Correlation Swap Pricing & Relative Value Arbitrage Lab

Price institutional correlation swaps and evaluate Sebastien Bossu (2005) implied correlation replication from index vs basket variance decomposition. Model the Correlation Risk Premium (CRP), systemic contagion, and mark-to-market profiles.

Market Regime Presets:
Implied Strike ($K_{\text{corr}}$)

48.2%

Bossu (2005) Replication
Correlation Risk Premium

+10.2 pts

$\rho_{\text{implied}} - \rho_{\text{realized}}$
Current MtM P&L

+$45.2k

Long Correlation Position
Diversification Ratio

1.56x

Basket Vol / Index Vol
Correlation DV01

$48,820 / pt

Discounted Notional / %
Accrual Timeline

25% Elapsed (63/252d)

Historical vs Forward
Contract & Market Parameters
%
Market variance strike $\sigma_I^2$ of the index.
%
Average implied volatility across basket components.
Equal-weighted components ($w_i = 1/n$).
%
Dispersion of single-stock vols around basket mean.
$
Dollar payoff per 1.0 correlation percentage point (0.01).
%
Historical / expected realized pairwise correlation.
%
Realized correlation observed over elapsed days.
%
Long vs Short Correlation Swap Terminal Payoff ($k) Linear Payoff: $\text{Payoff} = N_{\text{corr}} \times (\rho_R - K_{\text{corr}})$

Unlike variance swaps which have convex parabolic curves, correlation swaps are strictly linear in realized correlation points. Short correlation sellers harvest the premium in calm markets but suffer heavy drawdowns when correlations spike toward 80%-90%.

Resulting Index Volatility as a Function of Basket Correlation Bossu Decomposition: $\sigma_I = \sqrt{\bar{\sigma}^2 [w + (1-w)\rho]}$

At $\rho = 0$, index volatility drops to its lowest theoretical diversification floor $\bar{\sigma}/\sqrt{n}$. At $\rho = 1$, all diversification collapses and index volatility equals average single-stock volatility.

Implied Strike vs Expected Realized Correlation Across Tenors Correlation Risk Premium (CRP)

Implied correlation typically trades above realized correlation across maturities, reflecting the structural demand for correlation hedging by equity structured product desks (e.g. autocallables and reverse convertibles).

Basket Constituent Variance & Volatility Decomposition
Equal-Weighted Basket Representation
Constituent Basket Weight ($w_i$) Implied Vol ($\sigma_i$) Variance Contrib ($w_i^2 \sigma_i^2$) Vol vs Index ($\sigma_i / \sigma_I$) Regime Classification
Showing top 10 constituents of the basket. Variance contributions represent single-stock idiosyncratic variance components in the Bossu expansion.
Implied Strike Matrix: Index Vol vs Basket Vol
Evaluates Bossu implied correlation strike ($K_{\text{corr}} \%$)
Mark-to-Market Matrix: Accrued Corr vs Days Elapsed
Evaluates long correlation swap liquidation value ($k)

Institutional Theory & Market Mechanics

1. Sebastien Bossu (2005) Variance Decomposition

The fair implied correlation strike $K_{\text{corr}}$ is derived by decomposing the variance of a market index $I = \sum_{i=1}^n w_i S_i$ into the sum of single-stock constituent variances and cross-asset covariance terms:

$\sigma_I^2 = \sum_{i=1}^n w_i^2 \sigma_i^2 + \rho_{\text{implied}} \sum_{i \ne j} w_i w_j \sigma_i \sigma_j$

Solving for the implied correlation strike $\rho_{\text{implied}}$ yields:

$K_{\text{corr}} = \frac{\sigma_I^2 - \sum_{i=1}^n w_i^2 \sigma_i^2}{\sum_{i \ne j} w_i w_j \sigma_i \sigma_j} \approx \frac{\sigma_I^2 - \frac{1}{n} \bar{\sigma}^2}{(1 - \frac{1}{n}) \bar{\sigma}^2}$

This mathematical identity proves that an index variance swap can be dynamically synthesized by a long position in constituent variance swaps and a short position in a correlation swap.

