Guide

Volatility skew explained

Harbor Capital's tail-hedge budget bought one-month S&P 500 puts every quarter, but the risk desk noticed a problem: when VIX was already elevated, the puts they needed were even more expensive than ATM vol suggested. On 14 October 2008, 25-delta puts implied 62% vol while ATM implied 55% — a 7-point negative skew that turned a “cheap insurance” rule into an overpayment trap. After refactoring to size hedges off risk-reversal skew rather than VIX level alone, annual hedge drag fell from 1.4% to 0.9% of portfolio while 95th-percentile crisis drawdown improved 2.1 percentage points. Skew is not the same as the level of volatility; it is the shape of implied vol across strikes, and it tells you how much the market charges for left-tail protection relative to upside calls.

Volatility skew describes how implied volatility changes with strike price at a fixed expiry. Equity indices typically show negative skew (OTM puts trade at higher implied vol than OTM calls), while commodities and single stocks can show positive or smile-shaped surfaces. Skew interacts with but is distinct from the variance risk premium (average implied minus realized level) and from term-structure slope across maturities. This guide defines smile vs skew, shows how to measure 25-delta risk reversals and butterflies, explains economic drivers (leverage effect, crash demand, supply constraints), walks through Harbor Capital's skew-aware hedge refactor, compares techniques in a decision table, lists pitfalls, and ends with a production checklist alongside our options greeks and put-call parity guides.

Smile vs skew: reading the volatility surface

Plot implied volatility on the y-axis and strike (or moneyness) on the x-axis at one expiry and you get a cross-section of the volatility surface. Three patterns dominate:

Volatility smile — implied vol is lowest near ATM and rises for both OTM puts and OTM calls. Common in FX and after large moves when both tails look risky.
Volatility skew (smirk) — implied vol slopes monotonically: puts above ATM above calls (negative skew on equities) or the reverse (positive skew on some commodities expecting supply shocks).
Flat surface — rare in live markets; Black-Scholes assumes constant vol, which practitioners treat as a convenient fiction.

Traders quote skew with standardized delta strikes rather than raw moneyness because delta accounts for time to expiry and spot level. The workhorse quote is the 25-delta risk reversal: implied vol of a 25-delta call minus implied vol of a 25-delta put. On SPX, this number is usually negative (puts richer). A 25-delta butterfly (BF) measures curvature: how much ATM vol sits below the average of the 25-delta wings — useful when the surface is smile-shaped rather than purely skewed.

RR_25 = IV(25Δ call) − IV(25Δ put)     // negative = put skew on equities
BF_25 = 0.5 × (IV(25Δ call) + IV(25Δ put)) − IV(ATM)

Risk reversals are tradable: buy the call, sell the put (or vice versa) at the same delta to express a view on skew without much net delta. Market makers hedge skew exposure primarily through vanna and volga (second-order greeks), which is why skew can move sharply when spot gaps.

Why equity puts trade rich: drivers of negative skew

Several mechanisms reinforce persistent put skew on broad equity indices:

Leverage effect

When a stock index falls, company leverage rises mechanically (debt unchanged, equity down), increasing equity volatility. Black's (1976) leverage model links negative return shocks to higher subsequent vol. Options markets price this asymmetry: downside moves imply larger future vol than upside moves of equal size.

Crash insurance demand

Pension funds, structured products, and retail investors systematically buy OTM puts for portfolio insurance. Dealers who sell those puts are short skew and short gamma near strikes — they raise put implied vol to compensate. Demand spikes before known risk events (earnings, elections, FOMC) and after vol spikes when investors “chase” protection at already-elevated levels.

Supply and microstructure

Covered-call overwriting and call selling on indices add call supply, depressing call implied vol relative to puts. Corporate buyback call overlays have a similar effect on single names. Index option open interest clusters at round strikes, creating local kinks in the surface.

Jump and tail risk

Diffusion models underprice gap risk. Traders embed jump premium in OTM puts, steepening skew beyond what GARCH-style continuous vol models predict. This is one reason skew steepened into the 2008 and 2020 crises even when VIX was already high.

Measuring and monitoring skew in production

A practical skew monitor for allocators needs consistent inputs:

Tenor: one-month and three-month RR_25 are standard; match tenor to hedge horizon.
Surface source: vendor surfaces (CBOE, OptionMetrics) vs broker marks; align with execution venue.
Normalization: track RR_25 in vol points and as z-score vs five-year history for regime context.
Correlation with VIX: skew and level are correlated but not redundant; joint plots reveal when puts are expensive even for a given VIX.
Realized skew: compare implied RR to realized return asymmetry (downside semivariance vs upside) to see if insurance is overpriced.

Harbor Capital's dashboard plots 1M RR_25, VIX, and a skew percentile (current RR vs 10-year distribution). Hedge triggers fire on skew percentile > 80 and VIX < 30 — the combination that historically meant “calm surface, expensive tails” and cheap convexity was absent.

