Sentiment-driven stock price prediction: analysing green finance news using large language models for Tesla
Conference paper
Lorvand, A., Yetgin, H., Smith, S., Alkubaisy, D. and Piras, L. 2026. Sentiment-driven stock price prediction: analysing green finance news using large language models for Tesla. Ana Paula Rocha, Mattias Wahde and H. Jaap van den Herik (ed.) 18th International Conference on Agents and Artificial Intelligence. Marbella, Spain 05 - 07 Mar 2026 SCITEPRESS - Science and Technology Publications. pp. 2596-2605 https://doi.org/10.5220/0014306600004052
| Type | Conference paper |
|---|---|
| Title | Sentiment-driven stock price prediction: analysing green finance news using large language models for Tesla |
| Authors | Lorvand, A., Yetgin, H., Smith, S., Alkubaisy, D. and Piras, L. |
| Abstract | The rise of green finance has had a significant impact on the way investors evaluate a company’s performance, prompting them to consider environmental, social and governance (ESG) criteria alongside traditional financial metrics. Nevertheless, research investigating the potential of ESG-related news to improve stock price prediction models is still limited. This study examines the effectiveness of sentiment analysis derived from green finance and ESG-related news to improve stock price forecasts specifically for Tesla Inc. Historical stock price data was combined with sentiment scores extracted from ESG-related news articles using FinBERT. Two recurrent neural network models (RNN) and two long short-term memory (LSTM) models were developed and evaluated, both with and without the inclusion of sentiment features. The results show that sentiment features have only a modest impact on predictions during stable market conditions, while they provide significant improvements during periods of market volatility. Specifically, during a volatile period when Tesla’s share price fluctuated between $340 and $480, the inclusion of sentiment features in the LSTM model reduced the root mean square error (RMSE) from 9.87 to 4.06. This significant reduction underscores the improved accuracy achieved by incorporating textual sentiment data. Overall, these results emphasise the importance and potential benefits of combining traditional financial indicators with textual data analysis, especially in times of increased market uncertainty. |
| Keywords | Sentiment Analysis; Green Finance; Large Language Models; Long Short-Term Memory; Recurrent Neural Network; Artificial Intelligence |
| Sustainable Development Goals | 9 Industry, innovation and infrastructure |
| 13 Climate action | |
| Middlesex University Theme | Creativity, Culture & Enterprise |
| Research Group | Research Group on Development of Intelligent Environments |
| Conference | 18th International Conference on Agents and Artificial Intelligence |
| Page range | 2596-2605 |
| Proceedings Title | Proceedings of the 18th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART |
| Editors | Ana Paula Rocha, Mattias Wahde and H. Jaap van den Herik |
| ISSN | 2184-433X |
| ISBN | 9789897587962 |
| Publisher | SCITEPRESS - Science and Technology Publications |
| Publication dates | |
| 08 Mar 2026 | |
| Publication process dates | |
| Accepted | 19 Dec 2025 |
| Deposited | 09 Apr 2026 |
| Output status | Published |
| Publisher's version | License File Access Level Open |
| Copyright Statement | Lorvand, A., Yetgin, H., Smith, S., Alkubaisy, D. and Piras, L. |
| Digital Object Identifier (DOI) | https://doi.org/10.5220/0014306600004052 |
| Web address (URL) of conference proceedings | https://www.scitepress.org/ProceedingsDetails.aspx?ID=/WDmuZiE4ss=&t=1 |
| Language | English |
https://repository.mdx.ac.uk/item/367zq5
Download files
26
total views10
total downloads1
views this month0
downloads this month