Like stratified sampling, optimization is a technique for building an index-based portfolio using only a subset of the securities in the index. Optimized portfolios use a multi-factor risk model to measure the securities exposure to a variety of risks, then apply an objective function to specify the specific securities and weights such that tracking risk relative to the index remains within pre-defined levels. The risk factors may include beta, market capitalization, industry and even exposure to broader risks such as interest rates.
A major advantage of optimization compared with stratified sampling is that optimization considers covariances among the risk factors. A stratified sampling approach implicitly assumes that any two risk factors are unrelated to each other.
Disadvantages of optimization lie primarily in the fact that backward-looking models are subject to changing risks over time, and even in a constant risk environment the model can never be perfectly specified. Furthermore, as relative risks change over time the optimization function can require frequent trading and raise transaction costs.
Still, optimization tends to offer reasonable performance, and many index funds use full replication to match the largest and most liquid securities in a portfolio while using stratified sampling or optimization to more generally match the rest of the group.Asset Allocation, Fixed income investments, Investing in Stocks, Investing in bonds, Investment Returns, Portfolio Management See also: