Tuesday, September 14, 2010

Slightly Overcooked?

This is not the first time I have written about (whined about?) managing the volatility of Pinnacle portfolios when we insist on owning a diversified group of asset classes. We don’t have a dial to turn to adjust portfolio volatility, which is heavily dependent on the day-to-day correlations of the various asset classes we own. I just reviewed our performance for the first week of September, where the S&P 500 Index is already up about 5%, and the Pinnacle Moderate Growth portfolio captured about 40% of the move to the upside. The term used to describe relative performance to a benchmark is “beta,” so in this case the DMG portfolio had a beta of about 0.40 for the week. On down days the portfolio beta appears to be considerably less where we have observed downside capture of only 20% - 30% (beta of 0.20 – 0.30) of the broad market. The good news is risk to principal is clearly being managed in this market environment. The bad news is that we may have “overcooked” the amount of risk that we took out of the portfolios over the past few months.

We use several models to determine what portfolio volatility is likely to be based on the current mix of asset classes in our portfolio. One model that is intuitively obvious is a “beta” model that simply looks at the weighted beta of all of the securities in order to estimate the total portfolio beta. The problem is, of course, trying to estimate correlations. Two asset classes can have a the exact same beta relative to the market, but if the are zigging and zagging differently, then the total portfolio beta can be significantly less than a simple beta model might predict. We also use our “equity-like” models where we look at the risk attribution of each asset class in order to come up with our total equity or “risk” exposure. Because we own several asset classes that presumably have low correlations to the broad market, the equity-like models must be continually adjusted to reflect the current performance of the asset class and our subjective view of what correlations may be like in the future. The DMG portfolio current equity-like total is 61% of the portfolio, which implies that portfolio volatility could be substantially higher than we’ve been seeing if we get peak correlations going forward.

The spread between our “equity-like” calculation of 61% for DMG and downside beta of only 0.20 – 0.30 observed in down markets is troubling. I am certain that this will be one of the major areas of concentration for the investment team this week. At the moment the actual observed volatility in Pinnacle accounts implies a severely bearish view that is not consistent with our actual forecast. However, if correlations peak we may see an entirely different level of volatility in managed accounts. It should make for some interesting discussions.