Testing the Robust H-CLUB with Synthetic Data

We test the quality of the Robust H-CLUB by generating synthetic data and calculate estimates

The H-CLUB was originally proposed by Fan et al. in the paper "Robust Inference of Risks of Large Portfolios"
We use two different scenarios - the normal condition and the stress scenario.
In the normal condition we just test the quality of results when the data is according to the necessary conditions. In the stress scenario we introduce strong serial dependence, outliers etc. such that the conditions for the algorithm to work are not fulfilled anymore.
We then test how well it behaves under such strong conditions.