Trading and allocating with skew

Skew-aware strategies differ from pure level plays:

Protective puts — buy OTM puts when skew is flat (RR near historical median); avoid buying when skew is already steep unless crisis is imminent.
Collars — finance puts by selling OTM calls; call skew affects how much upside you surrender. See covered calls for the income side.
Risk reversal overlay — short RR (sell put skew, buy call skew) earns carry when skew mean-reverts; dangerous into crashes when skew steepens further.
Put spread vs outright put — when skew is very steep, vertical put spreads reduce vega cost by selling even-deeper OTM puts where implied vol is highest.
Skew vs VRP harvest — short variance risk premium via strangles sells both wings; you are short skew and short vol level simultaneously. Stress losses come from gamma and skew expansion together.

Worked example: Harbor Capital skew-aware tail hedge

Harbor's policy portfolio targets 55% global equities. A standing 2% annual budget funds tail hedges. Version 1 bought 3-month 95% puts whenever VIX < 20.

Version 1 problems (2009–2017)

Low VIX often coincided with steep put skew (post-crisis memory); puts were not cheap on a skew-adjusted basis.
Missed efficient entry in 2017 when VIX was low and RR_25 was near historical flats — puts were relatively inexpensive.
Average hedge drag 1.4% p.a.; three hedges expired worthless in calm 2013–2014.

Version 2 (2018 onward)

Size put notional off skew percentile: full budget when RR_25 z-score < −0.5 (flatter skew); half budget when z > +1.0 (steep skew).
Switch to put spreads when RR_25 percentile > 85 to cap vega spend on the steepest wing.
Pair with quarterly crisis regime check: if filtered crash probability > 0.4, buy outright puts regardless of skew (insurance when math says tail is live).
Result 2018–2025: hedge drag 0.9% p.a., one profitable March 2020 hedge (+3.8% portfolio contribution), avoided overpaying in late 2023 when VIX was moderate but skew was historically steep.

Technique decision table

Signal / tool	What it measures	Best when	Watch out for
25-delta risk reversal	Put vs call wing richness	Sizing OTM puts and collars; timing hedge entry	Tenor mismatch; stale surface after gaps
25-delta butterfly	Smile curvature vs skew slope	FX and post-event smile trades	Less informative when surface is pure skew
VIX level	ATM implied vol index	Quick fear gauge; vol targeting	Ignores strike shape; can mis-rank put cost
Variance risk premium	Implied minus realized level	Short-vol carry sleeves	Separate from skew; short vol is short skew too
Put-call parity	Arbitrage link calls/puts	Verify surface consistency	Borrow and dividend assumptions on singles
Skew swap / OTC	Direct RR exposure	Institutional skew mean-reversion	Counterparty, margin, crisis gap risk

Common pitfalls

Equating low VIX with cheap puts — skew can be steep in calm markets; puts are expensive on a wing-adjusted basis.
Using ATM implied for OTM hedge sizing — budget vega using 25-delta put implied, not ATM.
Ignoring tenor — one-month skew can invert vs three-month during events; match hedge expiry to risk horizon.
Confusing skew with smile — butterfly-rich surfaces need different trades than monotonic skew.
Short skew without crisis stop — selling RR earns carry until crashes steepen skew further; pair with drawdown limits.
Single-stock skew on index books — earnings skew on a constituent does not map to SPX RR.
Stale surfaces after halts — implied vol marks lag when underlying cannot trade; do not trust RR intraday around halts.
Overfitting skew z-scores — post-2008 regime shifted baseline skew; use rolling windows and structural breaks.

Production checklist

Define skew metrics (RR_25, BF_25) and tenors aligned to hedge horizon.
Source consistent volatility surface data; document interpolation method.
Plot skew percentile vs VIX level; identify joint “cheap put” zones.
Size tail hedges off skew-adjusted put cost, not ATM vol alone.
Use put spreads when skew percentile exceeds policy threshold.
Override skew rules when crisis regime probability exceeds cut-off.
Stress-test hedges for 2008, 2011, 2015, 2018, 2020 skew paths.
Report hedge drag and skew at entry in LP attribution notes.
Reconcile short-vol sleeves for implicit short-skew exposure.
Review skew model after major policy shocks (QE, vol targeting changes).

Key takeaways

Volatility skew is the slope of implied vol across strikes — equity indices typically show negative skew with OTM puts trading rich.
Measure skew with 25-delta risk reversals and butterflies; delta standardization makes quotes comparable across spot levels.
Skew is distinct from VIX level and variance risk premium — low fear does not mean cheap tail insurance.
Harbor Capital cut hedge drag by sizing puts off skew percentile and using spreads when wings were historically expensive.
Short-vol and short-skew strategies earn carry until crises steepen skew — size both with explicit tail protocols.