Heterogeneous Determinism Across Stocks and Scales: A Multiscale RQA Analysis of 5-Minute Financial Data
Autor(ka) práce:
Mashburn, Calum
Typ práce:
Diploma thesis
Vedoucí práce:
Brůna, Karel
Oponenti práce:
Tran, Van Quang
Jazyk práce:
English
Abstrakt:
Abstract The null hypothesis of linear stochasticity is rejected across the majority of the sample. At 30-minute resolution, all 18 securities achieved significance (100%), with z-scores orders of magnitude beyond what multiple testing or measurement noise could plausibly explain. Market dynamics are found to be strongly scale-dependent: high-frequency dynamics exhibit strong determinism, near-random or anti-persistent Hurst values, and vigorous chaos, while dynamics at lower frequencies show weaker determinism, persistent Hurst exponents, and reduced Lyapunov exponents. Wavelet preprocessing is demonstrated not to create spurious determinism; z-scores collapse by 90–92% with filtering, consistent with genuine high-frequency signal removal rather than artifact generation. Heterogeneity in the magnitude of dynamical structure across securities suggests that not all equities inhabit the same dynamical regime, with implications for which instruments and timescales are most amenable to quantitative analysis strategies. The findings support a characterization of equity and futures markets as exhibiting scale-dependent stochastic-chaotic dynamics, and challenge both the random walk hypothesis and scale-invariant fractal market models.
Heterogeneous Deterministic Signatures Across Stocks and Scales: A Multiscale RQA Analysis of Intraday Financial Data
Autor(ka) práce:
Mashburn, Calum
Typ práce:
Diplomová práce
Vedoucí práce:
Brůna, Karel
Oponenti práce:
Tran, Van Quang
Jazyk práce:
English
Abstrakt:
Abstract The null hypothesis of linear stochasticity is rejected across the majority of the sample. At 30-minute resolution, all 18 securities achieved significance (100%), with z-scores orders of magnitude beyond what multiple testing or measurement noise could plausibly explain. Market dynamics are found to be strongly scale-dependent: high-frequency dynamics exhibit strong determinism, near-random or anti-persistent Hurst values, and vigorous chaos, while dynamics at lower frequencies show weaker determinism, persistent Hurst exponents, and reduced Lyapunov exponents. Wavelet preprocessing is demonstrated not to create spurious determinism; z-scores collapse by 90–92% with filtering, consistent with genuine high-frequency signal removal rather than artifact generation. Heterogeneity in the magnitude of dynamical structure across securities suggests that not all equities inhabit the same dynamical regime, with implications for which instruments and timescales are most amenable to quantitative analysis strategies. The findings support a characterization of equity and futures markets as exhibiting scale-dependent stochastic-chaotic dynamics, and challenge both the random walk hypothesis and scale-invariant fractal market models.