Portfolio selection under changing market conditions

A Fund of Hedge Funds Under Regime Switching

“…The main difference in performance occurs in investment period 111, and arises from the early crisis recognition. The fact that investors using regime-switching models can react better to the crisis than the BS investor is in line with the findings in Ernst et al [2009] and Bernhart et al [2011], and is largely owing to the explicit assumption of two different market regimes.…”

“…In the high volatility regime (State 1), the MSCI World Index exhibits a negative mean, whereas the low volatility regime (State 0) has a positive mean return. This observation of an increased volatility in bearish times is consistent with the work of Ernst et al [2009], Hross et al [2010, and Maheu and McCurdy [2000]. In the following, we associate State 0 with a normal market regime and State 1 with a crisis.…”

“…Using almost 30 years of daily data, the resulting estimated crisis periods are highlighted in Exhibit 1. For the purpose of comparison, the exhibit also contains the crisis periods derived by the heuristic algorithm, which is proposed by Ernst et al [2009] and adopted in the estimation procedure as an initial guess. The crisis identifications made by the MSM and GMM are reasonably close to the heuristic scheme, but crisis periods determined by regime switching are often split up and interrupted by normal periods.…”

“…This routine employs the Maximum Likelihood approach described in Baum et al [1970]. We apply the heuristic crisis-detection algorithm proposed in Ernst et al [2009] in order to find realistic initial values for the Maximum Likelihood estimation. 10 After estimating the model, the most likely state sequence of S t can be inferred by applying Viterbi's Algorithm; see Viterbi [1967].…”

Asset Correlations in Turbulent Markets and the Impact of Different Regimes on Asset Management

“…Second, transition probabilities could be modelled dynamically and linked to financial and macroeconomic indicators in order to be able to predict financial crises to a certain extent (see, e.g., Ernst et al (2009)). First, other asset return distributions in different market regimes could be incorporated to enhance the fitting capability.…”