Paper: arXiv 2505.08654

Authors: Ziyang Xiong, Zhao Chen, Christina Dan Wang

Abstract

Estimating the leverage effect from high-frequency data is vital but challenged by complex, dependent microstructure noise, often exhibiting non-Gaussian higher-order moments. This paper introduces a novel multi-scale framework for efficient and robust leverage effect estimation under such flexible noise structures. We develop two new estimators, the Subsampling-and-Averaging Leverage Effect (SALE) and the Multi-Scale Leverage Effect (MSLE), which adapt subsampling and multi-scale approaches holistically using a unique shifted window technique. This design simplifies the multi-scale estimation procedure and enhances noise robustness without requiring the pre-averaging approach. We establish central limit theorems and stable convergence, with MSLE achieving convergence rates of an optimal \(n^{"-1/4"}\) and a near-optimal \(n^{"-1/9"}\) for the noise-free and noisy settings, respectively. A cornerstone of our framework’s efficiency is a specifically designed MSLE weighting strategy that leverages covariance structures across scales. This significantly reduces asymptotic variance and, critically, yields substantially smaller finite-sample errors than existing methods under both noise-free and realistic noisy settings. Extensive simulations and empirical analyses confirm the superior efficiency, robustness, and practical advantages of our approach.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper is mathematically dense, featuring advanced theoretical derivations, central limit theorems, and convergence rate analyses, warranting a high math score. It also demonstrates high empirical rigor through extensive Monte Carlo simulations, validation of theoretical properties, and real-world high-frequency data analysis on multiple assets.

Research Flowchart

  flowchart TD
  A["Research Goal: Efficient Leverage Effect Estimation<br>under Dependent Microstructure Noise"] --> B["Methodology: Multi-scale Framework<br>with Shifted Windows & Holistic Subsampling"]
  
  subgraph C ["Data & Inputs"]
      direction LR
      C1["High-Frequency Equity Data"]
      C2["Dependent &<br>Non-Gaussian Noise"]
  end
  
  B --> C
  C --> D["Computational Process:<br>MSLE & SALE Estimators<br>Optimal Weighting Strategy"]
  
  D --> E["Outcomes"]
  
  subgraph E ["Key Findings"]
      direction TB
      E1["Convergence Rates:<br>Optimal n^{'-1/4} (noise-free)<br>Near-optimal n^{'-1/9} (noisy)"]
      E2["Superior Efficiency & Robustness:<br>Lower asymptotic variance &<br>smaller finite-sample errors vs. existing methods"]
      E3["Simplified Multi-scale Estimation:<br>No pre-averaging needed"]
  end

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 two novel estimators for the leverage effect: the Subsampling-and-Averaging Leverage Effect (SALE) and the Multi-Scale Leverage Effect (MSLE).

Source sentence
We propose two novel estimators: the Subsampling-and-Averaging Leverage Effect (SALE) and the Multi-Scale Leverage Effect (MSLE).

Assumes 4

  • Microstructure noise is stationary, dependent, and may have nontrivial higher-order moments. statistical
    Source sentence
    allowing for stationary, dependent noise with nontrivial higher-order moments, which are commonly observed in empirical financial data.
  • High‑frequency price observations are contaminated by microstructure noise. data
    Source sentence
    estimating the leverage effect from high-frequency data contaminated by dependent, non-Gaussian microstructure noise.
  • The shifted‑window technique yields a noise‑unbiased base estimator. model
    Source sentence
    Central to our approach is a shifted window technique that constructs a noise-unbiased base estimator, significantly simplifying the multi-scale architecture.
  • The noise process may exhibit serial dependence and higher‑order moments, violating i.i.d. Gaussian assumptions. statistical
    Source sentence
    empirical evidence has consistently shown that microstructure noise exhibits serial dependence and higher-order moments

Data

  • TAQ high‑frequency trading data · TAQ database market-prices commercial
    2014 to 2023 · 30 assets (15 ETFs and 15 individual stocks)
    Source sentence
    The high-frequency trading data for a selection of assets, covering the regular trading hours from 2014 to 2023 (2,516 trading days), are collected from the TAQ database.

Recognised sources: NYSE TAQ

Costs

Not applicable The paper reports no trading or execution result.

Finds 3

  • The MSLE estimator consistently and substantially outperforms the pre‑averaging estimator in finite‑sample RMSE and efficiency across all sample sizes and noise levels. performance
    Source sentence
    The findings confirm that the MSLE estimator consistently and substantially outperforms the pre-averaging estimator in terms of finite-sample RMSE and efficiency across all sample sizes and noise levels.
  • In empirical analysis, a negative leverage effect is predominant across most assets, with exceptions for meme stocks. statistical
    Source sentence
    As summarized in Table 5 and visualized in Figure 8, a negative leverage effect is predominant across most assets, particularly within established large-cap and defensive stocks, consistent with financial theory.
  • The proposed weighting strategy yields substantial efficiency gains over existing benchmarks. theoretical
    Source sentence
    By optimizing the weights based on the asymptotic covariance structure across scales and incorporating finite-sample variance corrections, we achieve substantial efficiency gains over existing benchmarks.

Checked by others

Qualified by

  • International Financial Markets Through 150 Years: Evaluating Stylized Facts (2025)
    On the question: Is a leverage effect present in asset returns?
    Compare the findings

    This paper: In empirical analysis, a negative leverage effect is predominant across most assets, with exceptions for meme stocks.

    That paper: Mixed evidence is observed for the leverage effect, which is not uniformly immediate across all assets.

