Volatility Trading and Arbitrage Strategies › Dispersion Trading

Dispersion trading exploits the relationship between index volatility and the volatility of index constituents. The fundamental insight is that index variance depends on constituent variances and correlations between constituents. When you diversify by combining multiple assets, some of their movements cancel out, reducing overall portfolio volatility. This diversification effect creates a wedge between index volatility and the volatilities of individual stocks, and dispersion trading seeks to profit from this relationship.

For an index composed of $n$ stocks with weights $w_i$, the index variance is:

$$\sigma_{\text{index}}^2 = \sum_{i=1}^{n} w_i^2 \sigma_i^2 + \sum_{i=1}^{n} \sum_{j \neq i} w_i w_j \rho_{ij} \sigma_i \sigma_j$$

where:

• $\sigma_{\text{index}}^2$: variance of the index
• $n$: number of stocks in the index
• $w_i, w_j$: weight of stock $i$ and $j$ in the index
• $\sigma_i, \sigma_j$: volatility of stock $i$ and $j$
• $\rho_{ij}$: correlation between stocks $i$ and $j$

This formula applies portfolio variance mathematics to an index. It tells us that the total variance of the index comes from two distinct sources. Distinguishing these sources is the basis of dispersion trading.

The formula decomposes index variance into:

• $\sum w_i^2 \sigma_i^2$: the weighted sum of individual variances (diagonal terms), representing concentration risk
• $\sum \sum w_i w_j \rho_{ij} \sigma_i \sigma_j$: the weighted sum of pair-wise covariances (off-diagonal terms), representing correlation risk

The first component represents what would happen if each stock's movements were entirely independent: you would simply add up the variance contributions from each stock, weighted by the square of their index weights. The second component captures the interaction effects between stocks: when two stocks tend to move together (positive correlation), they add to index variance, while independent or negatively correlated movements partially cancel out.

This can be rewritten using average correlation $\bar{\rho}$:

$$\sigma_{\text{index}}^2 \approx \left(\sum w_i^2\right) \overline{\sigma^2} + \left(1 - \sum w_i^2\right) \bar{\rho} \bar{\sigma}^2$$

where:

• $\sigma_{\text{index}}^2$: variance of the index
• $\sum w_i^2$: sum of squared weights (measure of concentration)
• $\overline{\sigma^2}$: average variance of constituents
• $\bar{\rho}$: average pairwise correlation between constituents
• $\bar{\sigma}^2$: square of the average constituent volatility

This simplified form makes the role of correlation crystal clear. The first term depends only on individual volatilities and index concentration, while the second term scales directly with the average correlation. When correlation is zero, the second term vanishes entirely, and the index variance is driven purely by the concentration-weighted individual variances. As correlation increases, the second term grows, and the diversification benefit shrinks.

The key implication is that when correlations are high, index volatility approaches the average constituent volatility. When correlations are low, diversification reduces index volatility below constituent volatility.

A dispersion trade takes opposing volatility positions on an index versus its constituents. The two main configurations are:

Long Dispersion (Short Correlation):

• Sell index options (straddles or variance swaps)
• Buy constituent options

This trade profits when realized correlation is lower than implied correlation. It is typically the direction you choose because implied correlation tends to be higher than realized correlation, similar to the volatility risk premium. Index option buyers are often hedgers who pay a premium for protection against broad market moves. This demand pushes up index implied volatility relative to what the constituent implied volatilities would suggest.

Short Dispersion (Long Correlation):

• Buy index options
• Sell constituent options

This trade profits when correlations spike, typically during market stress. It is a tail hedge. During crises, correlations tend to spike as all assets sell off together, causing index volatility to surge relative to individual stock volatilities. A short dispersion position profits from this correlation spike.

In the low correlation scenario, the realized index volatility is significantly lower than the weighted constituent volatility, generating profit for the dispersion trade (short index, long constituents). When realized correlation is high, this diversification benefit disappears, and the trade suffers losses as the short index position moves against you. The dispersion trade is fundamentally a bet on correlation: if correlations come in lower than what option prices imply, the long dispersion trade profits.

You can express your views in terms of implied correlation, which can be backed out from index and constituent implied volatilities:

$$\rho_{\text{implied}} = \frac{\sigma_{\text{index}}^2 - \sum_i w_i^2 \sigma_i^2}{\sum_{i \neq j} w_i w_j \sigma_i \sigma_j}$$

where:

• $\rho_{\text{implied}}$: implied correlation derived from market prices
• $\sigma_{\text{index}}$: implied volatility of the index
• $w_i, w_j$: weights of constituents $i$ and $j$
• $\sigma_i, \sigma_j$: implied volatilities of constituents $i$ and $j$

This formula inverts the variance decomposition equation to solve for correlation. Given the observed implied volatility of the index and the implied volatilities of all constituents, we can back out the correlation level that would make these prices consistent. This implied correlation represents the market's collective expectation of how correlated stock returns will be over the option's life.

When implied correlation is high relative to expected realized correlation, a long dispersion trade (short correlation) is attractive. You are effectively betting that the market is overpricing co-movement and that individual stocks will move more independently than option prices suggest.