Richard O. Michaud, David N. Esch and Robert O. Michaud
The when-to-trade decision is a critical yet neglected component of modern asset management. Typical rebalancing rules are based on suboptimal heuristics. Rebalancing is necessarily a statistical similarity test between current and proposed optimal portfolios. Available tests ignore many real world portfolio management considerations. The first practical test for mean-variance optimality, the Michaud rebalancing rule, ignored the likelihood of information overlap in the construction of optimal and current portfolios. We describe two new algorithms that address overlapping data in the Michaud test and give examples. The method allows large-scale automatable noncalendar based portfolio monitoring and quadratic programming extensions beyond portfolio management.