Paper: arXiv 2406.07388

Authors: Markus Bibinger

Abstract

The scope of this manuscript is to review some recent developments in statistics for discretely observed semimartingales which are motivated by applications for financial markets. Our journey through this area stops to take closer looks at a few selected topics discussing recent literature. We moreover highlight and explain the important role played by some classical concepts of probability and statistics. We focus on three main aspects: Testing for jumps; rough fractional stochastic volatility; and limit order microstructure noise. We review jump tests based on extreme value theory and complement the literature proposing new statistical methods. They are based on asymptotic theory of order statistics and the Rényi representation. The second stage of our journey visits a recent strand of research showing that volatility is rough. We further investigate this and establish a minimax lower bound exploring frontiers to what extent the regularity of latent volatility can be recovered in a more general framework. Finally, we discuss a stochastic boundary model with one-sided microstructure noise for high-frequency limit order prices and its probabilistic and statistical foundation.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 2.5/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper presents advanced theoretical probability and statistics, focusing on semimartingales, minimax lower bounds, and asymptotic theory of order statistics, but is a review with no code, datasets, or backtests.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Statistical Methods for<br>Discrete Semimartingales"] --> B{"Methodology"}
  
  B --> C["Testing for Jumps<br>Extreme Value Theory &<br>Rényi Representation"]
  B --> D["Rough Stochastic Volatility<br>Minimax Lower Bounds &<br>Regularity Recovery"]
  B --> E["Microstructure Noise<br>Stochastic Boundary Model<br>for Limit Orders"]
  
  C --> F["Data: High-Frequency<br>Asset Returns"]
  D --> G["Data: Volatility Estimators<br>& Latent Processes"]
  E --> H["Data: Limit Order Prices<br>& Trade Data"]
  
  F --> I["Computational<br>Asymptotic Order<br>Statistics Analysis"]
  G --> I
  H --> I
  
  I --> J["Key Findings<br>Enhanced Jump Detection<br>Volatility Regularity Bounds<br>Microstructure Noise Model"]

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 new statistical jump‑detection methods based on extreme‑value theory and Rényi’s representation, establishes a minimax lower bound for recovering the regularity of latent volatility, and introduces a stochastic boundary model with one‑sided microstructure noise for high‑frequency

Source sentence
We review jump tests based on extreme value theory and complement the literature proposing new statistical methods. They are based on asymptotic theory of order statistics and the Rényi representation.

Assumes 5

  • The drift µt and volatility σt processes satisfy regularity conditions that can be generalized to weak assumptions. model
    Source sentence
    The proof is carried out under some assumptions on (µt) and (σt), which can be generalized to rather weak regularity conditions.
  • Under the null hypothesis of no jumps, the normalized maximal absolute return increments follow an asymptotic Gumbel distribution. statistical
    Source sentence
    On the null hypothesis H0 : supτ∈(0,1) |Xτ −Xτ−| = 0, based on high-frequency returns (∆n j X)1≤j≤n, Lee and Mykland (2008) proved in their Thm. 2 that q 2 log(2n)  n1/2 max 1≤j≤n ∆n j X ˆσj/n − q 2 log(2n) −log(4π log(2n)) 2 p 2 log(2n)  d −→Λ, (17) with a suitable estimator of the volatility (ˆσj/n), e.g., from (13).
  • The sub‑model used for the lower‑bound proof does not possess the dependence structure of fractional Brownian motion. model
    Source sentence
    The proof of the lower bound utilizes a sub-model which does not have the dependence structure of fBm.
  • The volatility process log(σt) can be modeled as a fractional Brownian motion or a more general fractional process. model
    Source sentence
    If (log(σt)) was a fBM, or as well if it was a more general fractional process, we expect to find q · H as the slope in these regressions.
Show 1 more
  • High‑frequency log‑prices are discretizations of continuous‑time stochastic processes, while limit‑order book data contain multiple price levels at each time point. market
    Source sentence
    5 Limit order microstructure noise While we modelled high-frequency log-prices so far as discretizations of continuous-time stochastic processes, when having available data from a limit order book, there is not only one price at some given time.

Data

  • S&P 500 ETF volatility estimates (daily, 7021 observations) · Risk Lab (Dacheng Xiu) / Oxford‑Man Institute market-prices public
    1996‑2023 · S&P 500 market ETF
    Source sentence
    We replicate the same behav-ior of statistics m(q, ∆) as in Figures 5-7 of Gatheral et al. (2018) for a time series of 7021 quasi maximum likelihood daily volatility estimates from 1996 to 2023, based on the method by Xiu (2010), inferred from intra-day high-frequency trade prices of the S&P 500 market ETF. The data is constructed from the Risk Lab on Dacheng Xiu’s website 4 and the S&P 500 is certainly a very
  • Nasdaq limit order book data for Apple · LOBSTER order-book public
    Apple (AAPL) asset
    Source sentence
    We use Nasdaq data from Lobster.

