Paper: arXiv 2505.02678
Authors: Othmane Zarhali, Cecilia Aubrun, Emmanuel Bacry, Jean-Philippe Bouchaud, Jean-François Muzy
Abstract
The Nested factor model was introduced by Chicheportiche et al. to represent non-linear correlations between stocks. Stock returns are explained by a standard factor model, but the (log)-volatilities of factors and residuals are themselves decomposed into factor modes, with a common dominant volatility mode affecting both market and sector factors but also residuals. Here, we consider the case of a single factor where the only dominant log-volatility mode is rough, with a Hurst exponent \(H \simeq 0.11\) and the log-volatility residuals are ‘‘super-rough’’ or ‘‘multifractal’’, with \(H \simeq 0\). We demonstrate that such a construction naturally accounts for the somewhat surprising stylized fact reported by Wu et al. , where it has been observed that the Hurst exponents of stock indexes are large compared to those of individual stocks. We propose a statistical procedure to estimate the Hurst factor exponent from the stock returns dynamics together with theoretical guarantees of its consistency. We demonstrate the effectiveness of our approach through numerical experiments and apply it to daily stock data from the S&P500 index. The estimated roughness exponents for both the factor and idiosyncratic components validate the assumptions underlying our model.
Complexity vs Empirical Score
- Math Complexity: 8.0/10
- Empirical Rigor: 6.5/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper presents advanced stochastic calculus and fractional Brownian motion theory with extensive mathematical derivations and theoretical proofs, indicating high math complexity. It validates its model with numerical experiments, statistical estimation procedures with theoretical guarantees, and empirical analysis on S&P500 data, demonstrating substantial empirical rigor.
Research Flowchart
flowchart TD Start["Research Goal: Why are single stock volatilities<br>rougher than S&P500 index volatilities?"] Method["Methodology: Nested Factor Model<br>with Rough Volatility"] --> Data["Data: S&P500 Daily Stock Returns"] Data --> CP["Computational Process:<br>1. Estimate Hurst exponent H<br>2. Analyze Factor vs Residual roughness"] CP --> Finding1["Key Finding 1: Factor H ≈ 0.11<br>(Rough, dominant mode)"] CP --> Finding2["Key Finding 2: Residual H ≈ 0<br>(Super-rough/Multifractal)"] Finding1 & Finding2 --> Conclusion["Outcome: Model explains why<br>index vol > single stock vol"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a statistical procedure to estimate the Hurst factor exponent and introduces the Nested log S‑fBM factor model (N‑SfFM) to explain rough volatility of stocks and indexes.
Source sentence
We propose a statistical procedure to estimate the Hurst factor exponent from the stock returns dynamics together with theoretical guarantees of its consistency.
Assumes 5
- Log‑volatility follows a Gaussian process. statistical
Source sentence
The GMM operates under the assumption that log‑volatility follows a Gaussian process.
- A Gaussianisation pre‑processing step corrects skewness in log‑volatility. model
Source sentence
we undertake a “Gaussianisation” procedure as a pre‑processing step in order to rectify the empirically determined log‑quadratic variations that are distorted by measurement noise.
- The dominant log‑volatility mode of the factor has H≈0.11 and residuals have H≈0. model
Source sentence
the only dominant log‑volatility mode is rough, with a Hurst exponent H ≃0.11 and the log‑volatility residuals are “super‑rough” or “multifractal”, with H ≃0.
- Daily Garman‑Klass estimator provides a reliable volatility measure for the factor. data
Source sentence
daily volatility estimator based on high‑frequency (intraday) data, such as the Garman‑Klass estimator
Show 1 more
- The sample contains a sufficiently large number of stocks to apply the GMM estimation. data
Source sentence
This approach is contingent upon having a sufficiently large number of stocks in the sample, as illustrated in our empirical study in Section 7.
Data
- Daily stock data from the S&P500 market-prices public
Source sentence
The calibration is applied to empirical daily data from YahooFinance spanning from 01/01/2015 to 21/03/2025 for 450 US stocks that are constituents of the S&P500 throughout the entire period.
- Garman‑Klass volatility estimator market-prices public
Source sentence
Figure 1: Hurst exponents estimation of 503 stocks and 49 indexes using the calibration method of Wu et al. [25] on daily Garman‑Klass estimation of the log‑volatility from 2013 to 2023.
Costs
Not applicable
Finds 4
- The factor log‑volatility exhibits a roughness exponent around H≈0.11 while idiosyncratic residuals are essentially super‑rough with H≈0. theoretical
Source sentence
the only dominant log‑volatility mode is rough, with a Hurst exponent H ≃0.11 and the log‑volatility residuals are “super‑rough” or “multifractal”, with H ≃0.