    A shows mixed evidence over 150 years, not uniformly immediate; B finds a negative leverage effect in most assets, except meme stocks.

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 (59)

  1. How Often to Sample a Continuous-Time Process in the Presence of Market Microstructure Noise (2005) — link
  2. Testing for Jumps in a Discretely Observed Process (2009) — link
  3. Ultra High Frequency Volatility Estimation with Dependent Microstructure Noise (2011) — link
  4. Testing for Jumps in Noisy High Frequency Data (2012) — link
  5. The Leverage Effect Puzzle: Disentangling Sources of Bias at High Frequency (2013) — link
  6. High- Frequency Financial Econometrics (2014) — link
  7. Estimation of the Continuous and Discontinuous Leverage Effects (2017) — link
  8. A Hausman Test for the Presence of Market Microstructure Noise in High Frequency Data (2019) — link
and 51 more
  1. Great Realizations (2000)
  2. Volatility Forecasting : Downside Risk , Jumps and Leverage Effect (2016) — link
  3. Jumps in High-Frequency Data : Spurious Detections , Dynamics , and News (2016) — link
  4. Separating Microstructure Noise from Volatility (2006) — link
  5. Microstructure Noise , Realized Variance , and Optimal Sampling (2008) — link
  6. Time-Varying Leverage Effects (2012) — link
  7. Realized Kernels in Practice: Trades and Quotes (2009) — link
  8. Stochastic Volatility, Jumps and Leverage in Energy and Stock Markets: Evidence from High Frequency Data (2021) — link
  9. Estimation of the Discontinuous Leverage Effect: Evidence from the NASDAQ Order Book (2019) — link
  10. Studies of Stock Price Volatility Changes (1976)
  11. Leverage and Volatility Feedback Effects in High-Frequency Data (2006) — link
  12. The Leverage Effect Puzzle under Semi-nonparametric Stochastic Volatility Models (2024) — link
  13. The Volatility of a Firm’s Assets and the Leverage Effect (2016) — link
  14. Volatility of Volatility and Leverage Effect from Options — → 2305.04137
  15. When Frictions Are Fractional : Rough Noise in High-Frequency Data (2025) — link
  16. Fact or Friction: Jumps at Ultra High Frequency (2014) — link
  17. The Stochastic Behavior of Common Stock Variances Value , Leverage and Interest Rate Effects (1982) — link
  18. Disentangling Sources of High Frequency Market Microstructure Noise (2021) — link
  19. Estimation for High-Frequency Data under Parametric Market Microstructure Noise (2021) — link
  20. Stochastic Leverage Effect in High-Frequency Data: A Fourier Based Analysis (2022) — link
  21. When Moving - Average Models Meet High - Frequency Data : Uniform Inference on Volatility (2021) — link
  22. Realized Variance and Market Microstructure Noise (2006) — link
  23. A Closed-Form Solution for Options with Stochastic Volatility with Applications to Bond and Currency Options (1993) — link
  24. On Continuous Conditional Gaussian Martingales and Stable Convergence in Law (1997) — link
  25. Limit Theorems for Stochastic Processes (2003)
  26. Microstructure Noise in the Continuous Case: The Pre-Averaging Approach (2009) — link
  27. Discretization of Processes (2012) — link
  28. Statistical Properties of Microstructure Noise (2017) — link
  29. Estimating Quadratic Variation Consistently in the Presence of Endogenous and Diurnal Measurement Error (2008) — link
  30. Subsampling High Frequency Data (2011) — link
  31. Nonparametric Estimation of the Leverage Effect : A Trade-Off Between Robustness and Efficiency (2017) — link
  32. Fundamental Theories and Their Applications of the Calculus of Variations (2021) — link
  33. Elements of Large-Sample Theory (1999)
  34. Efficient Estimation of Integrated Volatility Incorporating Trading Information (2016) — link
  35. Dependent Microstructure Noise and Integrated Volatility Estimation from High-Frequency Data (2020) — link
  36. A ReMeDI for Microstructure Noise (2022) — link
  37. The Asymmetric High-Frequency Volatility Transmission across International Stock Markets (2019) — link
  38. Inference for Continuous Semimartingales Observed at High Frequency (2009) — link
  39. The Econometrics of High-Frequency Data (2012) — link
  40. Between Data Cleaning and Inference: Pre-averaging and Robust Estimators of the Efficient Price (2016) — link
  41. Data-Based Ranking of Realised Volatility Estimators (2011) — link
  42. Bipower-Type Estimation in a Noisy Diffusion Setting (2009) — link
  43. Understanding Limit Theorems for Semimartingales: A Short Survey (2010) — link
  44. Estimation of Integrated Volatility of Volatility with Applications to Goodness-of-Fit Testing (2015) — link
  45. The Estimation of Leverage Effect With High-Frequency Data (2014) — link
  46. Quasi-Maximum Likelihood Estimation of Volatility with High Frequency Data (2010) — link
  47. Estimation of Leverage Effect : Kernel Function and Efficiency (2023) — link
  48. Leverage Effect in High-Frequency Data with Market Microstructure (2020) — link
  49. A Tale of Two Time Scales : Determining Integrated Volatility With Noisy High-Frequency Data (2005) — link
  50. Efficient Estimation of Stochastic Volatility Using Noisy Observations: A Multi-Scale Approach (2006) — link
  51. High- Frequency Data and Volatility in Foreign-Exchange Rates (1996) — link

1 of 59 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.