Complex Multi-Asset Class Momentum Strategy

Can investors beat a 60/40 stocks/bonds portfolio via a long-only momentum strategy applied to many asset class proxies? In their September 2022 paper entitled “Long-Only Multi-Asset Momentum: Searching for Absolute Returns”, Enrique Zambrano and Carlos Rizzolo explore variations of a long-only multi-asset momentum strategy. Strategy elements are:

• The asset universe is SPY, QQQ, IWM, VGK, EWJ, EEM, VNQ, DBC, DBA, GLD, LQD, HYG, TLT, SHV, IEF and Cash (or underlying indexes dovetailed with actual short fund histories). Cash, SHV and IEF are risk-off assets, and all others are risk-on assets.
• Measure momentum base on: (1) total return; (2) price relative to a simple moving average (SMA); and, (3) risk-adjusted returns that penalize assets with high return dispersion.
• Use signals for three lookback intervals for returns (3, 6 and 12 months) and three SMAs (50, 100 and 200 days). Aggregate signals from the three momentum measurements and three lookback intervals based on either an ensemble method or a scoring approach.
• For some tests, penalize (favor) assets with high (low) correlations with a portfolio of the rest of the assets using either a multiplier or a divisor.
• At the end of each month, rank the 13 risk-on assets according to aggregated-signal momentum and reform an equal-weighted portfolio of the top five assets. If any risk-on assets have negative momentum (with two ways to determine negative for the scoring approach), substitute for it the risk-off asset with highest momentum.

They focus on total strategy return over the full sample period, compound annual growth rate (CAGR), annualized standard deviation of returns (volatility), ratio of CAGR to volatility, maximum drawdown (MaxDD) and the probability of 1-year rolling returns being positive or higher than -5%. They also present performance data for the nine individual momentum measurement-lookback interval combinations. They further consider a strategy variation that limits exposure to risk-on assets based on the number of assets with negative returns (momentum breadth). Using daily prices for all assets as available from the end of December 2002 through the end of May 2022 (or daily associated index levels before respective funds are available), they find that:

(more…)