Paper: arXiv 2412.11602

Authors: Anton J. Heckens, Efstratios Manolakis, Cedric Schuhmann, Thomas Guhr

Abstract

Multivariate Distributions are needed to capture the correlation structure of complex systems. In previous works, we developed a Random Matrix Model for such correlated multivariate joint probability density functions that accounts for the non-stationarity typically found in complex systems. Here, we apply these results to the returns measured in correlated stock markets. Only the knowledge of the multivariate return distributions allows for a full-fledged risk assessment. We analyze intraday data of 479 US stocks included in the S&P500 index during the trading year of 2014. We focus particularly on the tails which are algebraic and heavy. The non-stationary fluctuations of the correlations make the tails heavier. With the few-parameter formulae of our Random Matrix Model we can describe and quantify how the empirical distributions change for varying time resolution and in the presence of non-stationarity.

Complexity vs Empirical Score

  • Math Complexity: 8.0/10
  • Empirical Rigor: 7.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper uses advanced Random Matrix Theory and multivariate distributions with heavy algebraic tails, demonstrating high mathematical density. It also shows strong empirical rigor by analyzing high-frequency intraday data from 479 stocks, implementing specific data cleaning and normalization procedures, and validating model predictions against observed non-stationarity effects.

Research Flowchart

  flowchart TD
  A["Research Goal: Quantify impact of non-stationarity<br>on multivariate stock return distributions"] --> B["Methodology: Random Matrix Model<br>for Multivariate PDFs"]
  B --> C["Data Input: Intraday returns<br>479 US stocks, S&P500, 2014"]
  C --> D["Computational Process: Fit model &<br>analyze tails across varying time resolutions"]
  D --> E["Key Finding: Correlations are non-stationary"]
  E --> F["Outcome: Non-stationarity increases<br>tail heaviness (Risk Assessment)"]

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 applies a Random Matrix Model to multivariate stock return distributions, providing few‑parameter formulae that capture heavy, algebraic tails and non‑stationarity.

Source sentence
Here, we apply these results to the returns measured in correlated stock markets.

Assumes 6

  • Returns are calculated from midpoint prices, assuming best ask and bid are available and comparable to transaction prices. data
    Source sentence
    From the midpoint price, we calculate the logarithmic returns
  • The set of stocks is continuously traded on every open day and remains in the S&P 500 index throughout 2014. market
    Source sentence
    We select stocks being continually traded on every open day while simultaneously being part of the S&P 500 index in 2014.
  • Correlation matrices are noisy and are optionally denoised with Ledoit–Wolf shrinkage, but the main analysis assumes the raw empirical matrices. model
    Source sentence
    We use Ledoit–Wolf shrinkage [77], which reduces noise in covariance matrices, on the distribution of the aggregated returns
  • Non‑stationarity of correlations is modeled by random matrices, implying that correlation fluctuations drive tail heaviness. model
    Source sentence
    The fluctuating correlations in the non‑stationary system are modeled by random matrices.
Show 2 more
  • Rotating returns into the eigenbasis of the correlation matrix preserves all information and facilitates analysis. statistical
    Source sentence
    We rotate the data into the eigenbasis of the correlation matrix.
  • Aggregating returns over longer horizons improves statistics without restricting results to the eigenbasis. statistical
    Source sentence
    By using the aggregated returns we facilitate the data analysis, but our results are not restricted to the eigenbasis

Data

  • Intraday returns of 479 US stocks · S&P 500 constituents market-prices public
    2014 · 479 stocks in the S&P 500 index
    Source sentence
    We analyze intraday data of 479 US stocks included in the S&P500 index during the trading year of 2014.
  • Daily adjusted prices of 308 US stocks · S&P 500 constituents market-prices public
    January 23, 1992 to December 31, 2012 · 308 stocks listed in the S&P 500 index
    Source sentence
    In total, it comprises 308 stocks for a time period ranging from January 23, 1992 to December 31, 2012. This data set lists daily adjusted prices.

