TURBULENTRISK
A robust statistical risk model using the Mahalanobis distance.
Last updated
A robust statistical risk model using the Mahalanobis distance.
Last updated
Using the Mahalanobis distance, this risk model characterizes the degree of unusualness in a cross section of asset returns and partitions the historical data into periods of financial turbulence and quiecense. These periods are usually marked by large asset movements (volatility) and/or unusual changes in correlations (e.g. when non-correlated assets become correlated) and considered statistical outliers.
The risk model is useful for stress testing and constructing resilient portfolios.
The following describes the function signature for use in Microsoft Excel's formula bar.
type
Required. Enumeration string to specify calculation type: "risk", "sigma", or "stdev" "correlation", "corr", or "rho", "covariance", "covar", or "cov"
assetReturns
Required. Time series or matrix of asset returns.
threshold
dataPeriodicity
Optional. Periodicity of the data, used for annualization. If you do not enter the argument, it defaults to 1. e.g. Daily = 255, Monthly = 12, Yearly = 1, Quarterly = 4.
Depending on the specified output type, the function will return the respective vector of risk estimates (annualized standard deviations), correlation matrix, or a covariance matrix.
The following video describes the conceptual application of this methodology.
Required. Probability threshold of turbulent periods (0.00 - 1.00). This threshold is the converted into the equivalent chi-squared,, score. Under a multivariate normal assumption, the cutoff can be interpreted as an approximation of the percentage of the sub-sample of outliers (turbulent). This approximation may vary depending on the underlying characteristics of the empirical data.