Fit an Independent Regime Switching Model

1 2 |

`R` |
xts object of asset returns |

`numRegimes` |
number of regimes to fit to the data |

`transMatbounds` |
bounds on the parameter tau as described in (Lee, 2010). Each paramter is defaulted to lie in the range (2,10) |

`dccBounds` |
bounds on the paramter theta as described in (Lee, 2010). Each paramter is defaulted to lie in the range (0,1) |

`w` |
proportion of entries to consider in initializing correlation for for each regime. It is defualted to split data equally across all regimes |

`...` |
addition control paramters that can be passed to the control object in DEoptim |

This method takes in returns data and the number of regimes and fits sepearate covariances to each regime using the Expectation Maximization algorithm decribed in (Lee, 2010). IS-DCC model avoids the path dependency problem observed in other regime switching models and makes the solution more tractable by running a separate DCC process for each regime.

Fitting the IS-DCC model to data corresponds works in two steps. In the first step a time varying univariate volatility process, GARCH(1,1) is fitted to each time series. In the second step DCC parameters for each state are estimated along with the transition probabilities corresponding to the Hidden Markov model. This is done by maximising the log-likelihood of observing the residuals

Rohit Arora

Lee, H.-T. (2010). Regime switching correlation hedging. Journal of Banking &

1 2 3 4 5 |

Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.

Please suggest features or report bugs with the GitHub issue tracker.

All documentation is copyright its authors; we didn't write any of that.