Skip to content

Managing Portfolio Volatility: summary and key takeaways

The paper tests volatility-managed portfolios built on exponentially weighted volatility forecasts across 29 market indexes. Faster-moving forecasts predicted volatility better, held realized volatility closer to the target and cut tail losses in most asset classes. Performance gains were clearer for equities and bonds than for currencies and commodities.

Stamos, M.; Zimmerer, T. · 2021 · The Journal of Portfolio ManagementSummary updated 20 November 2026
Published2021Journal article
Asset focusMulti-assetVolatility targeting
TypePortfolio MgmtPortfolio Construction & Risk
Reading time4 minthis summary

Managing Portfolio Volatility at a glance

The question
Does actively scaling portfolio exposure to volatility forecasts improve forecasting accuracy, risk-adjusted returns and tail risk across major asset classes?
Dataset & period
29 major market indexes: 8 equity markets, 6 currencies, 6 commodities and 9 bond markets, mostly traded futures plus carry-adjusted currency indexes, all as excess returns over the riskless rate. Daily returns, evaluated from January 1, 1990 to June 30, 2020. Results are reported for six equal-weighted portfolios: one per asset class, a stock-bond mix and a combined portfolio of everything. The equity volatility background section uses the MSCI World Index from 1980 to June 2020.
Method
Volatility is forecast either with a trailing estimate that weights all past daily returns equally, or with exponentially weighted moving averages (EWMA) of several half-lives from 12 months down to 0.5 months. Each managed volatility portfolio sizes its exposure as target volatility divided by forecasted volatility, with a 10% target. Forecast error is the mean squared gap between estimated and next-week realized volatility, using non-overlapping weekly windows built from five daily returns. For return and tail comparisons, all strategies are rescaled to exactly 10% annual volatility.

Key findings of Managing Portfolio Volatility

  • Equity volatility clusters: weekly volatility rises over about 10 weeks before a peak and takes about 10 weeks to fade, and its autocorrelation shape is similar in each decade since the 1980s.
  • Compared with trailing volatility, EWMA forecasts lowered forecast error by 15% at a 12-month half-life and by as much as 40% at 0.5 months, in all six portfolios.
  • Realized volatility of the managed portfolios drifted from the 10% target by roughly four percentage points with trailing estimates, and by slightly above 3% with the fastest EWMA.
  • Return gains from EWMA targeting over trailing targeting were strongest and most consistent for equities and bonds and mixed for currencies and commodities. The 0.5-month half-life gave positive alpha in every portfolio: above 1% for equities and bonds, slightly below 0.5% for currencies and commodities.
  • Tail risk, measured as conditional value-at-risk at 1% over 120 days, fell for equities from an almost 24% loss to a 17% loss in the best case. Currencies and commodities also improved substantially, while bonds improved only slightly and only with very fast half-lives.

Robustness, as reported in the paper

  • The study repeats the same volatility predictors and tests across equities, currencies, commodities and bonds, which the authors describe as a cross-validation of earlier work.
  • Autocorrelation of equity volatility is shown separately for each decade from the 1980s to 2020, with a similar pattern in all of them.
  • Results are shown for a range of EWMA half-lives, and the best forecasting accuracy and tail protection came from the shortest ones (0.5 and 1 month).

Caveats and limits

  • Return and tail results are shown mainly in charts, and the paper reports no transaction costs or turnover for the managed strategies (inferred).
  • Return improvements are measured against a trailing-volatility strategy rescaled to the same volatility, not against an unmanaged buy-and-hold portfolio (inferred).
  • The authors chose a simple EWMA model on purpose and do not test more complex volatility models.
  • Indexes enter the sample on different dates, and the evaluation covers one sample ending June 30, 2020 with no separate holdout period (inferred).

Items marked (inferred) are our reading of the design, not statements by the authors.

Why it matters

The paper gives a transparent benchmark for volatility targeting: one simple forecast family applied to many asset classes. It suggests the choice of forecast speed matters more for risk stability and tail losses than for average return. Practitioners can use it as a reference point when judging more complex risk-scaling overlays.

Questions about this paper

Which volatility forecast worked best in the paper?

EWMA estimates with short half-lives of 0.5 or 1 month gave the best forecasting accuracy across all asset classes tested.

Did volatility management help every asset class?

Not equally. Risk-adjusted gains were strong for equities and bonds and mixed to negative for currencies and commodities, while tail risk fell substantially for all but bonds.

How big was the forecast error reduction?

Against a trailing estimate, the error fell by 15% with a 12-month half-life and up to 40% with a 0.5-month half-life.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.3905/jpm.2021.1.207. We link to the paper and do not host the PDF.

Follow new quant research

Download MacroFinance market data as CSV or Excel, or pull it by API, and test the ideas in these papers yourself. New summaries are added as we review them.

Create a free account
Summaries are written by MacroFinance from the published papers, in our own words, for information only. They are not investment advice and may omit detail. Papers are not hosted here: each page links to the original at its publisher.