Anca Dimitriu

Indexing and Statistical Arbitrage

There are two basic methodologies for portfolio optimization: tracking error variance (TEV) minimization (the industry standard for indexing), and a cointegration–optimal strategy (advocated by econometricians). Cointegration is a statistical tool that seeks to exploit a long–run equilibrium relationship between a portfolio and a benchmark, ensuring that the two are connected in the long term. For simple index tracking, the additional feature of cointegration is found to provide no clear advantages or disadvantages over TEV. Both models produce optimal portfolios that outperform a price–weighted benchmark during market crashes, assuming a long enough model calibration period. When tracking becomes more difficult, ensuring a cointegration relationship enhances performance. Cointegration–optimal portfolios dominate TEV equivalents for all the statistical arbitrage strategies based on enhanced indexation in all market circumstances.

Detecting Switching Strategies In Equity Hedge Funds Returns

Equity hedge funds are thought to effectively operate market timing by implementing switching strategies conditional on market circumstances. In this article the authors use only the reported monthly returns on a set of funds to infer the type of switching strategies they follow, if any, as well as their switching times. A set of regime-switching models for each equity hedge funds returns against various benchmarks are estimated; subsequently the authors attempt to answer the following general questions: What proportion of equity funds seem to have switching strategies in place? Which are the most popular instruments for switching strategies? And what is the relationship between the switching times of different funds? The general methodology applied in this article may be useful to investors who wish to detect, from only reported returns, whether and when a particular fund has been timing the market.

Sources of Outperformance in Equity Markets

The general portfolio construction model described here is designed to replicate the first principal component of a group of stocks instead of a traditional benchmark, thus capturing only the common trend in stock returns. Reduction of the noise in stock returns facilitates the replication considerably, and the optimal portfolio structure is very stable. Analysis of the portfolio performance over different time horizons in European and U.S. equity markets reveals a time-varying structure. Throughout most of the period studied, the portfolio’s value component dominated the market and the volatility components, but during the volatile periods of the last few years the strategy earned a significant volatility premium. One explanation for the mean-reversion is behavioral; portfolio performance is influenced by the extent of investor herding toward the common trend in stock returns.

Equity indexing: Optimize your passive investments

Carol Alexander and Anca Dimitriu discuss two strategies for enhanced index tracking designed to best suit a passive investment framework.

Rank Alpha Funds of Hedge Funds

This article examines the performance of hedge fund portfolios when fund selection is based on the rank of a funds alpha, rather than the estimated value of the alpha. Estimated alphas can vary significantly depending on the model used and hence induce a high degree of model risk in portfolio optimization. The authors find, even for the simplest factor models, that ranking funds according to their alpha estimates is an efficient selection process. In an extensive out-of-sample historical analysis, funds of funds that are selected in this manner are shown to outperform an equally weighted index of all funds, minimum variance portfolios of randomly selected funds and portfolios that are optimized using estimated alpha values. Of the four factor models considered here the best out-of-sample performance is obtained using the rank alphas from the principal components factor model.

Rank alpha funds of hedge funds

This article examines the performance of hedge fund portfolios when fund selection is based on the rank of a funds alpha, rather than the estimated value of the alpha. Estimated alphas can vary significantly depending on the model used and hence induce a high degree of model risk in portfolio optimization. The authors find, even for the simplest factor models, that ranking funds according to their alpha estimates is an efficient selection process. In an extensive out-of-sample historical analysis, funds of funds that are selected in this manner are shown to outperform an equally weighted index of all funds, minimum variance portfolios of randomly selected funds and portfolios that are optimized using estimated alpha values. Of the four factor models considered here the best out-of-sample performance is obtained using the rank alphas from the principal components factor model.

Abnormal Returns in Equity Markets: Evidence from a Dynamic Indexing Strategy

This paper investigates the abnormal return generated through a dynamic equity indexing strategy and the extent to which this can be considered evidence against the efficient markets hypothesis. We introduce a new measure of stock price dispersion and show that it is a leading indicator for the abnormal return, where their relationship is based on a switching process of two market regimes. The entire abnormal return is associated with only one of the regimes and this is the prevalent regime during the last few years. The predictive power of the model is demonstrated over different time horizons and in different, real world and simulated stock markets. The strategy remains profitable even after introducing transaction costs, thus proving evidence of temporary market inefficiencies.