Topic 9 · ~9% Draw the payoff before you answer

Volatility & Complex Strategies

One question runs through the whole topic: who is short the tail? Identify that and most of these answers fall out.

Volatility as an asset class

Three volatilities — never mix them up. Realized volatility is computed backward from actual returns. Implied volatility is forward-looking, backed out of an option price given a pricing model — so it is a model-dependent statement about the market's expectation plus a risk premium. Forecast volatility comes from a model such as GARCH, which captures the two robust empirical facts: volatility clusters, and it mean-reverts.

The volatility risk premium. Implied volatility exceeds subsequently realized volatility on average and persistently, most clearly in equity indices. The economic explanation is insurance: investors are structurally long equities and pay up for protection, so option sellers earn a premium for supplying it. This makes short volatility a carry strategy — positive expected return, high hit rate, negative skew, and severe fat left tail. The distribution of returns is precisely that of an insurance underwriter.

Why it matters for an allocator. Many strategies that appear unrelated are short volatility in disguise: merger arbitrage, most relative value, carry trades in FX and credit, and anything earning a steady spread with rare severe losses. Sharpe ratios computed over a period without a crisis systematically overstate their quality, because the risk lives in a state the sample does not contain.

Memory hook
Long volatility = paying insurance premiums: many small losses, rare large gains, positive skew, negative carry. Short volatility = writing insurance: many small gains, rare large losses, negative skew, positive carry. Every strategy in this topic is one or the other, and saying which is usually the first sentence of a good answer.

Skew, smile & term structure

The surface exists because Black–Scholes is wrong. If returns were lognormal with constant volatility, one implied volatility would price every strike and maturity. Instead implied volatility varies systematically with both, and the shape encodes what the market believes about the return distribution.

Skew (equity indices). Downside puts trade at materially higher implied volatilities than upside calls. Two explanations, both examinable: the leverage effect (a falling equity price raises a firm's leverage and therefore its equity volatility), and crash-risk demand — structural hedging demand from investors who are long the index, met by dealers who require compensation. Practical implication: a collar financed by selling upside calls is selling cheap volatility to buy expensive volatility, which is why collars have an unfavourable structural pricing.

Smile (currencies, some commodities). A roughly symmetric elevation of both wings, reflecting two-sided jump risk — a currency can gap in either direction.

Term structure. Normally upward-sloping (contango), because longer horizons carry more uncertainty and because near-dated volatility is usually low in calm markets. It inverts in stress, when near-dated implied volatility spikes above long-dated. The practical consequence is severe: a constant-maturity long volatility position (a rolling VIX futures exposure) must repeatedly sell a cheaper near contract and buy a more expensive far one, bleeding roll cost in every calm period — the reason long-volatility ETPs decay so relentlessly.

Instruments and what they isolate

Straddle / strangle
Long volatility and long gamma, but the position acquires delta as the underlying moves, so without continuous delta hedging it becomes a directional bet. A delta-hedged straddle earns realized minus implied volatility — the cleanest expression of the volatility risk premium, and the reason hedging frequency and transaction cost matter so much.
Variance swap
Pays the difference between realized variance and a strike, with no delta hedging required and no path dependence in the exposure. Because it is linear in variance, it is convex in volatility: a doubling of volatility quadruples the payoff. A short variance swap therefore loses quadratically — the structural reason short-variance positions produce catastrophic single-day losses.
Volatility swap
Linear in volatility, so no convexity. Cheaper protection in a spike, and its fair strike sits below the variance-swap strike; the gap between the two is a function of the volatility of volatility.
VIX futures / ETPs
Reference expected 30-day implied variance of the S&P 500, not spot VIX, and you cannot hold spot VIX. Roll cost in contango is the dominant term over any medium horizon. Inverse and levered products carry path-dependency and gap risk — the February 2018 episode is the standard case study.
Collar / risk reversal
Buy a put, sell a call. Cheap or zero-cost protection at the price of the upside, and it interacts with skew as above. Useful for a concentrated position where outright sale is undesirable — subject to constructive-sale tax rules.
Structured note
A bond plus embedded options in a single wrapper. Decompose it: the investor is almost always selling optionality (a knock-in put, a cap on participation) to fund a headline coupon, and takes the issuer's credit risk on top. Also check the embedded fee — the difference between the note's issue price and the sum of its replicating parts.

