Why overlays?

During transition periods from growth to risky environment it is not straightforward to manage the risks carried by alternative allocations. This is especially the case for hedge fund allocations, which exhibit low level of liquidity (quarterly or monthly at best) and which exhibit significant inertia. In the case of UCITS or managed account allocations which are more liquid, market risk management can be tricky in the short term. Beyond the liquidity issue, large scale rebalancing can cause operational issues: significant concentration of the allocation in certain investment styles, frequent in and outs can be problematic with certain managers, exiting from closed funds...

The implementation of overlays can be a good alternative to large scale rebalancing. This type of hedging technique lies on the identification of the systematic risks embedded by the underlying funds, and by the implementation of effective hedges with liquid instruments (future contracts). The first advantage of this approach is that it allows the investor to keep its allocation unchanged. The hedge can then be triggered, waiting the confirmation of a trend which would validate the reallocation of the portfolio. The second advantage is that this approach does not require high amounts of cash (deposit x beta). Finally, the overlay can be priced in real time and be liquidated at anytime.

Discretionary vs. systematic overlays

In addition to the problems inherent to the identification of the risk exposures and the choice of hedging instruments, the timing of the trigger of the overlay is a central issue. Two options are available to investors and multimanagers. The first is based on the discretionary trigger of the overlay, depending on the investor views. From our point of view, this solution is not efficient, because medium term views are already reflected in the tactical allocation. The second approach is based on the use of a systematic signal to trigger the overlay. The major advantages of the latter are the objectivity of this approach and the ability to set up contrarian views.

Implementation

The overlay process requires the use of specific quantitative tools:

• a risk signal, which takes into account standard risk metrics (volatilities, credit spreads), but also second order risk metrics (dispersion, crossasset correlation...) ;

• a multifactor style analysis model, designed to identify the pertinent risk factors for each underlying fund, to take into account changes in risk exposures (tactical allocation, style drift...) and which allows the nonlinearity of the underlying risk structure..

See « A Dynamic Style Analysis Model for Hedge Funds », Research paper, Orion Financial Partners, 2011 and « A Non-Parametric Test of Market Timing for Hedge Funds: Beyond Alpha and Beta », Research paper, Orion Financial Partners, 2011.

• a transition matrix which links the identified risk factors with hedging instruments.

Implementing an overlay

Limits and risks

 In the case of poor market timing, the (non) trigger of the risk signal can be a source of loss. Note however that in the case in which the overlay is not triggered when needed, the performance of the portfolio will not be affected. At the opposite, i.e. if the overlay is triggered in the wrong timing, the performance of the portfolio will be cut from a part of its systematic component. In that case, alpha will not be affected.
 The quantitative analysis of the risk factors may imply a missspecification of the risks to hedge.
 Certain risks can only be partially hedged (high yield debt, emerging equities, commodities...).
 The changes in risk exposures imply that the computed hedge ratios can diverge significantly from optimal hedge ratios.
 Specific risks (i.e. independent from market risks) cannot be identified, nor hedged. As a consequence, overlays are not efficient when applied to strategies which exhibit no or limited exposures to systematic risks (especially relative value strategies).

By applying our model of systematic overlay on the Global HFRI index over the period January 2008 - August 2011, we show that the annualized performance of the benchmark portfolio is improved by 4.4%, the cumulative increase in performance over the period is estimated at 15%; the volatility level is reduced by 2%; the maximum drawdown is divided by 2.5 ; the return distribution is « normalized », i.e. skewness and excess kurtosis are estimated around 0 v.s. negative skewness and significant excess kurtosis without the overlay.

Systematic Overlay

Example: implementation of a systematic overlay on the HFRI Global index, from January 2008 to August 2011