Recognised sources: NYSE TAQ

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • Non‑stationary fluctuations of correlations make the return tails heavier, especially for longer time intervals. statistical
    Source sentence
    The non‑stationary fluctuations of the correlations lift the tails.
  • Four multivariate model distributions with heavy, mostly algebraic tails fit the empirical data well. theoretical
    Source sentence
    We obtained four multivariate model distributions with heavy, mostly algebraic tails which we fit to the data.
  • Aggregated‑return fits using Algebraic‑Algebraic, Algebraic‑Gaussian and Gaussian‑Algebraic models outperform the Gaussian‑Gaussian case. performance
    Source sentence
    the fits for ⟨p⟩(aggr)GA(er), ⟨p⟩(aggr)AG(er) and ⟨p⟩(aggr)AA(er) outperform the one
  • Fit parameters increase with larger return horizons, consistent with earlier studies. statistical
    Source sentence
    The fit parameters tend to be larger for a larger return horizon ∆t
  • Rotation into the eigenbasis captures full correlation information and does not limit applicability to that basis. theoretical
    Source sentence
    As opposed to the univariate distributions of the original, unrotated returns, the univariate distributions of the rotated returns depend on the eigenvalues of the correlation matrix and thus contain the full information on the correlated system.

Checked by others

Supported by

  • A Look at Financial Dependencies by Means of Econophysics and Financial Economics (2023)
    On the question: Do non‑stationary correlation fluctuations lead to heavier return tails that inflate the significance of correlation tests?
    Compare the findings

    This paper: Non‑stationary fluctuations of correlations make the return tails heavier, especially for longer time intervals.

    That paper: Parametric tests tend to overstate the significance of correlation coefficients because the tails of the permutated return distribution contain more extremes than those of independent normal noise.

    A finds parametric tests overstate significance due to extreme tails in permuted returns; B shows non‑stationary correlation fluctuations make tails heavier, especially over longer intervals, supporting A.

  • Beyond the Bid-Ask: Strategic Insights into Spread Prediction and the Global Mid-Price Phenomenon (2024)
    On the question: Do the log‑return distributions of TMOBBAS and GMP exhibit heavy‑tailed behavior?
    Compare the findings

    This paper: Four multivariate model distributions with heavy, mostly algebraic tails fit the empirical data well.

    That paper: Log‑return distributions of both TMOBBAS and GMP exhibit heavy tails and significant deviations from normality.

    Both findings report heavy tails: A notes heavy‑tailed log‑returns for TMOBBAS and GMP; B finds four multivariate models with algebraic (heavy) tails fit the data, supporting A.

Cited by 1 paper 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 (85)

