Method & calculation
The Dual Momentum rules, explained with Basic-DualMomentum as the example. How the advanced portfolios are built, and the data and calculation assumptions, are here too.
Dual Momentum
A rules-based strategy that holds the assets with the strongest recent returns and moves to a defensive asset when the market's own momentum turns negative. One decision a month. Taking Basic-DualMomentum as the example, each month end it decides in this order. 1. Relative momentum: between SPY and QQQ, take the one that rose more over the past year. 2. Absolute momentum: check whether LQD (US investment-grade corporate bonds) has beaten the 3-month US Treasury bill over the past year. 3. If it has, hold the chosen asset. If not, move to XLU, the defensive asset.
Compared: SPY / QQQ / Defensive asset: XLU
Rules fixed, then observed
Basic-DualMomentum is a simple starter portfolio for learning the Dual Momentum rules: the Dual Momentum proposed by Gary Antonacci, with the assets replaced by ETFs. Nothing in it was tuned to fit past results. The advanced portfolios (L0 to L3) use a designed selection process, so their results after the rules were fixed are recorded separately from the results before.
Advanced portfolios (FoF)
Each Dual Momentum strategy is treated like an asset. Strategies are combined, then those combinations are evaluated again at further stages, called DM, L0, L1, L2, L3 and the advanced portfolios. The structure is public; how the upper stages select what they hold is not.
Benchmarks
Comparisons use SPY (S&P 500), bought and held, calculated over the same period as whatever it is compared with.
Data and determinism
Price data: EODHD daily prices (unadjusted) and its dividend and split records. Split and dividend adjustments are calculated in-house. Dividends and splits are cross-checked against Tiingo and Alpaca records. ^VIX: the official CBOE series. Risk-free rate: 3-month U.S. Treasury bill (DTB3). Benchmark: SPY. System of record for signals and returns: PostgreSQL.
Portfolio selection is deterministic: the same inputs always produce the same selection. The selection method is not published.
Theory and papers
Dual Momentum combines relative momentum with absolute (time-series) momentum. The work it builds on: