arxiv: 2604.23961 · v1 · submitted 2026-04-27 · 📊 stat.AP · q-fin.MF· q-fin.TR

Recognition: unknown

Extended State-dependent Hawkes Process for Limit Order Books: Mathematical Foundation and the Reproduction of Volatility Signature Plots

Akitoshi Kimura

Authors on Pith no claims yet

Pith reviewed 2026-05-07 17:42 UTC · model grok-4.3

classification 📊 stat.AP q-fin.MFq-fin.TR

keywords state-dependent hawkes processlimit order bookvolatility signature plotmarketable limit orderslocal super-criticalityhigh-frequency tradingphysical constraints

0 comments

The pith

Extended state-dependent Hawkes process reproduces volatility signature plots by capturing local super-criticality from marketable limit orders

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes an Extended State-Dependent Hawkes Process for limit order books that permits state disappearances common in trading. It mathematically proves the maximum likelihood estimation stays separable under this extension using KKT conditions. In application to high-frequency data, the model reproduces the upward slope of volatility signature plots by modeling local super-criticality in disequilibrium, driven primarily by marketable limit orders. This succeeds where standard models fail due to instability, underscoring the need for physical consistency in such models.

Core claim

By allowing state disappearances and enforcing physical geometry that pauses residual accumulation during inadmissible periods, the ExsdHawkes model captures local super-criticality triggered by marketable limit orders, reproducing the characteristic upward slope in volatility signature plots, while models without these constraints exhibit explosive branching ratios and unstable simulations.

What carries the argument

The extended state-dependent Hawkes process with KKT conditions ensuring separable MLE and physical pause on residual accumulation in inadmissible states.

If this is right

Unconstrained models suffer explosive branching ratios and lose simulation stability.
Marketable limit orders act as the primary catalyst for forcing the LOB into unstable disequilibrium states.
Physical consistency is required to accurately link micro order flows to macro volatility patterns.
The two-step MLE procedure provides an efficient estimation method for the extended model.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Similar physical constraints might stabilize other Hawkes process applications in finance beyond limit order books.
The role of marketable limit orders could inform regulatory monitoring of market stress.
Testing the model on different stocks or exchanges would check if the MLO-super-criticality link holds generally.

Load-bearing premise

The load-bearing premise is that physical geometry must pause residual accumulation during inadmissible periods to preserve statistical integrity, and that marketable limit orders specifically cause the local super-criticality.

What would settle it

Removing the physical pause mechanism from ExsdHawkes and verifying whether the simulated volatility signature plots still show the upward slope or instead become unstable and fail to match data.

Figures

Figures reproduced from arXiv: 2604.23961 by Akitoshi Kimura.

**Figure 1.** Figure 1: Ensemble Volatility Signature Plot Comparison (MUFG 8306). The black line represents the empirical realized volatility over three months, showing the characteristic upward slope at high frequencies. Among the tested models, only ExsdHawkes (red) successfully prevents the explosive divergence observed in unconstrained models (SD-Hawkes) while accurately initiating the volatility capture at macro scales. Th… view at source ↗

**Figure 2.** Figure 2: Analysis of Local Super-criticality via Branching Ratios (Log Scale). All models exhibit branching ratios significantly exceeding 1.0, indicating a chronic state of super-criticality in high-frequency markets. ExsdHawkes captures this intensity while remaining stable through its structural ”physical gates.” To elucidate why unconstrained models explode while ExsdHawkes remains stable, we examine the degree… view at source ↗

**Figure 3.** Figure 3: Estimated Transition Probabilities ϕe(x, x′ ). The heatmaps empirically validate the KKT constraints. Specifically, price-improving orders (ALB, ALS) show near-zero probability at x = 1. proving that the model correctly ignores physically impossible events. 12 view at source ↗

**Figure 4.** Figure 4: Event-wise Aggregate QQ-Plots (Three-Month MUFG Data). The plots compare the residual distributions of ExsdHawkes (red), Constant Hawkes (green), and Poisson (dotted black) models against the theoretical unit exponential distribution. While ExsdHawkes and the Constant Hawkes model show comparable fit for high-volume events, ExsdHawkes provides superior alignment for price-moving aggressive orders, partic… view at source ↗

**Figure 5.** Figure 5: Total Residuals QQ-Plots: ExsdHawkes (a) vs. sdHawkes (b). A side-by-side comparison reveals the critical impact of physical constraints. In panels such as ALB, 1 and ALS, 1, where price improvement is physically impossible at x = 1, the unconstrained sdHawkes (b) exhibits significant deviation from the diagonal due to the accumulation of spurious residuals. Conversely, ExsdHawkes (a) maintains statistica… view at source ↗

