Do Blockchain Competent Investors’ Sentiments Drive Bitcoin Volatility: A Machine Learning and Nonlinear Analysis?
This paper extends the behavioral finance literature on the relationship between investor sentiment and bitcoin volatility in two ways. First, we propose a natural language-processing method to measure Blockchain-competent investor sentiment, while extracting information from Reddit forums. Second,...
Guardat en:
| Autors principals: | , , |
|---|---|
| Format: | Article ou chapitre numérique |
| Idioma: | Français |
| Publicat: |
2026
|
| Matèries: | |
| Accés en línia: | Accès Université d'Orléans et IFPM Accès Université d'Orléans et IFPM |
| Sumari: | This paper extends the behavioral finance literature on the relationship between investor sentiment and bitcoin volatility in two ways. First, we propose a natural language-processing method to measure Blockchain-competent investor sentiment, while extracting information from Reddit forums. Second, we extend the Heterogeneous Autoregressive Realized Volatility(HARRV) model to a nonlinear context, using a Markov switching approach; and we investigate the effect of Blockchain-competent investor sentiment on bitcoin volatility over the period 2018-2023. Further, using FinBERT (Financial Bidirectional Encoder Representations from Transformers) test, we comparatively analyze the effect of Blockchain-Competent (BC) and Non-Blockchain-Competent (NBC) investors’ comments on Reddit on Bitcoin’s volatility. Accordingly, we find that Blockchain-competent investor sentiment has a significant and nonlinear effect on bitcoin volatility. In addition, the distinction between positive and negative sentiment, as well as Blockchain-competent investor sentiment and Blockchain-non-competent investor sentiment significantly improves the forecast of bitcoin volatility. JEL Classification: C2, F10, G10 |
|---|