Analyzing market dynamics

: a comprehensive model using macroeconomic, sentiment and fundamental data for regime detection and asset allocation

Abstract

This dissertation examines the development and application of a comprehensive financial indicator that incorporates macroeconomic, sentiment, and fundamental data. It is demonstrated that various market features, such as high mean returns and low volatility periods, can be clustered into regimes using well-known models like the Two-State Markov Switching Model and Principal Component Analysis. Important discoveries are obtained by combining the findings in a Combined Indicator. For example, both the Dotcom Bubble and the Global Economic Crisis provided early signals of a downturn based on the classification of regimes. However, when applying these models to investment strategies, further results from the out-of sample dataset suggest that, despite the potential of certain individual models to improve market timing and achieve greater risk-adjusted returns in the training set, these cannot be replicated consistently. Realistic market conditions, such as the inclusion of a time lag due to signaling and the introduction of transaction fees, thus limit the viability of an effective investment strategy. Nevertheless, the results of the dissertation and the prediction of the Combined Indicator itself can be useful and serve as support for asset allocation.

Date of Award	2 Jul 2024
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Dan Tran (Supervisor)