Comparing Chronos-Bolt and ARIMA Models for Portfolio Optimization in the S&P 500 Information Technology Sector 💲 💵
Abstract
Stock return forecasting is critical for portfolio optimization in financial markets. Traditional models like ARIMA offer interpretability but struggle with nonlinear dependencies and long-term trends. In contrast, Chronos-Bolt, a transformer-based model, leverages pretraining and probabilistic forecasting to improve accuracy. This study compares ARIMA and Chronos-Bolt for predicting S&P 500 IT sector stock returns, evaluating Chronos-Bolt with technical indicators and sentiment analysis from social media discussions.This research demonstrates the potential of transformer-based forecasting and sentiment integration in financial decision-making.
Table of Contents
- Background & Research Question
- Data Collections
- Methodology
- Results
- Findings & Implications
- Limitations and Future Directions
Code & Poster
💻 You can find our Github respository here
🪧 You can find our poster here