Dispersion trading takes opposite volatility positions on an index and on its constituents. The classic version sells index volatility and buys single-stock volatility. It is in substance a correlation trade: it gains when the stocks move a lot individually but offset each other, so the index stays calm.
How it works
Index variance depends on the variance of each component and on how they co-move. A common proxy is: implied correlation ≈ index implied variance / (weighted average single-stock implied volatility)². Index options usually price a higher correlation than the one later realised, partly because investors buy index puts for protection. A short index / long components position, built with options or variance swaps, is short correlation. The legs are sized vega-neutral, theta-neutral or correlation-weighted, and each choice leaves a different residual exposure.
Example
Hypothetical: index implied volatility is 18% and the weighted average implied volatility of its components is 30%. Implied correlation is about (18 / 30)², or 0.36. Over the life of the trade, the index realises 15% while the stocks still average 30%. Realised correlation comes out at about (15 / 30)², or 0.25. The short correlation position profits from the 0.11 gap, before costs and hedging slippage.
On the desk
The classic pitfall is a market crash. Correlations jump toward 1, index volatility explodes, and the short index leg loses faster than the single-stock legs gain. Liquidity on the smaller names also caps the size that can be traded.
Educational definition. Worked examples are hypothetical. Not an offer or a recommendation.