**Figure 6.** Figure 6: Total Residual Correlograms for ExsdHawkes across 28 State-Event Pairs. The residuals exhibit near-zero autocorrelation across both equilibrium (x = 1) and disequilibrium (x = 2+) states. This successful “whitening” of the signal confirms that the statedependent framework, reinforced by physical constraints, effectively encompasses the complex temporal feedback loops and clustering inherent in the MUFG ti… view at source ↗

**Figure 7.** Figure 7: Total Residual Correlograms for the Unconstrained sdHawkes Model. In contrast to Figure 6, the residuals for sdHawkes show persistent, non-zero autocorrelations in several panels, particularly during disequilibrium phases. This failure to whiten the residuals indicates that without structural “gates”, unconstrained models leave behind systematic artifacts, failing to fully capture the explosive endogenei… view at source ↗

**Figure 8.** Figure 8: Event-by-event correlograms comparing ExsdHawkes, Constant Hawkes, and Poisson models. The consistent ‘whitening’ of residuals by ExsdHawkes (red) across all 196 panels underscores its robustness in capturing complex cross-event interactions without leaving systematic artifacts in the cross-sectional dynamics. 20 view at source ↗

read the original abstract

This paper proposes an Extended State-Dependent Hawkes Process (ExsdHawkes) to model the intricate dynamics of Limit Order Books (LOBs). Our theoretical contribution lies in relaxing traditional constraints by allowing for state disappearances -- a phenomenon frequently observed in high-frequency trading. We mathematically prove, using Karush--Kuhn--Tucker (KKT) conditions, that the maximum likelihood estimation remains separable, justifying an efficient two-step procedure. In the empirical section, we apply our model to three months of high-frequency tick data of Mitsubishi UFJ Financial Group (8306). We demonstrate that ExsdHawkes uniquely reproduces the volatility signature plot's characteristic upward slope by capturing the "local super-criticality" triggered during disequilibrium states. Crucially, we identify Marketable Limit Orders (MLO) as the primary catalyst that forces the LOB into these unstable states. Comparative analysis reveals that models lacking physical constraints (e.g., standard SD-Hawkes) suffer from explosive branching ratios and fail to maintain simulation stability. Our findings suggest that physical consistency is not merely a mathematical nicety, but a prerequisite for accurately modeling macro-level volatility. By enforcing the physical geometry to `pause' the residual accumulation during inadmissible periods, ExsdHawkes uniquely maintains statistical integrity where unconstrained models succumb to structural bias.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper adds a realistic extension to state-dependent Hawkes processes by allowing state disappearances and proves separability of MLE via KKT conditions, then shows it can reproduce volatility signature plots on one stock dataset.

read the letter

The main point is that this work relaxes standard state-dependent Hawkes models to let states disappear, which better matches limit order book behavior, and proves that maximum likelihood estimation stays separable under KKT conditions. That separability result supports a practical two-step fitting procedure. On the empirical side, the model applied to three months of Mitsubishi UFJ tick data reproduces the upward slope in volatility signature plots by generating local super-criticality in disequilibrium states, with marketable limit orders identified as the main trigger. Models without the physical constraints explode in simulations, so the paper argues that enforcing admissible periods is needed for stable macro-level volatility modeling. This is a concrete link between model features and a known empirical regularity in high-frequency data. The theoretical contribution looks solid and the empirical demonstration is direct, even if the abstract leaves the full derivation details out. The soft spots are clear: everything rests on a single stock and a three-month window, with no error bars, no out-of-sample checks, and no tests against alternative state definitions or other assets. The claim that physical consistency is a prerequisite and that MLOs are the primary driver could shift if the data window or event classification changes, which matches the stress-test concern about generalization. The separability proof itself appears independent of the data, but the attribution of the volatility slope does not. This is useful for researchers building constrained point-process models for order books or volatility. A reader working on high-frequency finance or Hawkes processes would get value from the separability result and the volatility-plot exercise. It deserves peer review because the math is grounded and the empirical target is well-defined, even though more datasets and robustness checks would strengthen it.

Referee Report

2 major / 2 minor

Summary. The paper proposes an Extended State-Dependent Hawkes Process (ExsdHawkes) for Limit Order Books that relaxes constraints to allow state disappearances. It proves via Karush-Kuhn-Tucker (KKT) conditions that maximum likelihood estimation remains separable, justifying a two-step procedure. Empirically, the model is fit to three months of tick data for Mitsubishi UFJ Financial Group (8306) and is shown to reproduce the characteristic upward slope of volatility signature plots by capturing local super-criticality (branching ratio >1) in disequilibrium states, with Marketable Limit Orders (MLO) identified as the primary trigger; models without physical constraints are reported to produce explosive behavior.

