Skip to content

Demystifying Time-Series Momentum Strategies: Volatility Estimators, Trading Rules and Pairwise Correlations: summary and key takeaways

The paper tests how three design choices affect a diversified futures trend-following strategy: the volatility estimator, the trend signal and a correlation-based leverage factor. Better volatility estimation and a significance-based signal cut turnover a lot with little Sharpe change. The correlation factor helped most in the post-crisis period.

Baltas, N.; Kosowski, R. · 2020 · Market MomentumSummary updated 20 November 2026
Published2020Journal article
Asset focusMulti-assetTrend following
TypeStrategyTrend & Momentum
Reading time4 minthis summary

Demystifying Time-Series Momentum Strategies: Volatility Estimators, Trading Rules and Pairwise Correlations at a glance

The question
How do the volatility estimator, the trading rule and average pairwise correlation affect the turnover and performance of time-series momentum strategies?
Dataset & period
Daily open, high, low and close futures prices for 75 contracts (commodities, equity indices, currencies and bonds) from Tick Data, December 1974 to February 2013; strategies are rebalanced monthly. Not all contracts start in 1974.
Method
The strategy goes long or short each futures contract and scales each position to a volatility target. The authors swap the standard close-to-close volatility estimate for the Yang-Zhang range estimator. They compare the past-year return sign rule with a rule that trades only when a fitted price trend is statistically significant. They then scale leverage down when average pairwise correlation is high. Costs are approximated as turnover times an assumed market impact.

Key findings of Demystifying Time-Series Momentum Strategies: Volatility Estimators, Trading Rules and Pairwise Correlations

  • The Yang-Zhang estimator cut turnover by 13% for the long-only constant-volatility strategy and 8% for the time-series momentum strategy, with no meaningful change in before-cost Sharpe ratios.
  • The turnover reduction from the range estimator appeared in all 75 contracts, with the largest effect in low-volatility contracts such as interest rate futures.
  • The trend-significance rule produced a before-cost Sharpe ratio of 0.99 versus 1.04 for the sign rule, a difference that is not statistically significant. Its turnover was roughly one third of the sign rule's.
  • The share of contracts with no significant trend rose after the crisis, and average pairwise correlation jumped and stayed elevated. The authors link both to the weak recent performance.
  • Over the full sample the correlation factor moved the Sharpe ratio only from 1.04 to 1.05 but improved downside-risk ratios, and turnover rose from 28.89% to 41.23%.
  • In the post-crisis period the correlation factor lifted the before-cost Sharpe ratio from 0.14 to 0.29.

Robustness, as reported in the paper

  • Estimation windows from one to twelve months were tested; the before-cost Sharpe ratio peaked at a three-month window, which is used for later results.
  • Return predictability regressions use standard errors clustered by time and asset; momentum shows up for roughly the first year of lags, followed by reversals.
  • With assumed market impact of 10 basis points, the post-crisis Sharpe ratios fall to 0.12 (unadjusted) and 0.26 (correlation-adjusted). At 50 basis points they fall to 0.05 and 0.14.
  • Simpler or less efficient range estimators were also examined, but the results are not reported.

Caveats and limits

  • Transaction costs use a constant market impact assumption; the authors state that real impact depends on order size and that results are illustrative.
  • Lower turnover from the better volatility estimator and trend rule did not translate into clearly higher risk-adjusted returns under the cost model, which the authors call surprising.
  • The correlation adjustment raises turnover and costs, and its strongest benefit is in a short post-crisis window (inferred: a single sub-period).
  • Most results sit in tables and figures that are not in the extracted text, so only prose-stated numbers are used here.

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

Why it matters

The paper separates three sources of turnover and performance in trend-following design. It shows that costs, not just signals, depend on how volatility and trends are measured. It also ties recent weakness to rising cross-asset correlation and offers a leverage rule to address it.

Questions about this paper

Does a better volatility estimator reduce trading in trend-following?

Yes. The Yang-Zhang range estimator lowered turnover by 13% for the long-only strategy and 8% for the momentum strategy, with similar before-cost Sharpe ratios.

What did the correlation adjustment do to performance?

In the post-crisis period it raised the before-cost Sharpe ratio from 0.14 to 0.29, at the price of higher turnover. After assumed 10 basis point costs the figures were 0.12 and 0.26.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.1002/9781119599364.ch3. 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.