Dynamic Momentum Learning for Adaptive Trend-Following

Summary

This article studies how trend-following signals can adaptively use different momentum look-back periods. Instead of relying only on fixed summaries of past returns, it applies a dynamic binary classifier to learn whether relationships between past momentum and future returns are changing or remain stable. The investor can then combine or select momentum speeds, particularly around market turning points.

The authors evaluate the method using data from 56 futures contracts and compare it with a traditional time-series momentum strategy. They report that a mean-variance investor would pay a considerable management fee for the dynamic approach, describing this as evidence of improved signal accuracy and portfolio performance. The provided account does not give the size of the fee, detailed test design, or risk-adjusted results, so it does not establish how robust the advantage is across periods or implementation costs.

Key ideas

• Fixed look-back momentum rules may miss changes in how past returns relate to future returns.
• A dynamic binary classifier learns whether these relationships vary over time.
• The method can select or combine momentum speeds around turning points.
• The evaluation uses 56 futures contracts and compares the method with traditional time-series momentum.
• Reported investor willingness to pay a management fee suggests an economic benefit, though its magnitude is unspecified.

Tags

# Trend-Following Strategies via Dynamic Momentum Learning


# Trend-Following Strategies via Dynamic Momentum Learning









Time series momentum strategies are widely applied in the quantitative financial industry and its academic research has grown rapidly since the work of Moskowitz, Ooi and Pedersen (2012). However, trading signals are usually obtained via simple observation of past return measurements. In this article we study the benefits of incorporating dynamic econometric models to sequentially learn the time-varying importance of different look-back periods for individual assets. By the use of a dynamic binary classifier model, the investor is able to switch between time-varying or constant relations between past momentum and future returns, dynamically combining or selecting different momentum speeds during turning points, improving trading signals accuracy and portfolio performance. Using data from 56 future contracts we show that a mean-variance investor will be willing to pay a considerable management fee to switch from the traditional naive time series momentum strategy to the dynamic classifier approach.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.