Non-Parametric Analysis of Hedge Fund Returns: New Insights from High Frequency Data

Abstract

The history of financial research often reflects the history of the changing nature of financial data. Simply put, one has a difficult time analyzing questions requiring empirical data that does not exist. In the area of hedge funds, currently-available databases provide only monthly returns. Moreover, various providers of hedge fund index data follow very different methodologies in structuring hedge fund indices. This article examines four daily hedge fund return indices: MSCI, FTSE, Dow Jones, and HFRX, all based on investable hedge funds, and three monthly hedge fund return indices: CSFB Tremont, CISDM, and HFR, which comprise both investable and non-investable hedge funds. This study, based on standard statistical analysis, non-parametric analysis of the return distribution, and non-parametric regressions with respect to the S&P500 index, shows that key biases like fund selection, asset liquidity,data frequency, sample period, and index construction methodologies are responsible for different statistical properties of hedge fund indices.One key variable that highly affects the statistical properties of hedge fund index returns is the “investability” of hedge fund indices.

TOPICS: Real assets/alternative investments/private equity, big data/machine learning, mutual funds/passive investing/indexing, performance measurement