The Consistency of Self-Declared Hedge Fund Styles — A Return-Based Analysis with Self-Organizing Maps

While hedge funds have common features, they remain an extremely diverse asset class. Despite this diversity, a consistent classification system is important for numerous purposes such as portfolio construction, performance attribution as well as risk management. This topic is also connected to the financial stability debate, which has recently dealt intensively with the issue of hedge funds. Diversified (fund) portfolios with an appropriate risk monitoring system in place will e.g. enhance risk-sharing among financial market participants. As fund self-declaration is prone to strategic misclassification, return-based taxonomies grouping funds along similarities in realized returns can be used to avoid this pitfall. In this paper we use Self-Organizing Maps (SOM) to find homogeneous groups of hedge funds based on similar (return) characteristics. Based on this technique, we can identify nine hedge fund classes. Whereas managed futures, sector financial and short-sell hedge funds are largely consistent in their self-declared strategies, we detect a number of declared hedge fund styles displaying no or very limited return similarities. Especially the so-called _equity hedge_ style encompasses too many different substyles with different return characteristics. Another important aspect that our paper addresses is the tendency of fund managers to perform undisclosed changes of their trading style or to strategically misdeclare their funds. Our results show that so called _style creep_ is an issue in the hedge fund business, with funds which misclassified themselves once being very likely to change their trading style again.