Tracking Portfolio Optimization

This article introduces a new approach to the tracking portfolio composition. Unlike traditional approaches, it doesn't require benchmark composition to be known and works on any sets of assets. Models presented in the article allow deriving a portfolio composition that results in the optimal value of a tracking performance indicator for the given sets of assets. An S&P 500 tracking portfolio composed of 16 arbitrary selected blue chip stocks generated with the models had in 2010 the annualized TEV of about 4%. Tracking accuracy is significantly affected by frequency of rebalancing and number of assets in portfolio for ex-post tests. Ex-ante tests during the same time period show lower tracking accuracy with the annualized TEV of about 4.4% for the same portfolio, and indicate no benefit in frequent rebalancing