By: Roland Austrup
Chairman & Managing Principal at WaveFront GAM
Just as value investing assumes mean reversion and growth investing assumes compounding innovation, trend-following assumes that price moves tend to continue before they reverse. This assumption is not speculative; it is empirically validated across decades and asset classes.
Rather than viewing trend-following as a hedge or tactical sleeve, investors should consider it a core exposure and a structural source of return alongside equity, credit, and real assets.
Trend-following strategies, particularly those employed by Commodity Trading Advisors (CTAs), have demonstrated persistent long-term performance across asset classes. There are two main factors that explain this performance:
1. Statistical dynamics of financial markets as complex systems, which exhibit leptokurtic and power-law behavior due to multiple factors affecting markets, interdependencies, feedback loops and emergent behavior from those feedback loops..
2. Structural and macroeconomic drivers of trend persistence, including long-term GDP growth, interest rate cycles, and commodity supply/demand imbalances.
Markets as Complex Systems
For context, natural phenomena can be broadly categorized by their statistical behavior as follows:
• Random (Gaussian Distributions)
• Mean-Reverting (Platykurtic Distributions)
• Trend-Persistent (Leptokurtic Distributions & Power Laws)
Random or Gaussian distributions, characterized by thin tails and a bell-shaped curve, generally describe phenomena where a number of small and independent factors lead to an outcome. Examples include height and weight, test scores and Brownian motion.
Platykurtic distributions have thinner tails and lower peak densities. They do not generally show up in nature and show up more often in designed or engineered systems.
Leptokurtic distributions, characterized by fat tails and sharp peaks, are the dominant descriptor of natural systems. Nature is a complex, dynamic system in which a multitude of interdependent factors and recursive feedback loops interact to produce nonlinear outcomes. Examples include population cycles of animals, climate events and financial markets.
Financial markets are driven by a wide variety of factors. These include, but not limited to, macroeconomic indicators, central bank and government policies, geopolitical events and tensions, politics, weather, commodity supply/demand and, importantly, investor psychology and herd behavior. In addition, there are both positive and negative feedback loops such as, respectively, momentum and herding that can amplify trends, and mean-reversion and valuation anchoring that can reverse trends.
In his argument against the Efficient Market Hypothesis in his book, “The (Mis)behaviour of Financial Markets, Benoit Mandelbrot demonstrated how markets are complex, adaptive systems characterized by leptokurtic distributions, volatility clusters and evidence of long-term memory, autocorrelation and feedback loops. And, in his book, “The Alchemy of Finance”, Soros demonstrated how feedback loops cause Individual decisions to aggregate into emergent collective behavior and perceptions that then influence prices which, in turn, influence perceptions.
These statistical and behavioral properties explain why markets trend, why they crash and why traditional models underestimate risk. These properties are the foundation for why trend-following strategies that cut losses have and will continue to perform over time. These strategies provide exposure to and capitalize from the most structural feature or DNA of markets.
Structural & Macroeconomic Sources of Trend Persistence
In addition to behavioral and statistical dynamics, markets also trend due to slow-moving structural forces:
Equities are fundamentally driven by long-term GDP and productivity growth, as stock prices reflect claims on future productivity and earnings. As economies grow, corporate profits rise, creating a secular upward drift in equity prices. However, it is worth noting that the belief that “stocks always go up” is contingent on global growth and productivity. Japan’s post-1989 stagnation is an example that stocks don’t always go up. Despite world-class companies, flat GDP and demographic decline led to decades of equity underperformance. Similarly, as was the case with the collapse of the Roman Empire and the multi-century Dark Ages that followed, if civilization were to regress for a long period of time, global equities could stagnate or decline accordingly.
Fixed-Income markets generally have a built-in positive expectancy from the cost of capital priced into fixed income instruments. And, as the yield curve is generally positively sloped, back-adjusted fixed-income futures tend to show upward drift over the long-term. However, interest rate cycles can cause secular bear markets.
* Disclaimer
* FOR ILLUSTRATIVE PURPOSES ONLY. Source: WaveFront
Currencies are mean reverting over the long-term within wide bands, driven by purchasing power parity and trade balances. Within those bands though, capital flows and carry-trades from sovereign interest rate differential cycles can create multi-year trends.
As with currencies, commodities tend to exhibit long-term trends within wide bands of mean-reversion. Price trends are anchored to the fact that supply/demand doesn’t flip-flop. While demand can shift quickly, supply moves in trends as it is slow to adjust due to capital intensity, regulatory constraints, and infrastructure bottlenecks. In addition, commodity futures also exhibit trend from persistent cycles of backwardation or contango.
Trend-Following Strategies
Trend-following CTA strategies perform over time because they are aligned with the deep statistical and structural realities of financial markets. As described, these markets are complex systems that exhibit leptokurtic and power-law behavior, driven by feedback loops and reflexivity. They are also shaped by slow-moving macroeconomic, geopolitical and government policy forces, as well as persistent pricing structures, that create persistent trends across asset classes.
By definition, time-series data that exhibit leptokurtic behavior have slight tendency towards serial correlation versus mean-reversion or randomness. And it can be demonstrated that a trend-following exposure to those time series will produce positive returns over time.
Furthermore, as the source of return from trend-following is exposure to statistical and structural characteristics of markets, a proliferation of trend-followers has not and will not arbitrage away these tendencies and sources of return. Rather than eliminate the tendency of markets to trend, a proliferation of trend-based managers will simply modulate and perhaps amplify the way in which markets trend. The same goes for AI. The use of AI to develop trading and investment strategies will only change the way in which markets trend, not that they do trend.
However, strategy herding, whereby managers gravitate towards similar trend-following models that loosely make decisions around the same inflection points, can create inefficiencies. Herding can cause localized gaps and jumps to occur at trend inflection points. Those inefficiencies can lead to arbitrage opportunities and there has been a growth of managers who look for these inefficiencies rather than focus on trend. The difference, though, between arbitrage and trend-following is that, again, trend is a structural feature that only gets modulated by an increase in participants. In contrast, with an increase in participants, arbitrage eliminates inefficiencies and the source of the arbitrage return.
A Meta-Strategy: Capturing Trendiness
A more robust approach than trend-following to capture the fact that markets trend over time and avoid the risk of herding inefficiencies may be a more agnostic, factor-based approach to simply having exposure to the “trendiness” of markets. There are a few options one could consider.
A trendiness score could quantify how prone an asset is to trending behavior over time. It can be constructed using several indicators, for example the ratio of net price movement to total volatility, the Hurst exponent that measures long-term memory in time series, autocorrelation of returns or the Sharpe or Sortino Ratio of a trend-following strategy as a proxy for trendiness. However, the problem with this approach may be that, while trendiness may be a good historical descriptor, it may not be a good predictor of future trendiness.
Another approach may be to maintain a passive, convex exposure to markets or the degree to which all markets trend over time. As traditional convex exposure via long options is costly due to time decay and bid-ask spreads, an alternative approach would be to synthetically replicate the payoff profile of a generic array of long options by delta-hedging a futures portfolio based on the evolving delta of options whose payoff profile one is trying to replicate.
Such an approach could be built using a “fly-trap” model. For example, imagine dozens of fishing lines (options to replicate) cast into a lake. Each line is cast from a different location (strike price) and tuned to a different depth (holding period), and each reacts to movement in its own way. Collectively, they don’t predict where the fish are, but they ensure exposure to wherever fish might be. This approach doesn’t chase or follow price. Rather, it casts a wide net across time and structure, thereby creating synthetic convexity structurally aligned with how markets trend and misbehave. More than a strategy, this approach is a meta-framework for harvesting the statistical nature of markets.
Conclusion
Trend-following CTA strategies perform over time because they are aligned with the deep statistical and structural realities of financial markets. These markets are complex systems that exhibit leptokurtic and power-law behavior, driven by feedback loops and reflexivity. They are also shaped by slow-moving macroeconomic forces that create persistent trends across asset classes.
While herding among trend-followers can distort short-term dynamics and create local risks at inflection points that could erode return potential, it does not eliminate the tendency of markets to trend. A meta-strategy focused on trendiness, rather than trend-following, may offer a more adaptive and resilient framework for capturing the true nature of market behavior.
About the WaveFront All-Weather Alternative Fund
Introducing Canada's premiere all-weather investment solution, the WaveFront All-Weather Alternative Fund. Engineered to deliver stability and consistent performance, regardless of the market environment, now available as both a mutual fund and ETF.
The WaveFront All-Weather Alternative Fund is engineered to deliver consistent, superior risk-adjusted returns across diverse market conditions. With a dynamic multi-asset and multi-strategy approach, it offers a next-generation liquid alternative designed for stability and growth.
