Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
134 tokens/sec
GPT-4o
10 tokens/sec
Gemini 2.5 Pro Pro
47 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Linear and nonlinear causality in financial markets (2312.16185v1)

Published 18 Dec 2023 in q-fin.ST and physics.data-an

Abstract: Identifying and quantifying co-dependence between financial instruments is a key challenge for researchers and practitioners in the financial industry. Linear measures such as the Pearson correlation are still widely used today, although their limited explanatory power is well known. In this paper we present a much more general framework for assessing co-dependencies by identifying and interpreting linear and nonlinear causalities in the complex system of financial markets. To do so, we use two different causal inference methods, transfer entropy and convergent cross-mapping, and employ Fourier transform surrogates to separate their linear and nonlinear contributions. We find that stock indices in Germany and the U.S. exhibit a significant degree of nonlinear causality and that correlation, while a very good proxy for linear causality, disregards nonlinear effects and hence underestimates causality itself. The presented framework enables the measurement of nonlinear causality, the correlation-causality fallacy, and motivates how causality can be used for inferring market signals, pair trading, and risk management of portfolios. Our results suggest that linear and nonlinear causality can be used as early warning indicators of abnormal market behavior, allowing for better trading strategies and risk management.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (56)
  1. F. Jovanovic, R. N. Mantegna,  and C. Schinckus, “When financial economics influences physics: The role of econophysics,” Available at SSRN 3294548  (2018).
  2. G.-J. Wang, C. Xie, S. Chen, J.-J. Yang,  and M.-Y. Yang, “Random matrix theory analysis of cross-correlations in the us stock market: Evidence from pearson’s correlation coefficient and detrended cross-correlation coefficient,” Physica A: statistical mechanics and its applications 392, 3715–3730 (2013).
  3. R. N. Mantegna and H. E. Stanley, Introduction to econophysics: correlations and complexity in finance (Cambridge university press, 1999).
  4. G.-J. Wang, C. Xie,  and H. E. Stanley, “Correlation structure and evolution of world stock markets: Evidence from pearson and partial correlation-based networks,” Computational Economics 51, 607–635 (2018).
  5. R. N. Mantegna and H. E. Stanley, “Scaling behaviour in the dynamics of an economic index,” Nature 376, 46–49 (1995).
  6. S. Ghashghaie, W. Breymann, J. Peinke, P. Talkner,  and Y. Dodge, “Turbulent cascades in foreign exchange markets,” Nature 381, 767–770 (1996).
  7. A. Haluszczynski, I. Laut, H. Modest,  and C. Räth, “Linear and nonlinear market correlations: Characterizing financial crises and portfolio optimization,” Physical Review E 96, 062315 (2017).
  8. J. Benesty, J. Chen, Y. Huang,  and I. Cohen, “Pearson correlation coefficient,” in Noise reduction in speech processing (Springer, 2009) pp. 1–4.
  9. C. W. Granger, Essays in econometrics: collected papers of Clive WJ Granger, Vol. 32 (Cambridge University Press, 2001).
  10. T. Schreiber, “Measuring information transfer,” Physical review letters 85, 461 (2000).
  11. G. Sugihara, R. May, H. Ye, C.-h. Hsieh, E. Deyle, M. Fogarty,  and S. Munch, “Detecting causality in complex ecosystems,” science 338, 496–500 (2012).
  12. X. Ge and A. Lin, “Dynamic causality analysis using overlapped sliding windows based on the extended convergent cross-mapping,” Nonlinear Dynamics 104, 1753–1765 (2021).
  13. M. Paluš and M. Vejmelka, “Directionality of coupling from bivariate time series: How to avoid false causalities and missed connections,” Physical Review E 75, 056211 (2007).
  14. J. Hlinka, D. Hartman, M. Vejmelka, D. Novotná,  and M. Paluš, “Non-linear dependence and teleconnections in climate data: sources, relevance, nonstationarity,” Climate dynamics 42, 1873–1886 (2014).
  15. A. L. Lloyd, “The coupled logistic map: a simple model for the effects of spatial heterogeneity on population dynamics,” Journal of Theoretical Biology 173, 217–230 (1995).
  16. J. B. DeLong and K. Magin, “A short note on the size of the dot-com bubble,”  (2006).
  17. H. P. Krishnan, A. Bennington, H. P. Krishnan,  and A. Bennington, “The vix “volmaggedon”, with exchange-traded notes destabilizing the market,” Market Tremors: Quantifying Structural Risks in Modern Financial Markets , 83–119 (2021).
  18. E. Zivot, J. Wang, E. Zivot,  and J. Wang, “Rolling analysis of time series,” Modeling financial time series with S-Plus® , 299–346 (2003).
  19. J. Runge, “Causal network reconstruction from time series: From theoretical assumptions to practical estimation,” Chaos: An Interdisciplinary Journal of Nonlinear Science 28, 075310 (2018).
  20. K. Pearson, “Vii. note on regression and inheritance in the case of two parents,” proceedings of the royal society of London 58, 240–242 (1895).
  21. D. G. Bonett and T. A. Wright, “Sample size requirements for estimating pearson, kendall and spearman correlations,” Psychometrika 65, 23–28 (2000).
  22. M. Mynter, “Evaluation and extension of the transfer entropy calculus for the measurement of information flows between futures time series during the covid-19 pandemic,”  (2021).
  23. F. Takens, “Detecting strange attractors in turbulence,” in Dynamical Systems and Turbulence, Warwick 1980: proceedings of a symposium held at the University of Warwick 1979/80 (Springer, 2006) pp. 366–381.
  24. D. Mønster, R. Fusaroli, K. Tylén, A. Roepstorff,  and J. F. Sherson, “Causal inference from noisy time-series data–testing the convergent cross-mapping algorithm in the presence of noise and external influence,” Future Generation Computer Systems 73, 52–62 (2017).
  25. S. Wallot and D. Mønster, “Calculation of average mutual information (ami) and false-nearest neighbors (fnn) for the estimation of embedding parameters of multidimensional time series in matlab,” Frontiers in psychology 9, 1679 (2018).
  26. M. B. Kennel, R. Brown,  and H. D. Abarbanel, “Determining embedding dimension for phase-space reconstruction using a geometrical construction,” Physical review A 45, 3403 (1992).
  27. J. M. McCracken and R. S. Weigel, “Convergent cross-mapping and pairwise asymmetric inference,” Physical Review E 90, 062903 (2014).
  28. L. Overbey and M. Todd, “Effects of noise on transfer entropy estimation for damage detection,” Mechanical Systems and Signal Processing 23, 2178–2191 (2009).
  29. P. Krishna and A. K. Tangirala, “Inferring direct causality from noisy data using convergent cross mapping,” in 2019 58th Annual Conference of the Society of Instrument and Control Engineers of Japan (SICE) (IEEE, 2019) pp. 1523–1528.
  30. B. Cummins, T. Gedeon,  and K. Spendlove, “On the efficacy of state space reconstruction methods in determining causality,” SIAM Journal on Applied Dynamical Systems 14, 335–381 (2015).
  31. C. Räth and R. Monetti, “Surrogates with random fourier phases,” in Topics on Chaotic Systems: Selected Papers from Chaos 2008 International Conference (World Scientific, 2009) pp. 274–285.
  32. C. Räth, M. Gliozzi, I. Papadakis,  and W. Brinkmann, “Revisiting algorithms for generating surrogate time series,” Physical review letters 109, 144101 (2012).
  33. E. O. Brigham, The fast Fourier transform and its applications (Prentice-Hall, Inc., 1988).
  34. D. Prichard and J. Theiler, “Generating surrogate data for time series with several simultaneously measured variables,” Physical review letters 73, 951 (1994a).
  35. E. Kasuya, “On the use of r and r squared in correlation and regression,” Tech. Rep. (Wiley Online Library, 2019).
  36. M. Maziarz, “A review of the granger-causality fallacy,” The journal of philosophical economics: Reflections on economic and social issues 8, 86–105 (2015).
  37. G. Vidyamurthy, Pairs Trading: quantitative methods and analysis, Vol. 217 (John Wiley & Sons, 2004).
  38. J. C. Hull, “Options, futures and other derivatives,”  (2019).
  39. H. M. Markowitz, “Foundations of portfolio theory,” The journal of finance 46, 469–477 (1991).
  40. D. Duffie and J. Pan, “An overview of value at risk,” Journal of derivatives 4, 7–49 (1997).
  41. P. Bérard, G. Besson,  and S. Gallot, “Embedding riemannian manifolds by their heat kernel,” Geometric & Functional Analysis GAFA 4, 373–398 (1994).
  42. A. A. Tsonis, E. R. Deyle, H. Ye,  and G. Sugihara, “Convergent cross mapping: theory and an example,” in Advances in Nonlinear Geosciences (Springer, 2018) pp. 587–600.
  43. L. Barnett, A. B. Barrett,  and A. K. Seth, “Granger causality and transfer entropy are equivalent for gaussian variables,” Physical review letters 103, 238701 (2009).
  44. T. Schreiber and A. Schmitz, “Improved surrogate data for nonlinearity tests,”  (1999), arXiv:chao-dyn/9909041 [chao-dyn] .
  45. D. Prichard and J. Theiler, “Generating surrogate data for time series with several simultaneously measured variables,” Phys. Rev. Lett. 73, 951–954 (1994b).
  46. “redm: An r package for empirical dynamic modeling and convergent cross mapping,” https://mran.revolutionanalytics.com/snapshot/2018-07-06/web/packages/rEDM, accessed: 2020-08-20.
  47. S. L. Bressler and A. K. Seth, “Wiener–granger causality: a well established methodology,” Neuroimage 58, 323–329 (2011).
  48. S. Ponczek, “To understand the wild u.s. stock rally, just forget about 2020,”  (2020).
  49. M. Paluš, V. Albrecht,  and I. Dvořák, “Information theoretic test for nonlinearity in time series,” Physics Letters A 175, 203–209 (1993).
  50. J. Hlinka, D. Hartman, M. Vejmelka, J. Runge, N. Marwan, J. Kurths,  and M. Paluš, “Reliability of inference of directed climate networks using conditional mutual information,” Entropy 15, 2023–2045 (2013).
  51. M. B. Kennel and M. Buhl, ‘‘Estimating good discrete partitions from observed data: Symbolic false nearest neighbors,” Physical Review Letters 91, 084102 (2003).
  52. D. A. Hsieh, “Nonlinear dynamics in financial markets: evidence and implications,” Financial Analysts Journal 51, 55–62 (1995).
  53. R. N. Mantegna, “Hierarchical structure in financial markets,” The European Physical Journal B-Condensed Matter and Complex Systems 11, 193–197 (1999).
  54. R. Cont, ‘‘Empirical properties of asset returns: stylized facts and statistical issues,” Quantitative finance 1, 223 (2001).
  55. D. P. Doane and L. E. Seward, “Measuring skewness: a forgotten statistic?” Journal of statistics education 19 (2011).
  56. J. C. Hull, Options futures and other derivatives (Pearson Education India, 2003).
Citations (1)
View on Semantic Scholar

