Dynamic Position Sizing: Insights, Implications, and the Outlier Hunting Perspective

Introduction

Quantica Capital has long been a beacon of rigorous quantitative research in systematic trading. Their latest paper, “Let it Ride or Lock it In? The Impact of Dynamic Versus Static Position Sizing on Trend-Following Performance,” offers a detailed and thoughtful exploration of the effects of three position-sizing methodologies on trend-following strategies: Static Contracts, Static Notional, and Dynamic (volatility-adjusted). Using data from 2,750 trades across 50 futures markets spanning three decades, the paper examines how these methodologies influence trade-level and portfolio-level performance.

Critique

1. The Limitations of Sharpe Ratio: Quantica’s preference for DPS largely stems from its superior Sharpe ratio. However, while the Sharpe ratio is a widely accepted measure of risk-adjusted returns, it is not fully aligned with the goals of trend-following. Trend-following strategies rely on capturing positive skew through a small number of large, profitable trades that often come with higher volatility. By penalizing all volatility equally, the Sharpe ratio fails to distinguish between detrimental fluctuations and the beneficial volatility that accompanies outlier trends, thus misrepresenting the strategy’s true performance dynamics. However, Sharpe’s penalization of all volatility–including beneficial volatility–makes it an imperfect measure for trend-following strategies. Trend-followers thrive on positive skew and fat-tailed distributions, where a few large winners dominate returns. By treating the volatility of outlier trends as risk, Sharpe overlooks the very essence of trend-following success. Path dependent metrics like the MAR ratio or trade-level equity measures (e.g., closed-trade equity) may better capture the true risk-return dynamics of trend-following. These metrics align more closely with the goal of preserving capital while embracing the upside potential of beneficial volatility.
2. Volatility as Opportunity, Not Risk: Classic trend-followers view volatility as a feature, not a bug. For instance, during the unprecedented price surge in cocoa futures between 2023 and 2024, the accompanying sharp increase in volatility was integral to capturing significant returns. Static position sizing allowed for consistent exposure to this trend, fully capitalizing on its potential, whereas a dynamic approach would have curtailed position sizes during the most volatile and profitable phases, thereby limiting the upside. This highlights how volatility, when managed at the portfolio level, can act as a powerful ally in trend-following strategies. Outlier hunting thrives on price dislocations, which are often accompanied by sharp increases in volatility. DPS, by scaling down positions during such periods, limits exposure to the very trends that drive long-term returns. While this approach mitigates concentration risk, it inadvertently sacrifices the “outlier effect,” where large winning trades disproportionately contribute to performance
3. Increased Complexity and Reduced Robustness DPS introduces additional layers of complexity by requiring more parameters and frequent adjustments. This increases the risk of overfitting and reduces the robustness of the strategy. Simplicity is a hallmark of successful trend-following systems, and the added intricacy of DPS may dilute this advantage.
4. Trade-Level Dynamics: Quantica’s analysis shows that DPS reduces position sizes by a median of 30% during the most profitable trades. This reduction impacts overall profitability by limiting the exposure to these highly lucrative trends, thereby capping potential returns. While this approach helps moderate risk during volatile periods, it also restricts the ability to fully capitalize on outlier trades, which are critical for driving long-term performance in trend-following strategies. Static models, by maintaining consistent position sizes, avoid this limitation and are better positioned to harness the full potential of such trends. While this moderates risk, it also caps upside potential. The trade-level data highlights that static models–despite their higher volatility–consistently capture larger returns from outlier trends. This reinforces the notion that managing risk at the portfolio level, rather than at the trade level, is a more effective approach for trend-followers. By focusing on portfolio-level risk, practitioners can maintain exposure to the types of trends that generate outsized returns while diversifying across markets to mitigate the impact of any individual trade’s volatility. This approach allows for broader risk distribution without dampening the core principle of capturing outliers. Trade-level adjustments like those seen in DPS may inadvertently curb the very opportunities trend-followers seek to exploit. Portfolio-level strategies, in contrast, align more closely with the overarching objectives of trend-following: to harness trends, preserve simplicity, and optimize long-term compounding potential.
5. Regime Dependency and Serial Correlation Clustering: While Quantica’s research draws from several decades of data, we question whether its conclusions fully account for regime dependency. Market conditions over the past 30 years, including extended periods of volatility suppression post-GFC, may bias these findings toward specific methodologies. For example, the post-GFC environment favored smoother, lower-volatility strategies, whereas earlier periods like the high-inflation 1970s or the dot-com bubble might have supported more aggressive or concentrated approaches. These regime differences suggest that the effectiveness of position-sizing methodologies can vary significantly depending on macroeconomic conditions and market structure. Different regimes will naturally favor different approaches, and a broader analysis that considers varying macroeconomic and market conditions could yield more nuanced insights. Additionally, the well-known principle that serial correlation clusters suggests that more aggressive, concentrated strategies could thrive during such periods. These clusters often signal the persistence of trends, making a static or concentrated approach better suited to capitalize on sustained price movements, particularly when regimes favor extended directional trends.
6. Primacy of Absolute Returns In trend-following as a standalone strategy, absolute returns often take precedence over the smoothness of returns. Trend-followers are not primarily concerned with minimizing volatility but with capturing the outsized gains that arise from outliers. A focus on smoothing returns, as seen with DPS, risks compromising long-term compounding potential in favor of short-term aesthetic appeal. For practitioners, the emphasis should remain on maximizing total returns while managing risk, rather than overly prioritizing Sharpe improvements.

Conclusion