  1. A new traders’ game? — Empirical analysis of response functions in a historical perspective — → 2503.01629
  2. Multivariate distributions in non-stationary complex systems I: A random matrix model and formulae for data analysis — → 2412.11601
  3. Efficient Multi-Change Point Analysis to Decode Economic Crisis Information from the S&P500 Mean Market Correlation — → 2308.00087
  4. Memory Effects, Multiple Time Scales and Local Stability in Langevin Models of the S&P500 Market Correlation — → 2307.12744
  5. Identifying dominant industrial sectors in market states of the S&P 500 financial data (2022) — link
  6. On financial market correlation structures and diversification benefits across and within equity sectors (2022) — link
  7. New collectivity measures for financial covariances and correlations (2022) — link
  8. Non-stationarity in correlation matrices for wind turbine SCADA-data and implications for failure detection (2021) — link
and 77 more
  1. Collective behavior in the North Rhine-Westphalia motorway network (2021) — link
  2. A new attempt to identify long-term precursors for endogenous financial crises in the market correlation structures (2021) — link
  3. Dynamics of the market states in the space of correlation matrices with applications to financial markets (2021) — link
  4. Exact multivariate amplitude distributions for non-stationary Gaussian or algebraic fluctuations of covariances or correlations (2020) — link
  5. Quasi-stationary states in temporal correlations for traffic systems: Cologne orbital motorway as an example (2020) — link
  6. Identification of structural multivariate GARCH models (2020) — link
  7. Uncovering the dynamics of correlation structures relative to the collective market motion (2020) — link
  8. Market states: A new understanding (2020) — link
  9. High-Dimensional Multivariate Forecasting with Low-Rank Gaussian Copula Processes (2019) — link
  10. Econophysics and sociophysics: Their milestones & challenges (2019) — link
  11. Reprint of: Mahalanobis, P.C. (1936) “On the Generalised Distance in Statistics.” (2018) — link
  12. Identifying long-term precursors of financial market crashes using correlation patterns (2018) — link
  13. Hawkes processes and their applications to finance: a review (2018) — link
  14. Extreme Portfolio Loss Correlations in Credit Risk (2017) — link
  15. A Review of Two Decades of Correlations, Hierarchies, Networks and Clustering in Financial Markets (2017) — link
  16. Multiplicative Conditional Correlation Models for Realized Covariance Matrices (2016) — link
  17. Concurrent credit portfolio losses (2016) — link
  18. Average cross-responses in correlated financial markets (2016) — link
  19. Cross-response in correlated financial markets: individual stocks (2016) — link
  20. Price response in correlated financial markets: empirical results (2015) — link
  21. Credit Risk: Taking Fluctuating Asset Correlations into Account (2015) — link
  22. Credit risk: taking fluctuating asset correlations into account (2015)
  23. Dependence structure of market states (2015) — link
  24. Constructing analytically tractable ensembles of stochastic covariances with an application to financial data (2015) — link
  25. Stability and hierarchy of quasi-stationary states: financial markets as an example (2015) — link
  26. Dynamics of quasi-stationary systems: Finance as an example (2015) — link
  27. Zooming into market states (2014) — link
  28. Modelling stochastic correlation (2014) — link
  29. Credit risk and the instability of the financial system: An ensemble approach (2013) — link
  30. PORTFOLIO RETURN DISTRIBUTIONS: SAMPLE STATISTICS WITH STOCHASTIC CORRELATIONS (2013) — link
  31. Non-stationarity in financial time series: Generic features and tail behavior (2013) — link
  32. The VIX, the Variance Premium and Stock Market Volatility (2013) — link
  33. Dynamic Conditional Correlation: On Properties and Estimation (2013) — link
  34. Identifying States of a Financial Market (2012) — link
  35. A copula approach on the dynamics of statistical dependencies in the US stock market (2011) — link
  36. Statistical causes for the Epps effect in microstructure noise (2010) — link
  37. The Conditional Autoregressive Wishart Model for Multivariate Stock Market Volatility (2010) — link
  38. Impact of the tick-size on financial returns and correlations (2010) — link
  39. Power mapping with dynamical adjustment for improved portfolio optimization (2010) — link
  40. Compensating asynchrony effects in the calculation of financial correlations (2009) — link
  41. X. On the criterion that a given system of deviations from the probable in the case of a correlated system of variables is such that it can be reasonably supposed to have arisen from random sampling (2009) — link
  42. A well-conditioned estimator for large-dimensional covariance matrices (2004) — link
  43. What really causes large price changes? (2003) — link
  44. The Long Memory of the Efficient Market (2003) — link
  45. Fluctuations and response in financial markets: the subtle nature of ‘random’ price changes (2003) — link
  46. A theory of power-law distributions in financial market fluctuations (2003) — link
  47. A Multivariate Generalized Autoregressive Conditional Heteroscedasticity Model With Time-Varying Correlations (2002) — link
  48. Dynamic Conditional Correlation (2002) — link
  49. A New Method to Estimate the Noise in Financial Correlation Matrices (2002) — link
  50. Quantifying and interpreting collective behavior in financial markets. (2001) — link
  51. Random matrix approach to cross correlations in financial data. (2001) — link
  52. An Introduction to Copulas (2000) — link
  53. An Introduction to Econophysics: Correlations and Complexity in Finance (1999)
  54. Scaling of the distribution of price fluctuations of individual companies. (1999) — link
  55. Scaling of the distribution of fluctuations of financial market indices. (1999) — link
  56. Noise Dressing of Financial Correlation Matrices (1998) — link
  57. Inverse cubic law for the distribution of stock price variations (1998) — link
  58. Finance (1991) — link
  59. Why Does Stock Market Volatility Change Over Time (1988) — link
  60. Comovements in Stock Prices in the Very Short Run (1979) — link
  61. DISTRIBUTION OF EIGENVALUES FOR SOME SETS OF RANDOM MATRICES (1967) — link
  62. Global industry classification sector (GICS) (2023)
  63. COVID-19 and March 2020 Stock Market Crash. Evidence from S&P1500 (2020) — link
  64. Data Reduction And Error Analysis For The Physical Sciences (2016)
  65. Multi-Asset Correlation Dynamics with Application to Trading (2015)
  66. New York stock exchange, Daily TAQ (Trade and Quote) (2014)
  67. Standard & Poor’s 500 stock data (2014)
  68. Pricing Foreign Equity Options with Stochastic Correlation and Volatility (2009)
  69. CHAPTER 2 – How Markets Slowly Digest Changes in Supply and Demand (2009) — link
  70. Modelling correlation as a stochastic process (2006)
  71. Asymmetric Volatility and Risk in Equity Markets (1997) — link
  72. Random variables, joint distribution functions, and copulas (1973)
  73. The Variation of Certain Speculative Prices (1963) — link
  74. Fonctions de repartition a n dimensions et leurs marges (1959)
  75. References
  76. S&P Dow
  77. MSCI, S

4 of 85 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.