Correlation versus Cointegration: Do Cointegration based Index-Tracking Portfolios perform better? Evidence from the Swedish Stock-Market

Passive portfolio management which aims to replicate a stock index faces basically two different optimization methods. Traditional portfolio management employs historical stock return data of preselected stocks in order to replicate the underlying stock index. The cointegration method employs time series data of stock prices instead, even though stock price data may statistically often exhibit random walk behavior. In this review the advantage of the latter method could be asserted. Thereby, different stock portfolios with respect to the Swedish stock market are constructed which rest upon both, the concept of correlation and the concept of cointegration. The cointegration based models dominate, which can be ascertained by comparing their Sharpe ratios as well as their Treynor ratios. The dominating stock portfolio beat the index by 79.08% within the overall 10-years out-of-sample period, whereas the annual volatility on average was 1.10 base points lower.