis on Energy Sector Stock Prices : A FinBERT Approach
Invited abstract in session MA-63: Applications in Finance and Economics, stream OR in Banking, Finance and Insurance: New Tools for Risk Management.
Monday, 8:30-10:00Room: S14 (building: 101)
Authors (first author is the speaker)
| 1. | Rania HENTATI KAFFEL
|
| CES : Centre d'économie de la Sorbonne, University Paris 1 Panthéon Sorbonne | |
| 2. | Sarra Ben Yahia
|
| CES : Centre d'économie de la Sorbonne, Université Paris 1 Panthéon Sorbonne | |
| 3. | José Ángel GARCÍA SÁNCHEZ
|
| CES : Centre d'économie de la Sorbonne, Université Paris 1 Panthéon Sorbonne |
Abstract
This paper provides accurate estimations of the impact of sentiment analysis on forecasting trends in stock prices. We use FinBERT, a pre-trained and fine-tuned NLP model to label and to score a large web-scrapped data (tweets posts and Reddit discussions) of 10 selected energy stock daily returns and the S&P Energy Index spanning from 2018 to 2023. Our approach confirms: i) the accuracy of stock price predictions going up top 85\%. ii) Results are highly sensitive to the test period and confirm the link between risk aversion and negatif or positif score. Also, a novel contribution is provided and consist on employing time series model to capture the temporal dependencies and autocorrelation structures within both the sentiment scores and stock returns data. Our study reinforces the concept of market efficiency and offers empirical evidence regarding the delayed influence of emotional states on stock returns.
Keywords
- Artificial Intelligence
- Forecasting
- Financial Modelling
Status: accepted
Back to the list of papers