Extreme Events – Specimen Question A.6.1(a) – Answer/Hints

[this page | pdf | references | back links]

Return to Question


Q. To what extent should you exclude such observations when developing a robust portfolio construction algorithm?


If the observations really are errors then they should be excluded from the underlying return series. In this respect ‘robust’ portfolio construction is not really like ‘robust’ regression in which we (usually) give reduced (but not nil) weight to outliers. Instead, with robust portfolio construction we should clean the dataset as far as possible, and include outliers if they really are valid data points but exclude them if they are not.


Contents | Prev | Next | Question

Desktop view | Switch to Mobile