Modeling volatility regimes in the Swedish stock exchange using a Hidden Markov Model

Sammanfattning

This thesis investigates whether financial market volatility can be describedas a regime-switching process rather than a constant or continuously evolvingone. Financial time series are characterized by volatility that varies over time.Understanding such dynamics is important to gain a better understanding offinancial markets. The study focuses on the OMXS30 index and examineswhether distinct and persistent volatility regimes can be identified from dailylog-returns. To address this, a Hidden Markov Model is applied. The Baum Welchand Viterbi algorithms are used to estimate the model parameters and inferthe most likely sequence of volatility states. Several model specifications arecompared to identify the best number of regimes.The results are compared with rolling volatility and GARCH to assesshow the identified regimes relate to continuous models. The results showthat HMM identifies multiple regimes mainly differentiated by their level ofvolatility. These regimes are persistent and generally align with the patternsobserved with the continuous models. The HMM also provides a transitionmatrix that provides insight into how the identified regimes evolve overtime. The results are discussed in relation to potential implications for riskassessment and decision-making under changing market conditions.