Arbitrage and relative value
Volatility trading
Volatility traders use options to bet on how much prices will move, not which way. The core trade compares implied volatility – what option prices assume – with realised volatility, buying options that look cheap and selling those that look dear, and hedging the direction away. Around it sit relative-value trades across strikes, expiries and between an index and its members.
At a glance
At a glance
- Makes money from
- The gap between implied and realised volatility, and mispricings across the surface
- Holds positions for
- Days to months
- Trades
- Options, variance swaps, volatility index futures, index against single-stock options
- Biggest risk
- Sudden volatility spikes when short volatility; steady bleed when long it
- Who does it
- Options trading firms such as SIG, Optiver and Five Rings, volatility funds and bank derivatives desks
- 19.3 vs 15.1 – average VIX against realised S&P 500 volatility (1990–2018, Cboe)
- 82.69 – record VIX close (16 March 2020)
- +116% – VIX’s largest one-day rise (5 February 2018)
- −96% – a short-volatility note’s fall that day (XIV)
- 2004 – VIX futures launched (options followed in 2006)
- 1% – daily move at 16% volatility (the rule of 16)
The business
An option’s price is a forecast of volatility. Buy one and hedge its delta, and you profit if the market moves more than the price assumed; sell one and hedge, and you profit if it moves less. A volatility trader’s whole job is deciding which side of that forecast to be on – and doing it across thousands of strikes and expiries where the forecast is not the same.
On average the forecast is too high for equity indices. From 1990 to 2018 the VIX – the S&P 500’s 30-day implied volatility – averaged 19.3%, while the volatility that followed averaged 15.1%. Investors pay up for protection, so selling options and hedging collects a volatility risk premium most of the time. It is not free money: the premium is compensation for losing a great deal in the occasional crash, and some studies find it much smaller in particular periods.
The players differ in what they do with it. Options market makers sit in the middle and try to stay flat; volatility funds take views; bank derivatives desks inherit volatility from the structured products they sell and trade around it; dispersion desks trade index volatility against the volatility of its members.
Implied volatility usually sits above realised – until it does not
A trade, step by step
Gamma scalping is the purest form of the trade: buy options, hedge the delta every day, and the hedging itself makes money if the market moves enough. Buying low and selling high is built into the hedge – when the price rises, a long straddle’s delta rises, so you sell shares to stay flat; when it falls, you buy them back.
Worked example
A hedged straddle over five days
Buy a one-month at-the-money straddle – a call and a put struck at 100 – on 100 shares, at 20% implied volatility. It costs 4.57 a share, $457 in all. Its gamma is 0.1391 and its time decay 0.0762 a share a day. Hedge back to flat at every close.
- Day 1: the stock rises 2.00+$20.20
- Day 2: it falls 1.50+$8.03
- Day 3: it rises 2.50+$35.85
- Day 4: it falls 1.00−$0.67
- Day 5: it rises 1.50As day 2+$8.03
Five days, +$71.44. The moves averaged 1.78 a day – about 34% a year against the 20% the straddle was priced at – so the gains from re-hedging beat the decay on four days of five. A day that moves less than about 1.05 loses, which is the implied daily move at 20%.
A long straddle
The maths that matters
A delta-hedged option earns, over a short interval, about . With gamma 0.05 on a 100 stock, 20% implied and 30% realised, that is a day per option. Notice it depends on where gamma is when the moves happen – which is why the same realised volatility can make or lose money depending on the path.
Variance swaps take the path out. They pay on realised variance directly, with , and can be replicated with a strip of options weighted by . Because the payoff is in variance, it is convex in volatility: a long variance swap gains more than a volatility swap when volatility is high and loses less when it is low.
Variance swap against volatility swap
Dispersion trades the gap between index volatility and the volatility of its members, which is a bet on correlation. For an index, , so index and stock implied volatilities together imply a correlation. Two equally weighted stocks at 30% each and an index at 24% imply . Selling index volatility and buying single-stock volatility is short correlation.
The VIX futures curve
Where the money comes from
- The volatility risk premium: selling index options, hedged, and collecting the gap between implied and realised – most of the time.
- Skew: out-of-the-money puts trade at higher implied volatility than calls; trade the steepness of that slope against its history and against realised skew.
- Term structure: short-dated against long-dated volatility, which usually slopes upward and inverts in a sell-off.
- Events: earnings, central-bank meetings and elections are priced as a jump in the volatility for that date; estimating the size of the move better than the market is a business in itself.
- Dispersion: index volatility against stock volatility – a view on whether stocks will move together.
Ways to hold volatility
| Instrument | Gives you | Watch out for |
|---|---|---|
| Hedged options | Gamma and vega, strike by strike | Path dependence; hedging costs |
| Variance swap | Realised variance, no hedging | Convex – short variance can lose without limit |
| VIX futures | 30-day implied volatility, forward | Roll cost in contango |
| Dispersion | Index against stock volatility | Correlation jumps in a crash |
Risk and what goes wrong
The danger of being short volatility is a single bad day; the danger of being long it is a slow bleed of time decay while nothing happens. The documented blow-ups are almost all volatility sellers.
