I am starting a new “live” portfolio that I have named the “Rutherford” Portfolio. Unlike the Hawking Portfolio that considers 100 ETFs for possible inclusion in the portfolio and uses Cluster analysis to narrow the list down to 4 ETFs, the Rutherford Portfolio will draw from a small list of only 10 ETFs (plus SHY) from which the portfolio will be constructed.
The method used for portfolio construction/asset selection is a momentum strategy that utilizes the “standard” momentum ranking sheet (available for download on this site) with which Platinum members will be familiar. “Standard” momentum periods of 91 and 182 (calendar) days ( 3 and 6 months) are used but the volatility (mean variance) period is changed from 10 to 63 (trading) days. Reasons for this are provided below. The relative weightings of the 3 parameters (to determine the “Overall Rank”) are kept “standard” at 50% (ROC1), 30% (ROC2) and 20% (Volatility).
This will be a VERY SIMPLE portfolio in that it will comprise only TWO ETFs (providing they rank higher than SHY; i.e. SHY momentum is used as a Risk Management filter.
In addition, the Asset List for the Rutherford Portfolio is based on the Swensen 6 portfolio that has been the subject of many posts on this site and should be familiar to regular readers.
ETFs used to represent the six “Swensen” asset classes are:
- VTI – Domestic (US) Equities;
- VEA – International, Developed Market Equities;
- VWO – Emerging Market Equities;
- VNQ – US Real Estate;
- TLT – US Treasury Bonds;
- TIP – US Treasury Inflation- Protected Securities.
In order to add more diversification I have added 4 additional ETFs:
- RWX – International Real Estate;
- PCY – Emerging Market Bonds;
- GLD – Gold (inflation hedge);
- DBC – Broad Commodity ETF (Oil & Gas, Agriculture, Metals)
Current Rankings are shown in the Figure below:
TLT is presently ranked #1 and will be purchased for the portfolio.
PCY is tied with SHY in the rankings and so is marginal at best. The 13 day EMA is currently below the 49 day EMA generating a “sell warning” signal. Also, the “absolute acceleration” is negative signifying decreasing momentum (in fact the 91 day momentum – ROC1 – is negative). Price has also only recently risen above the 195 EMA. For these reasons I will not be purchasing PCY at this time, but will put it on a “Watch List”.
I shall not be using a Strategic Asset Allocation (SAA) Plan as suggested by Swensen, rather, allocations are based on 3 important characteristics of asset behaviour, correlation with other assets, volatility and momentum.
This is a further modification of the Modified Risk Parity (MRP) weighting strategy that I have alluded to in a few earlier posts but uses the 91 day Sharpe Ratio (this is why the 63 (trading) day volatility is used in the ranking sheet) rather than volatility alone. Using the Sharpe Ratio integrates inverse volatility (naive Risk Parity) into the adjustment, since volatility is contained in the denominator of the Sharpe Ratio. The numerator of the Sharpe Ratio reflects the returns on the asset and so is proportional to momentum.
I shall refer to this as the Sharpe Weighting (SW) strategy.
Suggested allocations are shown in the figure below:
I plan to purchase 400 Shares of TLT to-morrow. I will keep the balance in Cash for now.
The Rutherford Portfolio will be reviewed every 33 days and data will be posted on this site.
I had intended to use the Sortino Ratio (using semi-variance – variance of negative returns) in preference to the Sharpe Ratio, reasoning that this might be more effective since it does not penalize positive returns (that we want). However, in back-tests there was really no discernible difference – with the Sharpe Ratio showing slightly better (but not statistically significant) results.
Allocations based on the Modified Risk Parity (MRP) method mentioned previously are also shown in the figure above. Using MRP allocations (in back-tests) results in a more conservative performance in that portfolio volatility is lower at the expense of returns – same old story. However, the Sharpe ratio of performance is better. It is my choice to accept the higher volatility (risk) in exchange for (hopefully) higher returns.
Late addition to Post:
In a recent post, Lowell mentioned the preference, by many investors, to choose low Beta assets. Note that the Beta of a portfolio containing TLT, PCY and SHY using SW weightings would be -0.10 or practically uncorrelated to VTTVX that I am using as the benchmark/reference. However, be a little careful since calculating the beta of a bond with reference to equities (although VTTVX contains 30% bonds) is not too reliable as indicated by the low r-squared values. Only VTI, VWO , VEA and RWX are high beta assets.
See Comments below – Stop Loss placement on TLT: