Random Portfolios: Practice and Theory

Main points:

• the history of random portfolios extends at least to Chicago 1965
• testing a fund versus a benchmark takes many years to get reasonable statistical power
• a benchmark will be easier to beat some years versus others
• peer groups depend on the fantasy that there is little noise in the rankings of the peers
• there are open probability questions about random portfolios
• random portfolios are sometimes used like the statistical bootstrap, sometimes like a random permutation test

annotated slides (pdf)

Presented 2009 April at the R/Finance Conference