Tuesday, January 13, 2009

Endowment Style Portfolios for December 2008

When I tried to optimize the performance of the asset mix used for the portfolios in my post about endowment style portfolios by maximizing the Sharpe ratio using historical data I ended up with a 100% allocation to the real estate fund. But this fund lost 7% in December. Now the optimal allocation is 9% managed futures, 4% hedge funds, 66% real estate, 20% bonds, and 1% gold. It has a beta of 0.03 to the MSCI stock index. No stocks of course as they have returned too little with too much volatility in the last 12 years. A portfolio with 1/7 in each of these assets plus Australian Dollars has about the same returns historically but double the volatility (but 1/3 the volatility of stocks still). It would have returned 2.7% in December in USD terms (it has a beta of 0.28 to the stock market)


Following the discussion in the comments I want to say that this isn't intended as a serious exercise in choosing a portfolio allocation but as a kind of thought experiment about what would have historically been the best portfolio with perfect hindsight. I'm pretty skeptical also about so-called "forward looking" portfolio optimizations too. They need to make some pretty strong assumptions. But all of this can be useful inputs into developing a portfolio allocation that works for you.


enoughwealth@yahoo.com said...

As they say, historical data is no guarantee of future performance. In fact it's not a very good estimate of future performance and volatility in most cases. If you limit the period of your "history" to the past 10, 20 or 30 years, you are liable to overweight in any asset class that is experiencing a bubble. And if you use very long term data (say 100 years of stock market data), you have problems with economic, investment and accounting changes making the oldest data irrelevant to current and future behaviour.

If you run the modelling using 10 different time periods (ending 1-10 years ago), what range of allocations and performance/volatility results do you get?

That sort of sensitivity analysis will at least give you an idea of whether you should be talking about "66%" real estate, or just think in terms of "30%-80%" real estate.

mOOm said...

Maybe I should have made clear that this kind of "analysis" shouldn't be used for designing an allocation. Just curious what would have been the highest Sharpe ratio portfolio....