UCLA

Portfolio resampling is a new approach to portfolio optimization. It generates a higher degree of diversification and smoothness on classic Markowitz's model, which in turn reduces risk and enhances forecast ability. This paper examines this technology on three financial models: single index model, constant correlation model, and multigroup model. Two sets of constraints regarding short sales are also implemented on the models above.

The analysis starts with the price data of 30 stocks and a market index. On the risk-return space, a concept called efficient frontier is introduced. Efficient frontier enables us to determine a set of optimal portfolios by setting a wide range of hypothetical risk-free rates. Some of the restrictions from the original assumption are also discussed. Through Monte Carlo simulation, we are able to visualize the resampling effect. The benefits and the limitations of this technique are put into the context of statistical analysis.