Monday, October 10, 2016
Drawdown-based risk parity - An improvement
Risk parity has been one of the most important advancements in portfolio construction over the last decade. It places the focus on equalization of risk and not on comparing expected returns which are notorious for being difficult to forecast. However, there is a problem with risk parity. It will allocate more to low risk asset classes that may be subject to a downturn. There is no opinion on market direction or valuation. It assumes the investor has no information on individual assets or view of factor risks. Outperformance relative to cap weighted portfolio is related to whether low risk assets have returns than higher risk assets.
A recent article, Investor Views, Drawdown-based Risk Parity, and Hedge Fund Portfolio Construction by Alex Rudin and Bill Marr tries to solve this problem with risk parity. The authors suggest a simple switch to expected maximum drawdown (EDD) over risk parity. What makes this attractive is that EDD uses the exact same framework as risk parity. In fact, risk parity may be considered a special case of EDD when there are no opinions on prices.
What EDD proposes is that investors should focus not on volatility, but on the maximum expected drawdown. The maximum drawdown will be affected by market direction; consequently, the market expectations can be used to measure expected drawdown. If you have a positive view on the stock market, the expected downside risk or drawdown over some horizons will be shifted up by some linear amount related to your expectation. Similarly, if you have a negative market view, your expected drawdown will get worse. Hence, you can directly use your expectations to place a tilt on drawdown exposure relative to the no view (simple risk parity) case. The authors center on the idea that expectations can be focused on factors such as market betas to provide the drawdown tilts.
What truly makes the EDD framework useful is that it is simple to implement and just an extension of what investors currently do with risk parity. Of course, there is no way around the problem of forecasting skill, but even a simple approach based on trend may help provide appropriate market tilts to control maximum drawdown.