Position Sizing in Trend-Following: Comparing Volatility Targeting, Volatility Parity, and Pyramiding

Introduction

Trend-following strategies are widely regarded for their long-term returns and diversification benefits. Yet, much of the literature often highlights entry and exit signals, leaving position sizing relatively underexplored. How positions are sized and managed can profoundly affect the shape and stability of a strategy’s equity curve, influencing not only total returns but also drawdowns, risk-adjusted metrics, and the overall investor experience.

This article addresses that gap by analyzing three different position sizing frameworks: Volatility Targeting (VT), Volatility Parity (VP), and Volatility Parity with Pyramiding (VP+P). We apply these methods to a proprietary trend-following engine trading 40 futures markets. Each position’s initial size is calibrated to target a 0.10% daily volatility contribution. VT maintains that level through weekly rebalancing, VP never adjusts the initial allocation, and VP+P adds more contracts as the market moves in multiples of 2x the initial risk distance, aggressively exploiting large trends.

We highlight aggregate results and a standout 2024 Cocoa trade, underscoring that position sizing is not a minor detail. Instead, it’s a strategic decision that can lead to vastly different return profiles.

Methodology

Data and Trading Engine: The study draws on 40 liquid futures markets representing diverse economic sectors. The proprietary trend-following engine identifies entries and exits based on medium-term momentum and volatility signals. While the engine’s specifics remain proprietary, our focus is on how distinct position sizing policies influence outcomes.

Risk Budget and Execution: At trade inception, each position is sized to target a daily volatility contribution of 0.10%. For a 100 million USD portfolio, this equates to roughly 100,000 USD of risk per trade per day. Under VT, positions are adjusted weekly to maintain this level. Under VP, the initial size remains fixed until the trade closes. Under VP+P, the initial size is the baseline, and each time the market price moves favorably by a 2x risk increment, we add another full layer of contracts, thereby pyramiding exposure to winning trades.

Results

Average P&L Trajectory of a Typical Trade

The above figure shows the average cumulative P&L evolution of individual trades, normalized over their typical life cycle. Each line represents one of the three sizing methods:

• Volatility Targeting (Green Line): Initially, all three lines start together since the initial sizing is identical. VT’s trajectory is relatively smooth and orderly. Its weekly rebalancing process dampens fluctuations—when volatility rises, the position scales down, preserving a stable risk profile. Conversely, if volatility falls, the position scales up slightly. This constant calibration tends to produce a steady but somewhat moderated performance curve. As the trade matures, the green line gently peaks and then tapers off as losing trades hit their stops.
• Volatility Parity (Blue Line): Without rebalancing, the blue line more closely reflects the unfiltered dynamics of the market. If the underlying trend remains stable, this approach can capture it fully without being “throttled” by volatility adjustments. While the average trajectory may still taper off at the end, it often does so from a slightly higher peak than VT, reflecting the method’s more pronounced exposure to occasional positive trend outliers. Yet, the absence of rebalancing means this approach can experience sharper ups and downs (though this may not be fully evident in the average plot).
• Volatility Parity + Pyramiding (Red Line): Initially similar to the other two, the red line diverges once a strong trend emerges. As the price moves in multiples of 2x the initial risk from the entry, the position size increases. This turbo-charges profitable trades, pushing the average peak well above the others. The catch is visible toward the end of the trade: when a reversal eventually comes, the larger position can suffer a swift drawdown as stops are triggered. Despite the dramatic retreat near the exit, the red line reaches an average profit almost 100bps higher than the other methods.

In summary, the average trade chart offers a snapshot of each strategy’s personality. VT is controlled and smooth, VP is slightly more raw and can eke out more gains on moderate trends, and VP+P explodes upward when trends are strong but can fall just as quickly at the end.

Aggregate Performance Over Four Decades

This long-term simulation spans decades (since 1980), revealing how compounding and rare large trends shape final outcomes:

• Volatility Targeting (Green Line): Produces substantial growth over time with a steady, more conservative equity path. Its volatility is well-managed, and while it achieves strong compounding, the ultimate terminal value is lower than that of the other strategies.
• Volatility Parity (Blue Line): Delivers even higher cumulative returns than VT due to its unadjusted exposure during favorably volatile trends. Although it may face somewhat larger fluctuations than VT, it benefits from not scaling down positions during high-volatility trends. Over decades, this results in significantly higher terminal equity than the volatility-targeted approach.
• Volatility Parity + Pyramiding (Red Line): The real spectacle unfolds here. By occasionally doubling and tripling up on winning trends, this method taps into rare but massive payoffs. Over a multi-decade horizon, a handful of these blockbuster trades can push the cumulative NAV into a completely different league. The red line soars far above the others, showcasing the power of leveraging fat-tail winners. However, it’s important to remember that these big wins come with more turbulent equity curves, deeper drawdowns, and prolonged volatility.

