Abstract: Investment decisions are often influenced by emotional biases and unstructured information, leading to misjudgment of risks. This research aims to address this challenge by developing an AI- based portfolio risk analyzer integrated with sentiment analysis of social media and financial news. Using Tesla’s stock as a case study, the project combines statistical models (ARIMA, SARIMA, and LSTM) with Natural Language Processing (NLP) techniques (FinBERT sentiment scoring). The dataset spans one year (30th September 2021 – 30th September 2022), covering daily Tesla stock prices and 5,050 Tesla-related tweets. The first objective is fulfilled by predicting stock price trends using ARIMA, SARIMA, and LSTM models and visualizing their performance comparison. The second objective is achieved by applying FinBERT sentiment scoring on tweets to analyze market perception and its correlation with stock price movements. The results indicate that LSTM provides the most accurate stock forecasting, while FinBERT sentiment analysis successfully captures daily market sentiment trends, which show a noticeable correlation with stock price volatility. These findings demonstrate the importance of combining financial time series models with NLP-driven sentiment indicators to provide deeper insights for retail investors. Keywords: Tesla Stock Prediction, ARIMA, SARIMA, LSTM, FinBERT, Sentiment Analysis, Stock Price Forecasting, Market Volatility, Time Series Analysis, Social Media Analytics, Investor Sentiment, AI-based Risk Analyzer
Modi et al. (Tue,) studied this question.