Dual momentum combines two decisions. Relative momentum selects an asset based on its return compared with other candidates. Absolute momentum checks that selected return against a specified threshold before holding the asset. Together they define an allocation rule that can be built and tested.
Specify one version
The illustrative prompt on this page uses SPY, EFA and BIL. At each monthly decision:
- Compare SPY and EFA over a 12-month lookback.
- Select the higher-return fund.
- Hold BIL if the selected return is below zero; otherwise hold the selected fund.
- Make tie handling, signal dates, fill dates, price basis and missing-data behavior explicit.
The lookback and zero threshold are example choices. Other implementations compare with Treasury bills or use a different defensive asset. They are different specifications and can produce different results.
Inspect a public model
Open the card's public record and read the current input schema and run. Check its actual threshold, universe, dates and data assumptions rather than assuming the strategy name proves a particular implementation. The card updates with the model; its metrics are not a frozen study. Source code access belongs to the model owner.
Understand the tradeoffs
The hypothesis is that recent relative performance can help choose an allocation. A backtest can evaluate that hypothesis over available history; it cannot establish that the pattern will persist.
A signal can reverse after trading, creating a whipsaw. A rule can also leave an asset after a decline and miss a recovery. Concentrating in one selected holding differs from a diversified fixed allocation. Inspect these behaviors in the allocation and drawdown histories instead of relying on a headline return.
Compare and maintain
Use a test run to change one supported input while preserving live defaults. Keep the run IDs, hold the comparison window fixed, and inspect costs, warnings and synthetic backfills. A different universe may have a shorter shared history. An apparent improvement can reflect those changed dates rather than the rule.
After reviewing a result, decide separately whether to adopt its inputs and enable scheduled refreshes. An enabled schedule does not prove the latest run succeeded; inspect the data date before using allocations. See the agent workflow and backtesting mistakes.
This prompt is illustrative. Review your own model's implementation and assumptions. ENSEMBLE is research software; backtests are hypothetical and past performance does not guarantee future results.
Frequently asked questions
- What are relative and absolute momentum?
- Relative momentum compares assets with one another. Absolute momentum compares an asset's return with a stated threshold, such as zero or a cash benchmark.
- Is this the same as every GEM implementation?
- No. Global Equities Momentum and other dual-momentum implementations can use different thresholds, defensive assets, data and timing. Specify those choices rather than relying on the label.
- Can I change the lookback or universe?
- Yes, within the model's supported inputs or through a reviewed model revision. Inspect coverage and compare matched windows; adding candidates can also add opportunities to overfit.
Related
- StrategiesThe 200-day moving average strategy: rules, evidence, and a backtest
Define a moving-average allocation rule, specify signal and execution timing, and inspect its tradeoffs in a model run. An illustrative prompt helps you build and compare your own version.
- StrategiesRisk parity: how it works, how to size it, and a backtest
Understand risk contributions and the simpler inverse-volatility weighting rule. Inspect a public model, test one assumption, and distinguish a model specification from a performance claim.
- BacktestingBacktest metrics explained: CAGR, volatility, Sharpe ratio, and max drawdown
Read ENSEMBLE backtest metrics with their actual conventions: CAGR, annualized volatility, a zero-risk-free Sharpe calculation, maximum drawdown and trading activity. Compare matched windows and inspect assumptions.
- BacktestingCommon backtesting mistakes and how to avoid them
The ways a backtest lies: lookahead bias, survivorship bias, overfitting, ignored costs, short or regime-limited samples, and misread metrics. What each one looks like, how to detect it, and what a backtesting tool can and cannot protect you from.