Index Dispersion (Sell Index Vol, Buy Single-Name Vol)

Overview

Here is the observation the whole trade rests on. If you look at the implied volatility of an index option, and then you look at the implied volatilities of the options on all the individual stocks inside that index, the index number is consistently lower than a simple weighted average of the components. That is normal and expected, because the stocks do not all move together and the diversification cancels some movement out.

But if you back out how much co-movement the market is pricing, that number, called implied correlation, is usually higher than the correlation that actually shows up. The market pays too much for index protection relative to single-name protection, because institutions hedge portfolios with index options.

Dispersion trading monetises that gap. You sell volatility on the index and buy volatility on the components. If the stocks end up moving around a lot but in different directions, you win: the single-name legs pay off and the index barely moves. If everything crashes together, you lose, badly.

Strategy logic

• Short leg: Sell an at-the-money straddle on the index, delta hedged. Or, if you have access, sell an index variance swap, which is cleaner because it gives pure volatility exposure without the path dependence of hedging a straddle.
• Long leg: Buy at-the-money straddles on a basket of the largest and most liquid components, also delta hedged. Weight them roughly by their index weights, adjusted for what you can actually trade.
• Neutrality: Size the two legs so total vega is balanced. If you match notional instead of vega you have accidentally taken a directional volatility bet rather than a correlation bet.
• Hedging: Every leg needs its delta re-hedged, typically daily. This is what turns a collection of option positions into a clean bet on realized versus implied volatility.

What you own after all of that is, in economic terms, a short position in implied correlation. You are betting the stocks will disperse more than the option market thinks.

Parameters (knobs)

• Basket size: Full replication of a large index is impossible in practice. Most desks trade the top 30 to 50 names. Fewer names means cheaper execution and a worse hedge.
• Structure: Straddles are available to everyone but require constant hedging and are path dependent. Variance swaps are cleaner but are over the counter and require a dealer relationship.
• Tenor: One to three months is typical. Longer tenors have less gamma noise but tie up capital and carry more correlation regime risk.
• Entry trigger: Some run it always on. Others only put the trade on when implied correlation is in the top quartile of its history, which is a genuinely sensible filter.

Where it works and where it hurts

It works in what practitioners call a stock-picker's market: earnings seasons, sector rotations, periods where individual companies have their own stories and the index itself grinds along quietly. Stocks move, they move in different directions, the index goes nowhere, and you collect on both legs.

It fails in exactly one scenario, and it fails catastrophically. In a genuine market-wide panic, correlation goes to one. Every stock falls at the same time. The index moves as much as the components do, which means the index straddle you sold explodes in value while the single-name straddles you bought gain much less than they need to. The diversification you were paid to bet on simply stops existing at the moment you need it.

This is not a hypothetical. It is what dispersion books did in 2008 and again in March 2020. The trade is a short position in systemic risk, dressed up in relative-value clothing.

Backtest design checklist

• Model the execution cost seriously. You are trading options on 30 or more names plus the index, entering and exiting, and re-hedging every leg daily. The spread costs are not a footnote, they are the difference between a profitable strategy and a losing one.
• Use real chains, not model prices. Single-name option spreads are much wider than index spreads. A backtest at mid prices will produce results you can never achieve.
• Include the hedging P&L. Delta hedging is not free and it is not neutral. The gamma path determines a large fraction of the outcome.
• Do not assume you can trade every component. Screen for liquidity as of the historical date, not today.
• Simulate the correlation spike. Run the position through a scenario where every stock drops 15 percent on the same day. That is your real risk number.

Common failure modes

• Vega neutral is not risk neutral. You can be perfectly vega balanced and still be enormously exposed to a correlation jump. Balancing vega hides the actual risk rather than removing it.
• Underestimating operational load. Dozens of legs, daily hedging, corporate actions, earnings dates, index rebalances. Operational error is a real and underappreciated source of loss here.
• Basket drift. The names you picked stop being representative of the index. Your hedge quality degrades silently.
• Selling correlation when it is already low. The premium is thin, the downside is unchanged. If implied correlation is already near historical lows, you are taking the same tail risk for much less compensation.

Our notes and suggestions

Be clear-eyed about what this is. Dispersion is presented as a market-neutral relative-value trade, which makes it sound safe. It is not. It is a short position in the thing that goes wrong in every crisis, and it has ended funds.

For most people the honest answer is that this is not implementable. It needs an option surface across many names, low institutional execution costs, a robust daily hedging system, and the balance sheet to survive a correlation event. If you do not have all four, the transaction costs alone will eat the entire theoretical edge before you get anywhere near the tail risk.

If you do run it, only put it on when implied correlation is genuinely elevated, size it as if a correlation-to-one event will happen during your holding period, and accept that it will. What would change our mind: implied correlation persistently trading below realized correlation, which would mean the premium has been arbitraged away and only the tail risk remains.