Bitcoin price sentiment analysis and forecasting: Integrating VADER and LSTM models

Abstract

Sentiment analysis and forecasting play a crucial role in understanding market emotions and their trends. When there is a high volatility or variability in the market, it becomes quite difficult for investors and market participants to predict market trends. In this study, VADER analysis is used for sentiment analysis, and the long short-term memory network (LSTM) model is applied for forecasting to predict Bitcoin prices during volatile periods, including the COVID-19 pandemic and the Russia–Ukraine war. In this research, we analyze the daily closing prices of Bitcoin from 2020 to 2023, extracted from Twitter news. The data are divided into two sub-periods: the first from 2020 to 2022, covering the COVID-19 period and the second from 2022 to 2023, covering the Russia–Ukraine war. This study aims to determine patterns and fluctuations in Bitcoin price sentiment over the observed time period and explain how sentiment dynamics are linked to Bitcoin price movements under erratic market conditions. The growing influence of social media on financial markets underlines the significance of this analysis. This study delivers valuable insights for investors, analysts, market participants and policymakers to manage their portfolios and mitigate risk during volatile periods.