Volatility Harvesting vs. Carry: Two Distinct Frameworks for Monetizing the Same Position

Abstract

Same $1.70M proxy basket. Same 36-month window. Two different ways to sell options on it.

Carry collected $1.73M cumulatively — $48,000 average monthly net, Sharpe 1.4, max drawdown −11.2%. Volatility harvesting collected $1.09M — $30,300 average monthly net, Sharpe 0.95, max drawdown −4.6%.

Carry earned 59% more premium. Harvesting cut drawdown by more than half. The tradeoff is the entire decision: steady collection with structural short-gamma exposure, or lumpy collection with most months flat. What follows defines each framework, compares the numbers, and states when each applies.

1. Carry: Systematic Collection at Fixed Delta

Carry is a schedule, not a forecast. Each month, sell the 25–30Δ call on each proxy component at 30 DTE, roll on expiry, and hold the 5–8% OTM put spread underneath at 60 DTE. Delta target stays fixed regardless of IV rank, skew, or recent realized vol.

Mechanics for the reference basket ($1.70M notional, 60% small-cap industrials ETF + 40% two liquid single-names):

• Premium leg: 30Δ short calls, 30 DTE, monthly roll. Gross collection averages $59,500/month at sample-period IV (ETF leg 30%, single-names 40–45%).
• Hedge leg: 6% OTM long put / 14% OTM short put, 60 DTE, on the ETF leg. Amortized cost averages $11,500/month.
• Net: ~$48,000/month average over 36 months. Position delta held at −0.05 to −0.10 vs. the basket.

The framework monetizes the persistent gap between implied and realized volatility. In the sample, 30Δ call IV averaged 32% against 27% subsequent realized — a 5-point spread collected 30 out of 36 months. Win rate: 83%. Losing months cluster in sharp rallies, when the short calls go ITM faster than theta accrues.

Carry’s exposure is continuous. The short calls are always on, which means short gamma is always on. In a +8% single-month rally in the proxy, the carry book lost $96,000 that month — two months of premium reversed in one move. That is the structural cost: the strategy is short the exact move that hurts the micro-cap least (upside) and collects steadily otherwise.

2. Volatility Harvesting: Selling Only Into Spikes

Volatility harvesting inverts the schedule. Default position: flat. No short calls, only the long put-spread hedge held for tail definition. The sale triggers only when proxy IV spikes above a threshold — here, 30-day ATM IV above its 252-day 75th percentile, confirmed by a 5-day realized-vol expansion of at least 30%.

When triggered, the sale is larger and shorter-dated than carry: 20–25Δ calls or 15Δ strangles at 14–21 DTE, sized at 1.5× the standard carry notional, closed at 50% of max profit or at 7 DTE, whichever comes first. Then back to flat.

Over the same 36 months, the trigger fired in 11 months. The other 25 months the premium leg did nothing:

• Active months: average net +$112,000 per triggered month ($123,500 gross premium, same $11,500 hedge cost).
• Inactive months: −$11,500/month — hedge cost with zero offset.
• Win rate on triggered sales: 73% (8 of 11 profitable). Average holding period: 9 trading days.
• Cumulative: $1.09M net over 36 months.

Harvesting monetizes dislocation, not the average IV-RV spread. It captures the weeks when 14-DTE premium annualizes above 60% IV, then steps aside. The cost is 25 months of unrecovered hedge spend: −$287,500 in aggregate drag.

3. Side-by-Side: 36-Month Comparison

Both frameworks run on the identical basket, identical hedge leg, identical $1.70M notional. Only the premium leg differs.

The table reads plainly. Carry wins on total premium, average month, median month, and Sharpe. Harvesting wins on drawdown depth, worst month, and best month. Median monthly net for harvesting is negative — the typical month loses $11,500 — while the mean is positive. That skew is the point: harvesting accepts chronic small losses to concentrate exposure in high- balanced expected-value windows.

4. Drawdown Signatures Differ Because Exposure Differs

Carry drawdowns are event-driven and correlated with proxy rallies. The −11.2% maximum drawdown came in months 14–15 of the sample: a two-month +13% run in small-cap industrials pushed three consecutive short-call tranches ITM. Rolling up and out preserved some premium but locked in intrinsic losses. Recovery took four months of uninterrupted theta.

Harvesting drawdowns are time-driven, not price-driven. The −4.6% maximum was not a market move — it was four consecutive months without a trigger, each deducting $11,500 in hedge cost. No gamma event, no forced roll, no intrinsic loss. Recovery required a single triggered sale.

This distinction matters for pairing with the underlying micro-cap. Carry’s worst months coincide with rallies, when the illiquid long is typically appreciating — the overlay loss is partially offset at the portfolio level. Harvesting’s worst months are calendar bleed during quiet markets, uncorrelated with the micro-cap’s direction. One framework’s drawdown hedges itself against the position it monetizes. The other’s does not.

5. Which Framework for Which Book

Use carry when the objective is maximum recurring income against a stable proxy and the book can tolerate short-gamma months. Requirements: IV-RV spread persistently positive (check trailing 12-month average above 3 points), proxy ADV sufficient for monthly rolls without crossing more than 10% of quoted spread, and operational capacity to manage 36 rolls a year across legs.

Use volatility harvesting when the constraint is drawdown or when the IV-RV spread is thin on average but wide in bursts. In the sample’s lowest-quartile IV regime (sector IV below 22%), carry net carry fell to +$18,000/month — barely above hedge cost — while harvesting simply sat out and lost $11,500/month. Neither earns in dead vol. Harvesting loses less badly and avoids paying away upside in grind-up markets.

A combined allocation is workable: 60–70% of premium notional to systematic carry, remainder reserved for spike sales at 1.5× size. In the sample, a 65/35 blend produced $1.51M cumulative with −8.4% max drawdown. The spike-sale trigger must be codified (IV percentile + realized-vol confirmation + DTE cap) or the harvesting sleeve becomes ad-hoc overwriting.

6. Limitations

Three boundaries apply to both frameworks equally.

Same residual. Neither framework touches the ~22% idiosyncratic variance of the micro-cap. A firm-specific −30% gap on no news in the proxy leaves both overlays unaffected — the proxy does not move, the options expire as priced, and the micro-cap loss stands alone.

Same vol dependence, different shape. Carry needs average IV-RV positive. Harvesting needs episodic spikes — at least 3–4 qualifying dislocations per year at the calibrated threshold. A market that grinds upward on steadily declining IV starves both: carry collects shrinking premium against rising assignment risk, harvesting never triggers.

Threshold fragility. The harvesting backtest is sensitive to the trigger definition. Moving the IV entry from the 75th to the 65th percentile adds 6 triggered months, lifts cumulative to $1.31M, and deepens max drawdown to −6.8%. Moving it to the 85th percentile cuts triggers to 6, drops cumulative to $0.72M, and flattens drawdown to −3.9%. Report the sensitivity alongside the headline — a single-threshold backtest without perturbation bands is incomplete.