A PDF version of this commentary can be downloaded here.
This week we wanted to spend some time digging into performance of our Risk Managed U.S. Sectors (“RMUS”) and our Multi-Asset Income (“MAI”) portfolios.
At Newfound, we generally break portfolio construction into two pieces: (1) the signals that drive our tactical decisions, and (2) the rules the turn these signals into portfolio allocations. We liken this to preparing a meal: you need both ingredients and a recipe. In our case, the ingredients are the tactical signals and the recipe is the set of rules we utilize.
Just as with preparing a meal, you only need either the ingredients to be bad or the recipe to be bad for the meal to be ruined. You can have the best ingredients in the world, but unless they are thoughtfully combined, you can end up with a pretty horrific tasting meal. On the other side, having the best recipe in the world won’t save you if all your ingredients have spoiled.
In our RMUS portfolio, the ingredients are the momentum signals we generate for each of the primary U.S. sectors. The recipe in this case is the rules of removing negative momentum sectors and equal-weighting positive momentum sectors.
When it comes to analyzing performance, the root cause of out-/under-performance can be either the ingredients or the recipe. In the case of 2015, under-performance to a market-cap weighted benchmark is largely due to the recipe.
Let’s take a look at year-to-date returns for the U.S. equity sectors.
We can see some fairly significant divergence in sector returns, with health care (“XLV”) leading the way (+9.17%) and utilities (“XLU”) bringing up the rear (-9.20%). In such an environment, our recipe of equal-weighting is additive when it is overweight, relative to the benchmark, the high-returning sectors and underweight, relative to the benchmark, the low returning sectors. We can think of this more simply as: equal-weight will out-perform when the biggest sectors in the S&P 500 are doing the worst and the smallest sectors are doing the best.
Rewinding the clock to 12/31/2014, we can look at the weight differential an equal-weight portfolio had to the S&P 500.
We can see large overweights to sectors like materials, utilities, and energy, but also large underweights to sectors like financials, technology, and healthcare.
With the full benefit of hindsight, we can see that at the turn of the year, our equal-weight recipe set us up to overweight all the underperforming sectors and underweight all the outperforming sectors.
In fact, with a little math, we can quantify, sector-by-sector, exactly how much this equal-weight recipe contributed to portfolio return.
For 8 of the 9 sectors, the equal-weight recipe was a detractor to portfolio performance versus a market-cap weighted benchmark. The biggest detractors were the equal-weight portfolios under- allocation to technology and healthcare as well as its over-allocation to utilities. Overall, an equal- weight portfolio is trailing the S&P 500 by approximately 130bp year-to-date.
Should we be concerned? Looking back, we find that an equal sector-weight portfolio has out- performed the S&P 500 in 63% of rolling six-month periods since 6/1999. In the periods it under- performed (the other 37% of the time), it did so, on average, by 125bp. Put slightly differently, we ca expect the equal sector-weight portfolio to under-perform the S&P 500 in roughly one out of three six month periods. The year-to-date performance of the equal-weight recipe, then, is both not uncommon and inline with historical norms.
To us, short-term performance is much less important than continually accessing if our recipe and ingredients are meeting expectations. Performing this type of evaluation requires understanding why we use equal sector-weights in the first place.
It is important to remember that the recipe does not exist independently of the ingredients: our tactical momentum signals. If we were to utilize a market-capitalization weighted recipe, the signals on sectors like financials and technology would have a much larger impact on portfolio performance than the signals on materials or utilities. To go back to our meal analogy: the outcome would rely much more on the quality of one or two ingredients rather than all the ingredients.
By using an equal sector-weight construction methodology, we are able to take a diversified momentum approach; if one ingredient has slightly spoiled, it may be masked by the freshness of the other ingredients. Doing this necessarily means there will be periods of underperformance – but we believe the benefit of diversifying the impact of our ingredients far outweighs this downside.
Turning our focus to Multi-Asset Income, performance in the investable universe for the model has been highly mixed – but the average return, year-to-date, is -0.49%. In comparison, the iShares Barclay’s Aggregate Bond ETF (“AGG”) is down -0.52%.
The model rebalanced this week for the first time since mid-May. The biggest changes were the complete removal of corporate bonds (“LQD”; from 5.2% to 0%) and long-dated U.S. Treasuries (“TLT”; from 1.1% to 0%) from the portfolio. These positions had already largely been reduced at the point of the last rebalance (5/15/2015) due to unattractive risk-adjusted yield levels.
We also reduced our position in mortgage REITs (“REM”) from 18.26% to 10.21%, due to a negative momentum signal. If the signal remains negative, the position will be transitioned out over the next several weeks. The most significant increases in allocations are in bank loans (“BKLN”), from 15.20 to 21.06%, convertible bonds, from 8.41% to 14.32%, and S&P buy write, from 7.82% to 10.94%. The resulting portfolio has a significantly reduced exposure to interest-rate sensitive investments (including domestic and international treasuries, corporate bonds, REITs, and mortgage REITs) in favor of more credit-sensitive investments (bank loans, convertible bonds, emerging market debt, preferreds, and S&P buy-write).
With an estimated forward yield of approximately 13.21%, reducing exposure to mortgage REITs necessarily reduced the yield profile of the portfolio. Nevertheless, we estimate that the forward expected annualized yield of this portfolio is still a relatively high 5.95%.
While the most recent drawdown (from 4/15/2015 to present) is only 25% of the steepest drawdown seen in the hypothetical backtest of this portfolio (2.47% versus just under 10%), we believe that recent reconfigurations help highlight the importance of taking a dynamic approach when investing in this universe of asset classes. By utilizing momentum as a filter and using risk-adjusted yield to drive our allocations, the portfolio has two layers of risk management. In this case, they worked together to reduce, or eliminate entirely, exposure to asset classes with higher interest rate sensitivity, reducing portfolio sensitivity to a rising rate environment.