Dispersion Trade: Short Index Vol, Long Single-Stock Vol

Key Takeaways

• Dispersion trade is short index variance swap plus long vega-weighted single-name variance swaps; the residual exposure after netting is realized vs implied correlation.
• S&P 500 implied correlation (tracked by Cboe's DSPX) tends to sit 5 to 15 points above realized correlation in calm regimes, the structural edge the trade harvests.
• A common mistake: replicating the index with too few names, practitioners use at least 30 to 50 constituents to minimize residual basis risk from single-name idiosyncrasy.
• March 2020 saw realized correlation spike from 0.30 toward 0.80 in days; long-dispersion books suffered double-digit losses as the diversification effect collapsed instantly.

What It Is

The classic structure is short an index straddle or variance swap and long vega-weighted straddles or variance swaps on a subset of the index constituents. Aggregate dollar vega on both legs is matched so that a parallel shift in the whole volatility surface is neutral.

What remains is a pure exposure to the gap between the index implied vol and the basket-weighted average of single-name implied vols. That gap, after some algebra, resolves to an implied correlation. Long dispersion is short implied correlation. The trade wins when single-name moves offset at the index level, and loses when everything moves together.

Desks at every major equity derivatives bank run dispersion books, and Cboe publishes a dedicated Dispersion Index (DSPX) tracking the expected implied correlation of the S&P 500 over the next 30 days.

The Intuition

Index variance decomposes into an average single-name variance term plus a correlation term:

Index Variance ~= sum(w_i^2 * sigma_i^2) + 2 * sum over i<j of w_i * w_j * sigma_i * sigma_j * rho(i,j)

If you sell index variance and buy the weighted single-name variance, the first term cancels. What is left is a bet on the average pairwise correlation rho. The implied correlation baked into option prices tends to be higher than realized correlation during calm periods, because end-users pay up for index puts as portfolio hedges. That structural overpayment is the harvestable edge.

Earnings season makes it worse. When companies report on staggered days, single-stock moves are large and idiosyncratic, while the index itself goes nowhere. Realized correlation falls and dispersion pays.

How It Works

Execution paths vary. The most common in practice:

• Variance-swap dispersion. Cleanest. Sell an index variance swap and buy a vega-weighted basket of single-name variance swaps. Covered in depth in the separate Variance Dispersion article.
• Vega-weighted straddles. Sell an at-the-money index straddle, buy ATM straddles on a representative subset (often 25 to 50 top-weighted names). Delta-hedged daily.
• Correlation swap. Go short a correlation swap directly. Cleanest vega profile, but OTC and illiquid.

Implied correlation implied by a dispersion trade is approximately:

ImpliedCorr = (sigma_index^2 - sum(w_i^2 * sigma_i^2)) / (2 * sum over i<j of w_i * w_j * sigma_i * sigma_j)

Traders monitor this number versus realized correlation over a rolling window. A wide spread invites long dispersion; a negative spread (realized above implied) signals that the setup is gone.

Worked Example

Ahead of a quarterly earnings season, a vol-arb desk computes S&P 500 1-month implied correlation at 0.22. Historical 1-month realized correlation during comparable earnings seasons has averaged 0.15. The spread looks attractive.

The desk sells $10 million vega of 1-month SPX variance and buys $10 million vega distributed across the top 50 S&P names, weighted by each stock's contribution to index vega. Deltas are hedged daily.

Over the month, earnings reactions are large and idiosyncratic: winners jump 8 percent, losers drop 10 percent, and the index moves only modestly on any single day. Realized correlation lands at 0.14.

Approximate P&L on the dispersion book, with a correlation drop of 0.08 times roughly $12 million per correlation point of sensitivity, is in the neighborhood of a $950,000 profit. The profit is noisy because both legs are individually volatile, but the correlation exposure is what drives the terminal number.

The opposite happened in March 2020. Realized correlation spiked from 0.30 toward 0.80 as every name sold off together. Long-dispersion books suffered double-digit losses in days. Many desks cut risk hard during the initial COVID decline.

Common Mistakes

1. Using too few names in the basket. Replicating index vol with a handful of constituents leaves large residual basis risk. Most practitioner papers suggest at least 30 to 50 names, and desks with real budgets use all 500.

2. Ignoring borrow and funding on short legs. Single-name options can be expensive to hold when short interest in the underlying is elevated. Dispersion books that assume flat funding understate the carry cost and overstate the edge.

3. Failing to account for skew. Index puts carry a richer skew than single-name puts, so selling index vol via straddles can underestimate the short skew exposure. Variance-swap dispersion sidesteps this problem.

4. Treating a correlation drop as a free lunch. Realized correlation can fall below 0.10 for months, then jump to 0.80 in a single session during a risk-off event. Sizing the book to survive a 2-standard-deviation correlation spike is the difference between a real vol-arb strategy and a blown account.

5. Mispricing end-of-life rebalance. As straddles approach expiry their gamma becomes concentrated and hedging costs rise. Rolling the dispersion book two to three weeks before expiry is standard practice.

Frequently Asked Questions

Q: What is a dispersion trade in simple terms? A dispersion trade bets that individual stocks will move more than the index. You sell volatility on the index and buy volatility on its components. When stocks move in different directions, the index stays calm while single-name vol pays off.

Q: How does the dispersion trade affect investment decisions? It harvests the structural gap between implied and realized correlation, index options have persistently higher implied correlation than constituent stocks realize, especially during earnings seasons. The trade is long idiosyncratic risk and short systematic risk.

Q: What is a real-world example of a dispersion trade paying off? During earnings season, a vol-arb desk observes SPX implied correlation at 0.22 versus a 5-year earnings-season average of 0.15. Winners jump 8%, losers drop 10%, but the index barely moves. Realized correlation lands at 0.14, and the spread of 0.08 produces roughly $950,000 on a $10M-vega book.

Q: How can investors manage the tail risk in dispersion trades? Size the book to survive a realized correlation spike to 0.80, what happened in March 2020. A single correlation regime change can wipe out six months of steady income in days. Most desks cap notional and use the DSPX index as a real-time monitor for implied correlation regime shifts.

Q: How is a dispersion trade different from a simple short straddle on the index? A short index straddle is short vol, short gamma, and has large directional exposure to the index. A dispersion trade nets the index short against single-name longs, removing most directional and level-vol exposure so the residual is a nearly pure correlation bet.

Sources

1. Bossu, S. A Primer on Correlation Trading via Equity Derivatives. http://docs.sbossu.com/bossu-correl_primer-20160408.pdf
2. Foresi, S. and Vesval, A. Equity Correlation Trading. Goldman Sachs Asset Management / NYU Stern. https://web-docs.stern.nyu.edu/salomon/docs/derivatives/GSAM%20-%20NYU%20conference%20042106%20-%20Correlation%20trading.pdf
3. Cboe. S&P 500 Dispersion Index (DSPX). https://www.cboe.com/us/indices/dispersion/
4. Quantpedia. Dispersion Trading. https://quantpedia.com/strategies/dispersion-trading

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.