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • The Lee‑Mykland jump test statistic, based on the maximal normalized increment, converges to a Gumbel distribution under the null of no jumps. theoretical
    Source sentence
    It is based on the maximal (absolute) normalized increment and exploits its asymptotic Gumbel distribution under the null hypothesis of no jumps. On the null hypothesis H0 : supτ∈(0,1) |Xτ −Xτ−| = 0, based on high-frequency returns (∆n j X)1≤j≤n, Lee and Mykland (2008) proved in their Thm. 2 that q 2 log(2n)  n1/2 max 1≤j≤n ∆n j X ˆσj/n − q 2 log(2n) −log(4π log(2n)) 2 p 2 log(2n)  d −→Λ, (17) with a suitable
  • A minimax lower bound of order n^{‑1/2+2α} for α<1/4 shows that a consistent estimator of the volatility regularity exists only for α₀<1/4. theoretical
    Source sentence
    In particular, we obtain n−1/2+2α, for α < 1/4, as a lower bound. This shows that a consistent estimator only exists for α0 < 1/4!
  • The proposed estimator ˆσ²_{ad} attains the same optimal convergence rate as an estimator that knows the regularity parameter α. theoretical
    Source sentence
    We conclude that ˆσ2,ad s attains the same optimal rate of convergence as the estimator which exploits known α.
  • Empirical analysis of the S&P 500 volatility series yields an estimated Hurst exponent ˆH≈0.16, supporting the rough volatility paradigm. statistical figure from elsewhere in the paper
    Source sentence
    From this illustration, we see the estimate ˆH ≈0.16 for this data.
  • A Gumbel‑type test for jumps under microstructure noise is derived, with the statistic n^{1/3}T_{BHR} converging to a standard Gumbel distribution. theoretical
    Source sentence
    It is based on T BHR = max k=1,...,h−1 n −1 mk,n −mk−1,n �ˆσ2 khn 1/2 . Based on extreme value theory, we show that under the null hypothesis H0 : sup τ∈[0,1] |∆Xτ| = 0 and for (σt) ∈Cα, i.e., Hölder continuous with regularity α, it holds with hn = 2 log(2h−1 n −2)n−2/3 that n1/3 T BHR −2 log(2h−1 n −2) + log �π log(2h−1 n −2)  d −→Λ , (32) with Λ the standard Gumbel distribution.

Checked by others

Supported by

  • Roughness Analysis of Realized Volatility and VIX through Randomized Kolmogorov-Smirnov Distribution (2025)
    On the question: What is the estimated Hurst exponent of the S&P 500 volatility series?
    Compare the findings

    This paper: Empirical analysis of the S&P 500 volatility series yields an estimated Hurst exponent ˆH≈0.16, supporting the rough volatility paradigm.

    That paper: The RV5 log series yields an average Hurst exponent of 0.1379 (variance 0.0085) with a highly significant p‑value.

    B supports A: A gives H≈0.16, B gives H≈0.1379 (variance 0.0085) with a highly significant p‑value; values are close, confirming rough volatility.

  • Rough volatility: evidence from range volatility estimators (2023)
    On the question: What is the Hurst exponent of range‑based volatility proxies for financial markets?
    Compare the findings

    This paper: Empirical analysis of the S&P 500 volatility series yields an estimated Hurst exponent ˆH≈0.16, supporting the rough volatility paradigm.

    That paper: The Hurst exponent for the range‑based proxies can be as low as 0.014, supporting the rough volatility hypothesis.

    Both findings report low H values (0.014 and 0.16) supporting rough volatility, differing in magnitude and sample (range‑based proxies vs S&P 500 series).

  • Algorithmic Market Making Can Lead to Rough Volatility Through the Order Book (2026)
    On the question: Do order‑flow and price volatility exhibit rough volatility, as indicated by a low Hurst exponent?
    Compare the findings

    This paper: Empirical analysis of the S&P 500 volatility series yields an estimated Hurst exponent ˆH≈0.16, supporting the rough volatility paradigm.

    That paper: Empirical analysis on six Binance markets shows order‑flow and volatility Hurst exponents between 0.07 and 0.09, within 0.02 of each other.

    A finds H≈0.16 for S&P 500; B finds H=0.07‑0.09 for Binance order‑flow and volatility, a smaller but still low value, both supporting rough volatility.

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.