Readers who have been following the discussion on future return projections will find the following information of interest. Using the Hoadley “Efficient Frontier” software, I set out to see what the optimal projections would be for three different time periods. In each case I selected two years of historical data as that is the default setting within the Hoadley add-in software. The first period ended on 11/30/2007 or near the high before the bear market of the Great Recession. The second period ran from 6/30/2006 to 6/30/2008 so as to capture a projection in the middle of the bear market. The last investigation included 3/31/2007 through 3/31/2009 or the end of the Great Recession.
When one begins as early as 11/30/2005 numerous ETFs are eliminated from the study as they do not have sufficient historical data. I ended up using these twelve ETFs and they are: VTI, EFA, VNQ, TLT, TIP, IJJ, IJK, SHY, IEF, VWO, ICF, and GLD. These ETFs provide a spread of equity and bond instruments.
I removed all constraints within the Hoadley software, something I never do when using this software with real portfolios. Removing constraints on both asset classes and individual ETFs turned the software loose to make its projects without outside influence. As it turned out, the percentage recommendations did not change all that much even though the market was going through turmoil. The “Markowitz” projections are rather consistent as readers will see below.
11/30/2005 – 11/30/2007 (Did not sum to 100% due to rounding error.) Even though the market was about to tank, the projection is to go full bore with equities.
- VTI – 87%
- EFA – 5%
- VNQ – 3%
- TIP – 1 %
- ICF – 4%
6/30/2006 – 6/30/2008 (Even though the market declined for another nine months, the optimized portfolio is still heavily invested in equities.)
- VTI – 74%
- EFA – 7%
- IJJ – 15%
- IJK – 3%
- ICF – 1%
3/31/2007 – 3/31/2009 (At the market bottom equities are still recommended, a correct call.)
- VTI – 83%
- EFA – 6%
- IJJ – 7%
- IJK – 2%
- ICF – 2%
While I did not expect the optimizer to pick up the coming bear market at the end of 2007, I did expect to see more of a shift in the recommendations in June of 2008. However, there was no shift to less volatile securities such as IEF, TIP, and TLT in June of 2008.
The next test is to use the CWM software with these same ETFs to see if the recommendations provide additional constraints so the draw-down is modified. It is hard to see the CWM bungling the call as much as the optimizer did.
Discover more from ITA Wealth Management
Subscribe to get the latest posts sent to your email.
Lowell, I guess this doesn’t surprise me. I don’t think there is anything in the Markowitz formulation that protects from bear markets. It is simply optimizing on the past return data and the past volatility.
This is where I would expect the “trend following” or triggers (or whatever one wants to call sell triggers) that are set by using SHY and/or the 190 day EMA would save one’s bacon to a large degree, as shown by HedgeHunter’s Feynman Studies in the backtesting. Oh, that I had had such triggers set in 2007! I would be able to take a trip around the world at least once on the money I would have saved and then profited from by getting back end after the bottom based on EMA signals.
Lowell, I just reread this study. What were you optimizing on? A “market” portfolio (maximizing the Sharpe ratio)? Minimum risk (minimizing the portfolio volatility)?
Ernie