Evaluating a binary relative-strength allocation

What the evaluation isolated

The archive distinguished three allocation designs. A strategic mix, or strategic-allocation, holds fixed long-horizon weights and rebalances back to them. A tactical mix, or tactical-allocation, makes large discrete weight changes from forecasts, sector-cycle knowledge, or a stock-versus-cash timing rule. A dynamic mix, or dynamic-asset-allocation, relocates capital across several classes as conditions change.

The evaluation defined dynamic allocation as a complete binary transfer from one representative asset class into one alternate class, not a partial reweight of a multi-asset book. That binary-switch keeps the entire stake in one class or the other, with no simultaneous multi-class holdings inside the test.

How the two books were built

Representative classes were formed by grouping funds by stated objective and geometrically averaging members on a daily basis as funds entered or left the database.

Alternate sleeves spanning lower to higher risk were tested against each primary class: a money-market sleeve, a long-term bond sleeve, a broad U.S. equity-index sleeve, an international-equity sleeve chosen for low U.S. correlation, and an aggressive-growth sleeve.

The historical window spanned more than one nominal business cycle. Reported statistics omitted commissions, transaction costs, and advisory fees, while fund-level management costs were already inside the prices used.

The rotation rule

The rotation rule is a momentum-rotation: it funds only the class whose smoothed daily percentage change currently leads its pair. The indicator is a relative-strength-switch. Each class’s daily percentage net-asset-value change is exponentially smoothed, the two smoothed series are subtracted, a shorter exponential average is applied to the difference, and the class with the higher relative strength is fully funded. Gaps that are small relative to the indicator’s peak are ignored.

The result is a binary-switch, a full transfer from one class into the other and back.

Scoring both books in the same unit

For every primary class and every alternate sleeve the study tabulated rate of return, monthly standard deviation, switches per year, fraction of time invested, Sharpe ratio, ulcer index, and maximum drawdown over the shared sample. The ulcer-index is the root-mean-square of equity retracements, used as a loss-only companion to standard deviation.

The Sharpe ratio was defined as yearly excess return over a money-market intercept, with monthly standard deviation converted to a yearly unit.

Cross-class correlations over the same window were used to screen pairs. Money-market and international-equity series were generally weakly related to most other classes, while like-equity groups moved more closely together.

Preferred pairings combined a higher Sharpe ratio with a low correlation between the primary class and its alternate, producing a two-state rotation rather than a simultaneously held multi-asset mix. Those preferred pairs form a dynamic-frontier, the return-versus-volatility envelope after the rotation rule is applied.