Adaptive Asset Allocation
Rule-based momentum and volatility allocation, tested rather than asserted.
Summary
Butler, Philbrick and Gordillo, who run a quantitative asset management firm, argue for a systematic allocation approach that adjusts holdings in response to observed returns and risk rather than holding fixed weights.
The critique of the conventional approach comes first. They argue that the standard sixty-forty portfolio and the strategic allocation framework depend on long-run return assumptions that cannot be estimated reliably, that they ignore observable changes in risk and correlation, and that the historical record supporting them covers a period of falling interest rates that will not repeat. They also argue that market-capitalisation weighting concentrates a portfolio in whatever has already risen.
Their method combines three elements applied to a universe of asset classes accessed through liquid instruments. Momentum selects the assets with the strongest recent returns, on the empirical finding that relative performance persists over intermediate horizons. Volatility estimation sizes positions so that each contributes comparable risk, since volatility is far more predictable than return. Correlation estimation adjusts the combination so the portfolio's total risk is minimised for the assets selected. The whole is rebalanced periodically.
They support this with backtests over long periods showing higher risk-adjusted returns and smaller drawdowns than static allocation.
The method's components are each supported in the academic literature: momentum, the low volatility effect and the predictability of volatility and correlation relative to returns. The combination's backtested performance depends on parameter choices made after seeing the data, which is the standard problem, and the authors' out-of-sample record since publication is where the evidence actually lies. Transaction costs and tax treatment matter more for this approach than for static allocation, and the book treats them lightly.
Key takeaways 7
- Volatility and correlation are far more predictable than expected returns, so allocation should rely on them rather than on return forecasts.
- Relative performance across asset classes persists over intermediate horizons, which is the basis of the momentum selection.
- Sizing positions by volatility makes each contribute comparable risk, which fixed-weight allocation does not achieve.
- The historical record supporting conventional balanced portfolios covers a period of falling interest rates that cannot repeat.
- Market-capitalisation weighting concentrates a portfolio in whatever has already appreciated most.
- Adjusting allocation in response to observed risk reduces drawdowns relative to holding fixed weights through a decline.
- The approach requires more turnover than static allocation, so transaction costs and tax treatment materially affect the realised result.
Who should read it
Investors and allocators considering systematic tactical allocation who want the method stated explicitly, weighed against its turnover costs and out-of-sample record.
Similar books 6
Chosen by shared topics, the entries that hold them, and the people who recommended both.
Katsenelson on buying and selling inside range-bound markets.
What markets looked like at the US stock bottoms of 1921, 1932, 1949 and 1982.
How to build a portfolio balanced to hold up in any economic environment.
Absolute momentum as a systematic substitute for judgement.
The evidence on what drives returns across asset classes: risk factors and behaviour.
The mathematics of risk parity: risk contributions and risk budgeting.
Where it sits in the reference 2
Topics