Summary

  • The paper introduces a dual-method framework that distinguishes linear from nonlinear causal relationships in financial markets using Transfer Entropy and Convergent Cross-Mapping.
  • It demonstrates that conventional measures like Pearson correlation overlook significant nonlinear market dependencies identified via Fourier transform surrogates.
  • The study shows that integrating causal measures can serve as early warning indicators for market volatility, improving trading strategies and risk management.

Linear and Nonlinear Causality in Financial Markets

The paper "Linear and Nonlinear Causality in Financial Markets" presents a comprehensive framework for identifying and quantifying co-dependencies among financial instruments, with a focus on both linear and nonlinear causal relationships. This research is particularly significant given the limitations of traditional linear measures, such as the Pearson correlation, which have long been used in financial markets to explain asset co-dependence. The insufficiency of these linear measures in capturing the full spectrum of market dynamics necessitates more sophisticated approaches.

At the core of this research are two causal inference methodologies: Transfer Entropy (TE) and Convergent Cross-Mapping (CCM). Both methods are employed to measure causal relationships, with Fourier transform surrogates used to separate the linear from the nonlinear components. This distinction is crucial as it highlights the potential oversight of significant nonlinear dependencies by conventional linear measures.

The authors apply their framework to major German and U.S. stock indices, uncovering substantial nonlinear causalities which have been largely underestimated in previous analyses. This reveals an important insight: that correlation serves as a good proxy merely for linear causality and neglects the nonlinear effects that are fundamental to an accurate understanding of financial markets. By incorporating causality into their analysis, the authors provide a more nuanced approach, emphasizing the correlation-causality fallacy and its implications for market interpretation.

The authors further reveal that both linear and nonlinear causality can be instrumental as early warning indicators of abnormal market behavior. This has practical implications, suggesting that more effective trading strategies and improved risk management of portfolios can be achieved by integrating these measures. For instance, during significant economic events such as the dotcom bubble burst and the 2008 financial crisis, the nonlinear contribution to market dynamics becomes more pronounced. By continuously monitoring these causal measures, practitioners can better anticipate and respond to potential market disruptions.

The paper not only offers an innovative approach to understanding market dynamics but also proposes tangible applications, including market signal inference, pair trading, and portfolio optimization. The authors demonstrate that incorporating these causality measures into traditional financial models can lead to better performance outcomes. Particularly in pair trading and portfolio optimization scenarios, causality measures provide a more robust framework compared to traditional reliance on Pearson correlation alone.

In conclusion, the framework outlined in this paper sheds light on the complexity of financial market interactions by presenting a clear method to distinguish between linear and nonlinear causalities. As the financial landscape continues to evolve, the proposed methodologies offer promising avenues for future research, potentially leading to the development of advanced trading algorithms and enhanced prediction models. While this paper is oriented towards identifying causal relationships, its broader implications can contribute to an improved theoretical understanding of market dynamics, offering a richer context for financial decision-making and risk management.

File Document Download Save Streamline Icon: https://streamlinehq.com PDF File Document Download Save Streamline Icon: https://streamlinehq.com Markdown