The Greeks in practice

Delta
Sensitivity to the underlying. Also the hedge ratio, and approximately the risk-neutral probability of finishing in the money for a call.
Gamma
Rate of change of delta. Highest at-the-money and near expiry. Long gamma means the hedge is rebalanced profitably (buy low, sell high); short gamma means the reverse, which is why short-gamma books lose most in fast markets.
Vega
Sensitivity to implied volatility. Highest at-the-money and for long-dated options, because more remaining time means more exposure to a change in the volatility assumption.
Theta
Time decay. Roughly the mirror of gamma: a long-gamma position pays theta, a short-gamma position collects it. This is the carry of a volatility position.
Rho
Sensitivity to interest rates. Usually second-order except for long-dated options.
Exam trap — gamma vs vega
Both measure sensitivity to "movement", but to different movements. Gamma responds to a move in the underlying and peaks near expiry; vega responds to a change in implied volatility and peaks for long-dated options. So a short-dated at-the-money option is high gamma and low vega; a long-dated at-the-money option is the reverse. A trader who wants exposure to a volatility repricing buys long-dated; one who wants exposure to actual movement buys short-dated.

Hedging in practice. Delta hedging is continuous in theory and discrete in reality, so hedging error grows with gamma, with the interval between hedges, and with transaction costs. In a gap move, delta hedging fails entirely — which is the mechanism behind every "our hedge did not work" post-mortem, and the reason static option hedges are preferred for genuine tail protection.

Complex strategies & the hidden short put

Map each strategy to its option-equivalent payoff. This is the single most productive framing in the topic.

Convertible arbitrage
Long the convertible (cheap embedded call), short the equity, hedged for delta. Long convexity/gamma, but short credit and short liquidity — the 2008 collapse came from financing withdrawal and forced deleveraging, not from a modelling error.
Merger arbitrage
Long the target, short the acquirer in a stock deal. Payoff is a small, high-probability spread with a large loss if the deal breaks — economically a short put on deal completion, correlated with market stress because financing and antitrust risk rise together.
Fixed-income relative value
Small, model-identified mispricings amplified with heavy leverage. Short liquidity and short funding; convergence trades widen before they converge, so the risk is not being wrong but being unable to hold — the LTCM lesson.
Carry (FX, credit, vol)
Earn a spread while the world is calm, lose it violently when it is not. Negative skew by construction.
Trend following / managed futures
The exception: a long-volatility, positively skewed profile resembling a long straddle, with many small losses and occasional large gains. This is why it is a genuine crisis diversifier — provided the crisis is a sustained trend rather than a one-day gap.

The measurement consequence. Sharpe ratios treat upside and downside deviation identically and reward strategies with high hit rates. Negatively skewed strategies therefore look outstanding in benign samples and are systematically over-allocated. Add skewness, kurtosis, maximum drawdown, and an explicit crisis-period return to any evaluation, and prefer a factor model that includes an option-like payoff term so the hidden short position is visible.

Tail-risk hedging & CPPI

The honest framing. A standing put program pays a reliable negative carry to buy convexity. Over long periods, that carry is expensive — index put protection is structurally rich because of the same crash-risk demand that creates the skew. So the question is never "does it work?" but "what does a unit of drawdown mitigation cost, compared with the alternatives?".

The alternatives to compare against: simply holding less equity (free, but gives up expected return symmetrically); holding high-quality long duration (works when the shock is deflationary, fails when it is inflationary — as 2022 demonstrated); trend-following as a convexity proxy (cheaper carry, but no protection in a one-day gap); and a dynamic de-risking rule. A complete answer compares the cost, the reliability, and the scenario in which each fails.

CPPI and portfolio insurance. Constant proportion portfolio insurance sets a floor, computes the cushion (portfolio value − floor), and holds risky exposure equal to a multiplier × cushion, rebalancing as values change. It replicates an option payoff synthetically without buying one. Its failure mode is gap risk: if the market falls faster than the rule can de-risk, the cushion is breached and the floor is broken — and once exposure is at zero, the strategy is "cash-locked" and cannot participate in a recovery. The same dynamic-hedging fragility amplified the 1987 crash.

Implementation notes worth marks: tail hedges should be sized against the specific liability or drawdown they defend, not against a notional; a monetisation policy — the rule for when profits on the hedge are crystallised and redeployed — is as important as the hedge itself, since an unmonetised hedge that expires is pure cost; and basis risk between the hedged index and the actual portfolio is a real residual exposure.

Arbitrage strategies in depth

Each of these is a long-convexity or short-convexity position dressed as a spread trade. Learn the legs, the hidden short option, and the event that breaks it.

Convertible arbitrage

Long the convertible bond, short a delta-hedged amount of the issuer's equity, often with the credit leg hedged via CDS. The manager is buying cheap embedded volatility and earning gamma from delta re-hedging, while carrying credit and interest-rate exposure.

Short option: financing and liquidity. The trade needs stock borrow and repo leverage; when both are withdrawn — 2008 — the position is liquidated regardless of its valuation. Short-selling bans are a specific, examinable tail risk.

Merger arbitrage

In cash deals, long the target and capture the spread to the offer; in stock deals, long the target and short the acquirer at the exchange ratio. The return is the spread, annualised over the expected time to close, times the probability of completion.

Short option: a deal-break put. Payoffs are many small gains and rare large losses when regulatory, financing or shareholder approval fails — and deal breaks cluster with market stress, so the strategy has equity-tail beta despite a low reported beta.

Fixed-income relative value

Small, model-identified mispricings between closely related instruments — on-the-run versus off-the-run, swap spreads, basis trades, curve trades — expressed with heavy leverage because the gross spread is a few basis points.

Short option: repo financing and margin terms. Convergence is usually right eventually; leverage decides whether you survive the divergence. This is the LTCM lesson, and the exam expects it named as a funding liquidity failure, not a modelling failure.

Capital structure & distressed

Relative value between claims on the same issuer — senior versus subordinated, debt versus equity, CDS versus cash bond. Distressed adds a legal dimension: absolute priority, fulcrum security identification, and the recovery outcome of a restructuring.

Short option: process and legal risk. Outcomes depend on negotiation and jurisdiction, holding periods are long and uncertain, and marks are model-based — so measured volatility understates the true risk.

Common thread for essays: these strategies harvest a premium for supplying liquidity and bearing event risk. Their Sharpe ratios flatter them because the sample usually excludes the event, their reported volatility understates risk when marks are stale, and their correlation with each other rises sharply in a funding shock — which is why a "diversified multi-strategy" allocation can prove to be one trade.

Confusion pairs

Gamma vs vega
Sensitivity of delta to the underlying (highest at-the-money and near expiry) vs sensitivity to implied volatility (highest at-the-money and long-dated).
Realized vs implied vol
Backward-looking from returns vs forward-looking from option prices. The persistent gap between them is the volatility risk premium.
Skew vs smile
Asymmetric — index puts richer than calls, from crash-risk demand vs symmetric, both wings richer, typical of currencies.
Variance vs volatility swap
Pays on realized variance and is convex in volatility (losses grow quadratically on a short) vs linear in volatility. Variance swaps blow up faster.
VIX vs VIX futures
Spot 30-day implied variance of the index, not investable vs a forward on it that bleeds roll cost in contango.
Contango vs backwardation
Upward-sloping curve, negative roll yield for a long vs downward-sloping, positive roll yield. Volatility curves invert in stress; commodity curves invert on scarcity.

Practice

Six multiple-choice questions in exam style, with the reasoning — not just the letter.

1. A trader wants exposure to a repricing of implied volatility rather than to actual movement in the underlying. The most appropriate position is:

A. Short-dated at-the-money options
B. Long-dated at-the-money options
C. Deep out-of-the-money short-dated puts
D. A delta-hedged one-week straddle
B. Vega is largest for long-dated at-the-money options. Short-dated options carry high gamma and low vega, so they express a view on realized movement rather than on the implied-volatility level.

2. Why does a short variance swap position lose more than a short volatility swap for the same volatility spike?

A. Variance swaps require delta hedging
B. The payoff is linear in variance and therefore convex in volatility
C. Variance swaps have longer maturities
D. Variance swaps carry counterparty risk and volatility swaps do not
B. Squaring the volatility term makes losses grow quadratically: a doubling of realized volatility quadruples the payoff owed. Neither instrument requires delta hedging, and both are OTC contracts with counterparty risk.

3. An investor holds a constant-maturity long VIX futures position for a year in a persistently calm market. The dominant driver of returns is:

A. Changes in spot VIX
B. Negative roll yield from a contango term structure
C. Dividend yield on the underlying index
D. Convexity gains
B. Maintaining constant maturity requires repeatedly selling the cheaper near contract and buying the more expensive next one. In sustained contango this roll cost dominates and produces the well-documented decay of long-volatility ETPs.

4. Merger arbitrage is best characterised as:

A. A long straddle on the target
B. A short put on deal completion, correlated with market stress
C. A market-neutral strategy with symmetric risk
D. A long-volatility diversifier
B. The payoff is a modest capped spread against a large loss if the deal breaks — a written put. Deal breaks cluster with financing stress and regulatory shifts, so the strategy's worst outcomes coincide with market drawdowns rather than diversifying them.

5. A CPPI strategy with a 90 floor and a multiplier of 5 faces a 25% overnight gap. The most likely outcome is:

A. The floor holds because the rule de-risks automatically
B. The cushion is breached, the floor is broken, and the strategy may be cash-locked
C. Exposure increases as the multiplier amplifies the cushion
D. The strategy converts into a long straddle
B. Dynamic replication assumes continuous trading. A gap prevents the de-risking from executing, so losses exceed the cushion. Once exposure is forced to zero, the portfolio cannot participate in any recovery — the structural argument for buying an actual option instead of replicating one.

6. Which strategy has a return profile most similar to a long straddle?

A. Fixed-income relative value
B. FX carry
C. Managed futures / trend following
D. Convertible arbitrage after credit hedging costs
C. Trend following produces many small losses in choppy markets and occasional large gains in sustained trends — positive skew, long-volatility-like. The others are all short-volatility profiles. Note the caveat: trend protects against extended drawdowns, not against single-day gaps.

Constructed-response practice

Write these under time. Each outline is the shape the rubric rewards, not a model answer to memorise.

Prompt A (15 minutes). An investment committee proposes funding a standing index-put tail hedge costing approximately 90bp per year from the equity allocation. Evaluate the proposal against two alternatives and make a recommendation.

Outline:
  • Frame the decision as cost per unit of drawdown mitigated, not as "protection versus no protection".
  • Explain why puts are structurally expensive: skew and crash-risk demand mean the insurance is priced by a market of forced buyers.
  • Alternative 1 — hold less equity: costs expected return symmetrically but has no carry, no basis risk, and no monetisation problem.
  • Alternative 2 — trend following: cheaper carry and positive skew, but fails in a single-day gap and requires a sustained trend.
  • Note the operational requirements of any option program: sizing against a defined drawdown, a written monetisation rule, roll management, and basis risk between the index and the portfolio.
  • Recommend, and state the condition that would change the recommendation (e.g. a near-term liquidity event that makes a specific dated drawdown intolerable).

Prompt B (12 minutes). A manager presents a relative-value credit strategy with a 2.4 Sharpe ratio over six years, no month worse than −1.1%, and a stated market-neutral profile. Explain the payoff profile you suspect, describe how you would test for it, and state what you would require before allocating.

Outline:
  • Name the suspicion precisely: a short-volatility / short-put profile earning carry, with risk concentrated in a state the sample omits.
  • Test 1 — moments: skewness, excess kurtosis, and the shape of the worst decile of months relative to the best.
  • Test 2 — regression against an option-like payoff (e.g. returns on a short index put) alongside credit spread and liquidity factors; a significant loading identifies the hidden position.
  • Test 3 — crisis-period behaviour and stress: how did the strategy or its predecessor behave in 2008, 2020 and 2022, and what leverage and financing terms support it now?
  • Requirements before allocating: transparency on gross leverage and financing, position-level liquidity, stress results at a defined shock, and sizing set against the tail loss rather than the reported volatility.
← 8 · Due Diligence Formula sheet →