The skew appears · 19 October 1987
The Dow fell 22.6% in a day. Index implied volatility had been roughly flat across strikes; since then, downside puts have always traded at a premium.
The VIX’s intraday record · 24 October 2008
At the height of the financial crisis the VIX reached 89.53 intraday, a record that still stands.
“Volmageddon” · 5 February 2018
The VIX rose 116% in a day, from 17.31 to 37.32. XIV, an exchange-traded note that was short VIX futures with about $1.9 billion in it, lost about 96% and was closed; a fund selling S&P options lost about 80% in two days.
The record close · March 2020
The VIX closed at 82.69 on 16 March. A variance-swap fund wound down, leaving a bank with a $214 million loss, and a large manager’s options-selling funds lost up to 97%.
A spike in thin markets · 5 August 2024
The VIX printed 65.73 before the US open, then closed near 40. Thin liquidity had widened option quotes, which lifted the index more than the market itself had moved.
Technology and data
- Clean option prices and implied-volatility surfaces, with arbitrage-free fits, across every expiry.
- Realised-volatility estimates from intraday data, and models of how volatility clusters and mean-reverts (GARCH and its relatives).
- Event calendars and models of the move each event implies.
- Risk by strike and expiry, with scenarios that shock spot and volatility together – they move in opposite directions for indices.
Who works here
Volatility trader
Decides where the book should be long or short volatility, by expiry and strike, and executes it through options, variance swaps or futures. Watches realised volatility, event pricing and skew against history every day.
Quant researcher
Builds volatility forecasts, event-move estimates and skew models; studies the volatility risk premium and when it fails; tests hedging rules and their costs.
A day on the desk
Re-mark the surface before the open, check overnight moves and the event calendar. Through the day, hedge the delta, trade around flows and re-assess the book’s exposure to a large move. After the close, attribute P&L to theta, gamma and vega, and compare realised with implied.
In interviews for this desk
Expect the Greeks, how the volatility surface behaves, Black–Scholes intuition rather than derivation, and statistics on realised volatility. Options games where you must hedge as the price moves are common.
You hold a hedged option with gamma 0.05 on a 100 stock. Implied is 20%, realised 30%. What do you make a day?
About per option. Realised above implied pays the long-gamma holder.
What is a variance swap’s payoff with $100,000 vega notional, strike 20, if realised volatility is 25?
Variance notional is per variance point, so the payoff is . A volatility swap would pay $500,000; the extra is the variance swap’s convexity.
Two equally weighted stocks each have 30% implied volatility and the index has 24%. What correlation is implied?
, so and .
Why do index puts trade at higher implied volatility than calls?
Investors are net buyers of downside protection, and index crashes come as jumps that a lognormal model underprices. Index volatility also rises when the index falls, so a put is worth more than a constant-volatility model says.
Why does a long VIX futures position tend to lose money over time?
The curve is usually in contango: each future is above the VIX and drifts down towards it as expiry nears. Holding long and rolling means repeatedly selling low and buying high – the roll cost that short-volatility products collect.
Roughly what does a one-month at-the-money straddle on a 100 stock at 20% volatility cost?
About . Black–Scholes gives 4.57 with no rates.
Firm guides: SIG (Susquehanna) · Optiver · Five Rings
Glossary
- Implied volatility
- The volatility an option’s price implies.
- Realised volatility
- The annualised standard deviation of actual returns.
- Volatility risk premium
- The average excess of implied over subsequently realised volatility.
- Gamma scalping
- Re-hedging a long-gamma position to capture realised moves.
- Straddle
- A call and a put at the same strike: a bet on movement, not direction.
- Variance swap
- A forward contract on realised variance.
- Vega notional
- A variance swap’s size, in dollars per volatility point near the strike.
- Dispersion
- Trading index volatility against the volatility of its members.
- Implied correlation
- The correlation that makes index and stock implied volatilities consistent.
- Contango
- A futures curve sloping upward with expiry.
- Backwardation
- A futures curve sloping downward with expiry.
- VIX
- Cboe’s index of 30-day implied volatility on the S&P 500, from a strip of options.
Further reading
- Trading Volatility: Trading Volatility, Correlation, Term Structure and Skew – Colin Bennett, 2014. A practitioner’s survey of every volatility trade
- Volatility Trading – Euan Sinclair, 2013. Second edition, Wiley
- More Than You Ever Wanted to Know About Volatility Swaps – Kresimir Demeterfi, Emanuel Derman, Michael Kamal and Joseph Zou, 1999
- Variance Risk Premiums – Peter Carr and Liuren Wu, 2009
- Option Volatility and Pricing – Sheldon Natenberg, 2014