An adaptive regime-switching regression model for hedge funds

We propose a switching regression model for hedge funds to capture the characteristics of trading strategies through time. The coefficients are governed by a discrete-time Markov chain and are able to switch between regimes. The states of the Markov chain represent different states of the economy. Hedge fund indices from main trading strategies and market indices are chosen as regressors. A filtering technique by Elliott (1994) is used to filter out hidden information and optimal parameter estimates are derived through a filter-based Expectation–Maximization algorithm. Our switching regression model is applied to individual hedge fund series from the Hedge Fund Research database.