Sunday, February 1, 2015

Hedge fund replication - still not there

For those that want low cost alternative investments, hedge fund replication has been research to watch. There have been some significant analysis that has helped with hedge fund style identification, but there still has not been a way to "bottle" the behavior of hedge funds. The replication process still under-performs the hedge fund indices developed.

The recent paper Hedge Fund Replication: A Model Combination Approach provides an improvement over simple linear factor models for replicating hedge funds. They use a novel approach of combining models to generate optimal weights. Unfortunately, the results still show that replication is not possible. Hedge funds generate better returns. Their results table shows the underperformance even with their improved model.



So what makes hedge funds unique? The factors related to a strategy can be found and provide a good first approximation. What cannot be replicated is the dynamic diversification used and the risk management employed. The alpha is in the details of active management and not the broad strategy rules. Hedge funds can change allocations faster than a model which needs past data to measure factors. The factor analysis cannot effectively replicate risk management which provides option like pay-offs. It is the dynamic switching of asset classes and factor with risk management which is the secret sauce.

No comments: