ReSolve 2017 Year-End Report
Calendar 2017 was an exemplary year for ReSolve strategies.
Strategies targeting long-term equity-like volatility produced almost double the return of U.S. stocks, with some mandates achieving over 40% growth.
Strong results are a function of proven process + relentless discipline + favourable conditions. We offer a comparative analysis to cement the point.
Net performance of ReSolve Strategies in 2017 (daily scale). Click on links to see strategy fact sheets.
|AAA 8% Levered
|Risk Parity 12%
|Risk Parity 6% Un-Levered
SOURCE: ReSolve Asset Management. PAST PERFORMANCE IS NOT NECESSARILY INDICATIVE OF FUTURE RESULTS. See disclaimer.
Download the Complete 2017 Report HERE
Calendar 2017 was an exemplary year for ReSolve strategies. Per Table 1, at a 12% annualized volatility target, consistent with the long-term risk of a traditional 60/40 global balanced portfolio1, performance of Adaptive Asset Allocation (AAA) and Global Risk Parity (RP) mandates rivaled most global equity benchmarks2. Adaptive Asset Allocation mandates targeting a volatility closer to the long-term profile of global equities (16%-20%) produced almost double the return of U.S. stocks, with some mandates achieving over 40% return.
Figure 1: Live geometrically linked net daily returns to ReSolve USD mandates in 2017. SOURCE: ReSolve Asset Management. See disclaimer.
Figure 1 compares the net total return trajectories of ReSolve’s Adaptive Asset Allocation, Risk Parity, and Tactical Equity mandates against global equities3.
Long-time investors know that ReSolve takes an ensemble approach to strategy design to minimize the risk of specifying the exact wrong model for the current market environment. The current version of Adaptive Asset Allocation uses resampling to draw many combinations of lookback and weighting parameters to estimate momentum, correlations, and volatilities, each time we create an optimal portfolio. In addition, the Strategy uses four techniques to measure momentum and trend; three transforms of momentum data; and five portfolio optimization methods.
One of these methods is thematically consistent with the toy model we described in our seminal whitepaper, Adaptive Asset Allocation: A Primer (2012 rev 2015), where assets are ranked on average momentum, and the top half of assets are held in minimum variance weights. But we also apply forty other combinations of momentum measures, transforms, and optimization methods when we form portfolios. Final portfolios represent a thoughtful combination of all forty-one sub-strategies. While the long-term performance profiles of these sub-strategies are statistically indistinguishable, we observe fairly wide dispersion in performance from year-to-year.
Figure 2: Return trajectory of Adaptive Asset Allocation sub-strategies in 2017. Live aggregate strategy performance is highlighted in gold. SOURCE: ReSolve Asset Management. See disclaimer.
Figure 2 illustrates how each of the forty-one sub- strategies that comprise ReSolve’s Adaptive Asset Allocation approach performed this year, without any volatility targeting. The worst strategy produced less than 9%, while the best generated almost 21%. This is quite a range given that the strategies all seek to maximize momentum and trend while minimizing portfolio volatility. Meanwhile, the live aggregate strategy4 delivered net returns of 15.5%.
Some investors may be wondering why the unlevered version of Adaptive Asset Allocation lagged global stocks. After all, global equities produced one of the best trends ever in 2017. The reason is that ReSolve’s strategies, in contrast with many other ostensibly similar strategies, explicitly account for the fact that adding assets with lower expected Sharpe ratio, but low correlation to other portfolio assets, will improve the expected risk-adjusted performance of the portfolio.
Take the case of a two asset portfolio, where one asset has twice the expected Sharpe ratio of the other asset. If the two assets are perfectly correlated, an investor should simply choose to invest all her capital in the asset with the best expected performance. However, if the assets are uncorrelated, the investor is always better off, in terms of expected risk-adjusted performance, to invest in a combination of the two assets rather than simply investing all her funds in the asset with the best expected performance5.
Figure 3: Optimal risk budget for a lower Sharpe asset given different correlation assumptions. SOURCE: ReSolve Asset Management.
Figure 3 considers the case of a two asset portfolio, and quantifies the risk budget6 that should be allocated to an asset with a lower expected Sharpe ratio (with the remainder in the higher Sharpe asset), to maximize the portfolio’s expected risk-adjusted performance. Given the example where one asset has half the expected Sharpe ratio of the other asset, the violet line in the chart shows that the most efficient portfolio would combine about 63% in the higher Sharpe asset with a 37% position in the lower Sharpe asset, when the correlation between the two assets is zero. Indeed, assets with negative Sharpe ratios may improve the efficiency of a portfolio given sufficient negative correlation. The light blue line tracks the optimal allocation to a lower Sharpe asset given a correlation of -0.5 with the higher Sharpe asset. You can see that an asset with a Sharpe ratio that is negative and 0.25 times as large as the higher Sharpe asset would still deserve a risk budget allocation of almost 30%!
Why does this matter? Many popular Global Tactical Asset Allocation strategies simply emphasize assets with favorable characteristics, which often leads to portfolios that are fully allocated to just one asset class. In years like 2017, this approach works well, because there are no trend reversals, and the best approach in retrospect was simply to hold 100% in emerging stocks all year. However, under conditions of uncertainty, portfolios that account for correlations will hold more diverse assets, which positions them to capture a smaller proportion of losses on reversals, and to transition into alternate asset classes more quickly.
This is one of the reasons why even a simple Adaptive Asset Allocation strategy is the top ranked strategy by Sharpe ratio, and second-best strategy by annualized total return, over the past twenty years (in simulation) according to a popular GTAA portal7.
PERFORMANCE ATTRIBUTION ANALYSIS
The relatively strong performance of ReSolve’s strategies this year accrued from a confluence of factors. First, most global assets delivered positive returns on the year, and several markets produced spectacular growth. Indeed, the top performing assets produced their returns with low volatility and high Sharpe ratios, so that it was possible to amplify returns using the Capital Market Line. Second, there was a large disparity between the top and bottom- performing assets, which made it easy for our momentum indicators to separate the wheat from the chaffe. In addition, the best performing assets outperformed the worst performing assets very consistently all year, with few rank-reversals.
At the end of 2015, we produced a report to explain some of the reasons why ReSolve’s strategies had struggled that year. We described the necessary conditions for long-only active asset allocation strategies to thrive, and compared the conditions that prevailed in 2015 with conditions during other calendar years over the previous two decades. The analysis made clear just how extraordinarily challenging that year was relative to other years on a wide variety of dimensions. We thought that this year it would be interesting and instructive to contrast conditions in the current year with conditions that prevailed in calendar 2015, using the same analytical tools.
This exercise is useful because, while we have made incremental improvements to our strategies over the past few years, the fundamental mechanics that inform our strategies have not changed materially. By revisiting the metrics that we used to explain the unexciting performance in 2015, and demonstrating how current conditions would have predicted strong performance this year, it will become clear that results in any one year do not reflect on the value of our strategies, but rather reflect the role of luck on short term investment results.
To kick-off our analysis, let’s first examine just how amazing the current year was for equity investors. While investors in U.S. equities have enjoyed outstanding returns for the past nine years, 2017 rewarded investors from virtually every corner of the globe. Moreover, per Figure 4, equities produced their mouth-watering returns with vanishingly low monthly volatility, leading to stratospheric Sharpe ratios at monthly scale.
Figure 4: Rolling 12-month annualized Sharpe ratio. SOURCE: Global Financial Data from 1926 – 2015, CSI Data for VT ETF from 2015 – 2017. PAST PERFORMANCE IS NOT NECESSARILY INDICATIVE OF FUTURE RESULTS. See disclaimer.
In fact, 2017 was the only calendar year since 1926 that global equities offered investors positive returns in every single month8.
Recall that ReSolve’s Adaptive Asset Allocation and Risk Parity mandates are long-only. These strategies will produce positive returns only if one or more markets are rising. In most years, this is not an issue, as there is almost always a strong bull market somewhere. During periods of strong global growth surprises, equity markets should produce good returns. In other years characterized by negative growth or outsized inflation shocks, other assets, like high grade bonds or commodities, should do well. But there are usually good opportunities for dynamic, long-only strategies to prosper, as evidenced by the top row of Figure 5.
Figure 5: Total returns to major asset classes in calendar years 2000 – 2017, sorted from highest to lowest. SOURCE: Calculations by ReSolve Asset Management. Data from CSI, S&P Dow Jones, Deutsche Bank. PAST PERFORMANCE IS NOT NECESSARILY INDICATIVE OF FUTURE RESULTS. See disclaimer.
Most years present good opportunities for long-only active asset allocation strategies to prosper, but some years, like 2001 and 2015, conspire to produce few profitable opportunites.
A historical case study will help cement the point. Imagine that on January 1st of each year, we knew in advance which major global asset classes9 would deliver returns above the group’s average, as well as positive returns, in the coming year. Of course, knowing the future in this way is impossible, but it helps to illustrate the available opportunity set for an active long-only global asset allocation strategy. Perhaps unsurprisingly, an investor with this type of perfect foresight – who allocated capital equally among the top half of positively performing assets each year – would have produced average annual returns of 21.5% since 1991.
Moreover, as can be seen in Figure 6, this strategy would have generated returns over 15% in 21 out of 27 years, and gains of more than 10% in 23 of the years. Fully 26% of years provided returns of greater than 25%! Any investor in his right mind would gleefully invest in this ‘crystal ball’ strategy…
Continued in the report…
Download the Complete 2017 Report HERE
1. A portfolio consisting of 60% MSCI Global All-Cap World Index, 20% U.S. Aggregate, and 20% S&P/Citigroup International Treasury Ex-US total return indexes exhibited an annualized daily volatility of 11.7% since 1990. 2. The Vanguard Total U.S. Stock Market ETF (VTI) returned 21.28%, and the global Vanguard Total Market ETF (VT) returned 24.57%. 3. Total return index of the Vanguard Total Stock Market ETF (VT) 4. ReSolve Adaptive Asset Allocation: 8% Volatility (USD) unlevered mandate. 5. Note the repeated emphasis of the word expected. Obviously if an investor has perfect information about which asset will outperform in the coming period, they should invest all their capital in that asset, with maximum leverage. Unfortunately, markets rarely present investors with perfect information, so investors must act to optimize their expected future wealth under conditions of uncertainty, which supports diversification. 6 .If the assets have equal expected volatility then risk budget is the same as portfolio weight. 7. See the “Strategies” page in the member’s section at Allocate Smartly https://allocatesmartly.com 8. Actually, the twelve months ending November and December 2017 represent two of just sixteen twelve-month rolling periods in the past 1075 months where global equities offered investors positive returns in every single month.