Trend following, or time-series momentum, is a rules-based investment strategy that uses both long and short positions to generate returns. It is based on the assumption that a trend (positive or negative) is likely to persist—that winning assets are more likely to continue gaining while losing assets are more likely to continue underperforming, says Eddie Cheng, CFA, Allspring Global Investments.

Equities and bonds are expected to compensate investors for bearing systematic risk directly linked to economic cycles. Equity investors are exposed to growth risks and bond investors are exposed to inflation risks. 

The drivers for this trend strategy are somewhat different. They can largely be categorized as behavioral biases existing in financial markets such as the anchoring effect and herding phenomenon. 

Anchoring happens when markets underreact to news, while herding occurs when markets eventually overreact to news beyond a level justified by fundamentals. Because these behavioral biases are not necessarily synced with growth and inflation cycles, we expect trend to deliver robust performance regardless of the economic environment.

How trend strategies can deliver robust performance across different economic environments

To illustrate how traditional long-only stocks, bonds, and commodities expose investors to growth and inflation risks while trend may cut through these different risks, we classified each month from February 1965 through June 2022 into one of four different economic regimes:
-    High inflation, high growth
-    High inflation, low growth
-    Low inflation, high growth
-    Low inflation, low growth

When we look at the average monthly excess returns for both trend and the underlying assets grouped by the four regimes, we found that the profitability of trend is pervasive regardless of the regime we are in.

This is a stark contrast to the economic sensitivity for the underlying assets that the trend signal is based on. For example, commodities are very profitable in the high growth and high inflation regime, but they're not so effective when both growth and inflation are low.

Equities performed poorly when inflation was high, while bonds did not fare well when inflation was high and growth was low. In the most recent period in 2022, which has seen high inflation and low growth, equities and bonds had a meaningful drawdown this year while commodities gained. Trend provided a robust positive return in the same period. 

Resolving trend-specific challenges 

Managing a trend strategy has its own unique challenges. The biggest challenge may be in how to calibrate the signal as to which assets to hold long and which to hold short, so the strategy can balance between reactivity and stability. Setting the trend signal to calibrate at a rate that is fast tends to make the strategy reactive during turning points, but a fast signal is also more likely to react to noise instead of capturing the true trend that will persist in the future.

The opposite is equally true; while calibrating the trend signal at a slower rate can filter out abnormal price movements that are deemed to be noise, it can result in failing to react as quickly if the noise turns out to be an actual signal of a change in the trend. 

A dynamic approach using machine-learning techniques 

 

Sometimes the trade-off between reactivity and stability is dealt with by a static weighting between fast and slow signals. A dynamic approach is based on some conditioning variables that allow the weights to change with time and conditions.

While the simplicity of the static approach is attractive, we find the dynamic approach more appealing and also more intuitive: When markets are more volatile, we believe investors tend to be more short-sighted and the shorter-term signal should be overweighted. When markets are calmer, investors may be prone to just letting their winners run and the longer-term signal should be overweighted.  

We use a machine-learning technique to help us identify regimes in which fast or slow trends are likely to outperform. We found that using this approach has been useful in adjusting the weights to the fast and slow signals. (For detailed analysis, see "Trending Fast and Slow," The Journal of Portfolio Management 48, no. 3, February 2022.)

Making trend your friend

Financial assets are typically sensitive to economic cycles. In a high-inflationary environment, equities and bonds tend to underperform. On the contrary, the drivers for some alternative strategies such as trend are less tied to economic cycles and can provide good diversification across different regimes. 

Managing a trend strategy does come with the unique challenge of balancing signal reactivity and robustness. We find that investors can achieve a better outcome through machine learning-based techniques by dynamically allocating the trend signal weight between fast and slow signals.

By Eddie Cheng, CFA, head of International Portfolio Management for the Systematic Edge Multi-Asset team at Allspring Global Investments.