2. The Correlation Risk Premium & Crash Asymmetry

The Correlation Risk Premium (CRP) is defined as the persistent positive difference between the implied correlation strike priced in index options and the subsequent realized correlation:

$\text{CRP} = \rho_{\text{implied}} - \rho_{\text{realized}}$

Historically, CRP averages +5% to +15% across equity indices (S&P 500, Euro Stoxx 50). This premium arises because institutional structured product issuers are structurally short correlation due to autocallables and reverse convertibles. To hedge, desks bid up index options and buy correlation swaps.

However, correlation short strategies face extreme crash asymmetry: in market panics, correlations spike toward 85%-90% ("correlations go to 1"), inflicting catastrophic losses on correlation sellers.

3. Mark-to-Market Timeline Blending

During the life of a correlation swap, its fair value is determined by blending the historical realized correlation observed over $t$ elapsed days with the prevailing implied correlation for the remaining $T - t$ days:

$\rho_{\text{blended}} = \left(\frac{t}{T}\right) \rho_{\text{accrued}} + \left(1 - \frac{t}{T}\right) \rho_{\text{forward}}$

The mark-to-market (MtM) value of a long position is discounted back to present value using the risk-free rate $r$: $\text{MtM} = N_{\text{corr}} (\rho_{\text{blended}} - K_{\text{corr}}) e^{-r (T-t)}$.

4. Pure Correlation Swaps vs. Dispersion Trades

In traditional equity dispersion trading, quantitative funds buy single-stock options and sell index options. While intuitively a short correlation trade, dispersion returns are heavily contaminated by realized single-stock volatility, gamma rebalancing, and vol-of-vol drag.

A correlation swap eliminates these contaminants by isolating pure, unadulterated correlation. It exhibits zero delta to the underlying stocks, zero vega to parallel volatility shifts, and delivers pure linear P&L on the realized co-movement of assets.

Quantitative Correlation Mastery Quiz

1. What mathematical identity defines the fair strike of an implied correlation swap?

2. What is the Correlation Risk Premium (CRP) and why is it historically positive?

3. What happens to pairwise stock correlations during systemic market panics?

4. How is a correlation swap hedged relative to an index dispersion strategy?

Frequently Asked Questions

A correlation swap is an over-the-counter financial derivative where the buyer pays a fixed fair correlation strike (K_corr) and receives the annualized average realized pairwise correlation (rho_R) of a specified stock basket: Payoff = N_corr * (rho_R - K_corr). Unlike dispersion trades, correlation swaps offer pure, linear correlation exposure unpolluted by single-stock volatility drift or gamma rebalancing.

By Sebastien Bossu (2005) and Demeterfi et al., index variance decomposes into weighted constituent variances plus cross-asset covariance terms: sigma_I^2 = sum(w_i^2 * sigma_i^2) + rho_implied * sum_{i != j}(w_i * w_j * sigma_i * sigma_j). Solving for rho_implied gives the fair implied correlation strike equating the market variance of the index to its replicating basket.

The Correlation Risk Premium (CRP) is the persistent spread where implied correlation trades 5% to 15% above expected realized correlation: CRP = rho_implied - rho_realized. It exists because index option market makers and structured product desks pay a premium to buy correlation to hedge adverse correlation risk in exotic guaranteed products and worst-of autocallables.

In severe macro sell-offs and liquidity crunches, market-wide margin calls and index-level ETF liquidations dominate idiosyncratic company fundamentals. As investors dump broad market baskets simultaneously, stock co-movement surges, causing pairwise correlations across diverse sectors to converge toward 80%-90% ('all correlations go to 1').

During the lifespan of a correlation swap, mark-to-market (MtM) pricing blends the realized pairwise correlation accrued over elapsed trading days with prevailing implied forward correlation for remaining days: rho_blended = (t_elapsed / T) * rho_accrued + ((T - t_elapsed) / T) * rho_forward. The discounted difference against the original strike K_corr determines current liquidation value.
Related Volatility & Derivatives Simulators:
Corridor Variance Swap Lab Volatility Swap Lab Variance Swap Lab Dispersion Trading Lab CBOE VIX Index Lab Dupire Local Vol Lab