Correlation, Volatility-of-Volatility, and Sector Implied Volatility

How Correlation and Volatility Affect Portfolio Risk

Author

Nam Nguyen
September 20, 2026 • Estimated Reading Time: 5 minutes

Correlation is an important component of portfolio and risk management. However, unlike volatility, which has received significant attention and for which numerous models have been developed, correlations have received considerably less attention from a modeling perspective.

In this edition, we give correlations the attention they deserve and examine their role in volatility dynamics, portfolio construction, and trading.

In this issue:

Latest Posts

  • Reflexivity and the Dynamics of Option Markets (10 min)

  • The Effects of ETF Investing on Financial Markets (10 min)

  • Market Regimes and Changing Market Dynamics (10 min)

  • Making Option Pricing Models More Practical (10 min)

  • Algorithmic Trading, HFT, and Market Stability (10 min)

State-Dependent Correlation Between the S&P 500 and the VIX

It is well known that the correlation between the S&P 500 and the VIX index is usually negative. However, practitioners also recognize that this correlation can break down and even turn positive.

Reference [1] addresses this issue by examining the time-varying nature of the correlation. The paper first posits and demonstrates that volatility-of-volatility (VoV) risk is distinct from volatility (Vol) risk, and then classifies the market into four regimes:

1. Low VOL risk and low VOV risk (LVOL–LVOV)

2. High VOL risk and low VOV risk (HVOL–LVOV)

3. Low VOL risk and high VOV risk (LVOL–HVOV)

4. High VOL risk and high VOV risk (HVOL–HVOV).

Findings

  • The authors distinguish between two types of uncertainty: volatility (VOL) risk in the S&P 500 and volatility-of-volatility (VOV) risk in the VIX.

  • They propose a theoretical framework in which the correlation between the S&P 500 and VIX varies across different combinations of VOL and VOV risk states.

  • The authors develop a state-dependent regime-switching DCC model that identifies four states: high/low VOL risk combined with high/low VOV risk.

  • The empirical results indicate that shocks driving VOL risk are not necessarily the same as those driving VOV risk, and the two risks respond differently to market shocks.

  • The strongest negative correlation between the S&P 500 and VIX occurs in the high-VOL/high-VOV state, while the weakest negative correlation occurs in the high-VOL/low-VOV state.

  • The authors find that their state-dependent correlation model provides greater portfolio risk reduction than conventional time-dependent correlation models.

In short, the paper develops a regime-switching model to show that the relationship between the S&P 500 and the VIX varies across distinct volatility (VOL) and volatility-of-volatility (VOV) risk states. It identifies a four-state system and demonstrates that correlation strength depends on the joint VOL–VOV regime.

This is an important contribution, as it quantifies the dynamic relationship between the S&P 500 and the VIX, thereby providing useful guidance for refining hedging and arbitrage strategies.

Reference

Predicting Volatility Risk Premium Through Sector Implied Correlation

Despite the frequent use of the volatility risk premium (VRP), it appears, however, that there exists no reliable method for forecasting the VRP. In the literature, we have seen methods ranging from a simple one such as IV-HV (historical volatility) to a more sophisticated method such as GARCH.

Article [2] examines the volatility risk premium through implied volatilities and the correlation of sector ETFs. It concludes, surprisingly, that correlation premium is a good predictor of the volatility risk premium.

Findings

  • The authors examine options on sector ETFs within the S&P 500 and find predictable variation in sector implied volatilities associated with unusually high or low implied correlations.

  • They argue that abnormal short-term option pricing can be identified more effectively through implied correlation than through implied volatility alone.

  • The authors develop a sector-specific correlation premium that adjusts for aggregate correlation levels and isolates the idiosyncratic component of sector correlation.

  • The sector-specific correlation premium predicts future changes in sector implied volatility more reliably than conventional volatility- or correlation-premium measures.

  • One-day reversals in sector implied volatility are associated with reversals in the sector-specific correlation premium.

  • The authors find that information extracted from individual-stock implied volatilities has little or no predictive ability for sector implied volatility.

  • Trading signals based on the abnormal sector-specific correlation premium are profitable and outperform conventional strategies based on sector volatility premiums.

Reference

[2] Koticha, Apoorva and Li, Chen and Marks, Joseph M., Abnormal Sector Option Correlation Premiums and Predictable Changes in Implied Volatility, SSRN 3862777

Closing Thoughts

Together, these studies highlight the importance of correlations in understanding volatility dynamics and improving investment decisions. The first shows that correlations between the S&P 500 and VIX vary across different volatility and volatility-of-volatility regimes, with state-dependent correlations improving portfolio risk management. The second demonstrates that implied correlations contain information about future sector implied volatility beyond conventional volatility measures and can form the basis for profitable trading signals.

Overall, both studies suggest that explicitly modeling correlations can provide valuable information for volatility forecasting, portfolio construction, and trading.

Additional Reading

For further discussion on correlations, refer to the previous issues:

Educational Video

Correlation, DCC-GARCH model, copulas, market networks

In this video, Dr. Kiss Gábor Dávid discusses several approaches to modeling correlation, beginning with classical correlation and its limitations. A major weakness of conventional correlation is that it summarizes dependence over an entire sample and therefore ignores changes through time. He discusses exponentially weighted moving averages as one way to give greater importance to recent observations, before turning to Dynamic Conditional Correlation (DCC) models, which allow the variance-covariance matrix and correlations to evolve over time. Using MATLAB examples, he demonstrates how DCC can be applied to asset returns and emphasizes that the estimated correlations can depend significantly on data frequency and model specification. 

He then extends the discussion beyond linear correlation to copulas, which allow dependence to be modeled through joint probability distributions and can accommodate different marginal distributions and forms of tail dependence. However, he highlights an important trade-off: standard copula implementations can capture non-normal dependence structures but may assume correlations that are constant through time, potentially missing changing dependence during periods of stress. He also demonstrates how correlation structures can be represented through networks to study the propagation of shocks and discusses their application to credit-risk and default modeling. Overall, the lecture illustrates why correlation modeling requires more than a single historical correlation estimate, particularly when dependence changes over time or becomes stronger during extreme events. 

Around the Quantosphere

  • How India Built (and Broke) the World’s Biggest Options Casino (bloomberg)

  • On the Cutting Hedge: Generating Alpha While Managing Risk in the Current Regime (nb)

  • How AI Is Reshaping Quantitative Finance (wharton)

  • Mark Carney wants the world's biggest funds to invest billions in Canada. Will they buy in?? (financialpost)

  • How AI Tools Are Reshaping Stock Trading, Investment Research in China (scmp)

  • Hedge fund QRT interviewed heavily for its discretionary macro team. Now it's made a decision (efinancialcareers)

  • Bridgewater's co-CIO warned that AI will kill people before it gets regulated (qz)

  • 2026 Hedge Fund Allocation Survey (bnpparibas)

  • How AI and the hedge fund talent war pushed Man Group to reimagine its $156 billion quant empire (businessinsider)

  • The Growth and Performance of AI-Driven Investing (dealroom)

  • How one hedge-fund manager built his firm to be powered entirely by AI agents (cnbc)

  • How to make $32bn in hedge funds by letting people do what they like. The hot hand of Greece (efinancialcareers)

  • The wittiest quant describes how AI is changing his job. Iliad wrestling parties are a thing (efinancialcareers)

Disclaimer

This newsletter is not investment advice. It is provided solely for entertainment and educational purposes. Always consult a financial professional before making any investment decisions.

We are not responsible for any outcomes arising from the use of the content and codes provided in the outbound links. By continuing to read this newsletter, you acknowledge and agree to this disclaimer.