- Empirical estimates show most individual stock Hurst exponents are very close to zero, confirming the model’s assumptions. statistical
Source sentence
Note that most of the Hi are very close to zero.
- Stock indexes have larger Hurst exponents (≈0.1) than individual stocks, reproducing the puzzling stylized fact. statistical
Source sentence
While the roughness index of stock indexes is compatible with the values reported by Gatheral et al. (H ≈0.1), individual stocks exhibit significantly greater roughness (H ≈0) as illustrated in Figure 1.
- The calibrated N‑SfFM reproduces the empirical findings of Wu et al., matching the observed Hurst exponent patterns for both factors and idiosyncratic components. performance
Source sentence
The estimated roughness exponents for both the factor and idiosyncratic components validate the assumptions underlying our model.
Checked by others
Supported by
- A unified theory of order flow, market impact, and volatility
Compare the findings
This paper: The factor log‑volatility exhibits a roughness exponent around H≈0.11 while idiosyncratic residuals are essentially super‑rough with H≈0.
That paper: Volatility shows extreme roughness with Hurst parameters between 0.05 and 0.15, aligning with the model’s prediction for H = 2H₀‑3/2.
- Roughness Analysis of Realized Volatility and VIX through Randomized Kolmogorov-Smirnov Distribution
Compare the findings
This paper: The factor log‑volatility exhibits a roughness exponent around H≈0.11 while idiosyncratic residuals are essentially super‑rough with H≈0.
That paper: The RV5 log series yields an average Hurst exponent of 0.1379 (variance 0.0085) with a highly significant p‑value.
2 papers in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (47)
- Individual and collective stock dynamics: intra-day seasonalities (2011)
- The variation of certain speculative prices (1967)
- Fractional Brownian motion with zero Hurst parameter: a rough volatility viewpoint (2018)
- A new spin on optimal portfolios and ecological equilibria (2021)
- Rough volatility (2023)
- Correlation structure of extreme stock returns (2001)
- Symmetry alteration of ensemble return distribution in crash and rally days of financial markets (2000)
- Decoupling the short-and long-term behavior of stochastic volatility (2022)
and 39 more
- Pricing under rough volatility (2016)
- Volatility has to be rough (2021)
- Rough volatility: evidence from option prices (2018)
- Roughness in spot variance? A GMM approach for estimation of fractional log-normal stochastic volatility models using realized measures (2023)
- Is volatility rough? (2019) — link
- Turbulent cascades in foreign exchange markets (1996)
- A multifractal model of asset returns (1997)
- The Riemann–Liouville field and its GMC as H→ 0, and skew flattening for the rough Bergomi model (2022)
- The Rough Bergomi model as H→ 0–skew flattening/blow up and non-Gaussian rough volatility (2020)
- Fractional Brownian motion with Hurst index H=0 and the Gaussian unitary ensemble (2016)
- The multiplicative chaos of H= 0 fractional Brownian fields (2022)
- Log-infinitely divisible multifractal processes (2003)
- Multifractal stationary random measures and multifractal random walks with log infinitely divisible scaling laws (2002)
- Multifractal random walk (2001)
- The Large Sample Behaviour of the Generalized Method of Moments Estimator in Misspecified Models (2003)
- Generalized Method of Moments (2005)
- Measuring the Sensitivity of Parameter Estimates to Estimation Moments (2017) — link
- Volatility is rough (2018)
- Sensitivity analysis using approximate moment condition models (2021)
- From rough to multifractal volatility: The log S-fBM model (2022)
- A nested factor model for non-linear dependencies in stock returns (2015)
- The evaluation of the collision matrix (1950) — link
- Random cascade model in the limit of infinite integral scale as the exponential of a nonstationary $1/f$ noise. Application to volatility fluctuations in stock markets (2013) — link
- The joint distribution of stock returns is not elliptical (2012)
- Modelling fluctuations of financial time series: from cascade process to stochastic volatility model (2000)
- Modelling financial time series using multifractal random walks (2001)
- Long memory in continuous-time stochastic volatility models (1998)
- Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk (1964)
- The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets (1965)
- Common Risk Factors in the Returns on Stocks and Bonds (1993)
- On Persistence in Mutual Fund Performance (1997)
- A Five-Factor Asset Pricing Model (2015)
- Risk and Return in an Equilibrium APT: Application of a New Test Methodology (1988)
- Determining the Number of Factors in Approximate Factor Models (2002)
- …and the Cross-Section of Expected Returns (2016)
- Log-Normal continuous cascades: aggregation properties and estimation. Application to financial time-series (2008) — link
- Rough volatility: evidence from range volatility estimators — → 2312.01426
- On the Estimation of Security Price Volatility from Historical Data (1980)
- Th 'e orie de la sp 'e culation (1900)
1 of 47 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.