This long-term chart highlights the starkly different growth trajectories these sizing methods produce. Smooth and steady versus explosive and volatile—there’s a clear trade-off at play.

Trade-Level and Portfolio-Level Statistics

The tables offer a numerical deep dive:

Volatility Targeting:

Trade-Level: Average trade P&L is modest (13 bps), with a median negative (-10.8 bps), indicating that many trades are losers or break-evens. However, the worst trades are relatively contained, and the hit ratio (~42.5%) combined with moderate skewness (2.4) suggests a well-distributed set of outcomes with some positive outliers.

Portfolio-Level: Total return of ~16,828% since 1980 with an IRR p.a. of 11.46%. MDD ~25.65% and a monthly hit ratio at 60% show stable and relatively predictable performance. Sharpe and Sortino ratios are healthy, reflecting a good balance between return and risk.

Volatility Parity:

Trade-Level: Slightly better average P&L (14.7 bps) and a slightly higher best trade, showing that not trimming positions during favorable conditions helps capture more upside from moderate trends. The hit ratio remains similar to VT, around 42.4%, but the distribution skews a bit more positively.

Portfolio-Level: Total returns leap to ~30,014% with a higher IRR p.a. (12.83%). The MDD is similar to VT, but the overall risk-return profile is marginally improved. The monthly hit ratio is around 59%, showing consistent yet slightly bumpier performance.

Volatility Parity + Pyramiding:

Trade-Level: Average trade P&L (26.2 bps) is nearly double that of the other approaches, but the median is lower (-13.6 bps), indicating a greater reliance on rare, huge winners. The best trade (16,863.4 bps = 168%) is extraordinary, but the worst trades can be quite severe, and the hit ratio drops to 39.3%. This confirms a “lottery-ticket” style distribution, where most trades don’t pay off, but a few can massively boost the average.

Portfolio-Level: Total return surges to ~556,106%, with an IRR p.a. of 20%. However, annualized volatility more than doubles compared to the other methods, and the Sharpe ratio is lower due to that volatility. MDD (48.69%) is also deeper. The monthly hit ratio is a decent 56%, but monthly skewness (3.74) is very high, reflecting a portfolio that thrives on rare big winners.

These statistics translate the qualitative observations into hard numbers. VT and VP are more stable, while VP+P offers a shot at astronomical returns at the cost of higher volatility and less consistency.

Recent Example: Cocoa in 2024

The Cocoa market in 2024 offers a real-world case study. A powerful rally allowed the VP+P strategy to shine, pyramiding additional contracts as the market moved through each 2x risk increment, turning a single strong trend into a blockbuster event.

Deploying a pyramiding approach, this single trade delivered over 50% at the portfolio level, nearly ten times what a volatility-targeted approach would have captured. While VT and VP profited from the rally, their conservative or fixed positions paled in comparison to the windfall generated by pyramiding.

The Cocoa trade exemplifies the promise and peril of pyramiding. Catching a monster trend can yield head-turning returns, but such events are rare and come with significant emotional and financial costs during quieter periods.

Conclusion

Our analysis shows how position sizing methods can dramatically reshape a trend-following strategy’s risk and return profile. Volatility Targeting ensures stable risk exposure and smoother outcomes, while Volatility Parity removes the rebalancing constraint, slightly boosting long-term returns at the cost of somewhat higher volatility. Pyramiding changes the game entirely—it turns a trend-following program into a quest for the next “fat-tail” event, capable of generating massive returns but also imposing higher volatility, deeper drawdowns, and long stretches of lackluster performance.

Choosing the right approach depends on your objectives and risk tolerance. If you value stability, higher hit ratios, and a smoother equity journey, you might prefer Volatility Targeting or Volatility Parity. On the other hand, if you can accept prolonged underperformance and aren’t fazed by steep drawdowns in pursuit of exceptional windfalls, Pyramiding may be the right fit.

Position sizing is not an operational footnote—it’s a key strategic choice that shapes the entire return distribution. Just as signals determine when to be in or out of the market, how you scale your bets determines whether you achieve steady compounding or swing for the fences.

For any further details, feel free to contact me at carlo@concretumgroup.com