Significance. If the empirical reproduction of the volatility signature plot and the attribution to MLO-driven local super-criticality hold under broader validation, the work would advance applied statistics in high-frequency finance by linking enforceable physical LOB geometry to macro-level volatility features. The separability result via standard KKT conditions is a clear strength, as it supports efficient estimation without circularity in the theoretical component.

major comments (2)

[Empirical section] Empirical section: the reproduction of the volatility signature plot's upward slope and the identification of MLO as the primary catalyst for local super-criticality both rely on two-step MLE applied to the same 8306 dataset, with no reported out-of-sample validation, alternative state definitions, sensitivity to admissible/inadmissible period rules, or quantitative metrics (e.g., slope values with error bars or branching-ratio distributions) to support uniqueness or robustness.
[Comparative analysis] Comparative analysis: the claim that unconstrained models (e.g., standard SD-Hawkes) suffer from explosive branching ratios and fail to maintain simulation stability is asserted without specific reported values for branching ratios, instability metrics, or implementation details of the baseline, which is load-bearing for the conclusion that physical constraints are a prerequisite for statistical integrity.

minor comments (2)

[Abstract] The abstract and introduction should provide a brief equation or reference for the intensity function modification that enforces pausing of residual accumulation during inadmissible periods.
[Introduction] The manuscript would benefit from additional citations to prior state-dependent Hawkes models for LOBs to better situate the extension.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which highlight important areas for strengthening the empirical and comparative sections of the manuscript. We address each major comment point by point below, outlining the revisions we will incorporate to improve robustness and transparency while preserving the core contributions.

read point-by-point responses

Referee: [Empirical section] Empirical section: the reproduction of the volatility signature plot's upward slope and the identification of MLO as the primary catalyst for local super-criticality both rely on two-step MLE applied to the same 8306 dataset, with no reported out-of-sample validation, alternative state definitions, sensitivity to admissible/inadmissible period rules, or quantitative metrics (e.g., slope values with error bars or branching-ratio distributions) to support uniqueness or robustness.

Authors: We agree that the current empirical demonstration relies on in-sample fitting to the full three-month 8306 dataset and lacks explicit out-of-sample checks or quantitative robustness metrics. While the in-sample reproduction of the upward slope in volatility signature plots and the attribution to MLO-driven local super-criticality are directly supported by the two-step MLE procedure and the physical constraints, we will revise the manuscript to include: (i) out-of-sample validation on a held-out portion of the tick data, (ii) sensitivity analyses to alternative state definitions and admissible/inadmissible period rules, and (iii) quantitative metrics such as slope estimates with error bars and empirical distributions of branching ratios. These additions will better substantiate the uniqueness and robustness of the findings. revision: yes
Referee: [Comparative analysis] Comparative analysis: the claim that unconstrained models (e.g., standard SD-Hawkes) suffer from explosive branching ratios and fail to maintain simulation stability is asserted without specific reported values for branching ratios, instability metrics, or implementation details of the baseline, which is load-bearing for the conclusion that physical constraints are a prerequisite for statistical integrity.

Authors: We acknowledge that the comparative claims regarding explosive behavior in unconstrained models (such as standard SD-Hawkes) are presented qualitatively in the manuscript without accompanying numerical values or instability metrics. To strengthen this section, we will add specific reported branching-ratio values for the unconstrained baseline, quantitative instability metrics (e.g., the rate of explosive simulations across repeated runs), and additional implementation details on how the baseline was simulated and estimated. This will provide concrete evidence supporting the necessity of the physical constraints for maintaining simulation stability and statistical integrity. revision: yes

Circularity Check

0 steps flagged

No significant circularity; mathematical proof and empirical demonstration remain independent

full rationale

The paper's derivation chain begins with a standard application of KKT conditions to prove separability of the MLE for the extended Hawkes process, which is a self-contained optimization result independent of any fitted parameters or data. The subsequent empirical application fits the model to three months of tick data for one stock and then simulates to reproduce the volatility signature plot's upward slope while attributing it to local super-criticality from MLOs; this is a conventional post-fit validation exercise rather than a quantity that reduces by construction to the inputs (no parameter is fitted directly to the slope and then relabeled a prediction). No self-citations appear as load-bearing premises, no ansatz is smuggled, and no known empirical pattern is merely renamed. The physical-geometry constraint is an explicit modeling choice whose necessity is tested by comparison to unconstrained variants, keeping the central claims falsifiable against the data rather than tautological.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Review performed on abstract only; no explicit free parameters, axioms, or invented entities are stated in the provided text. The model extension implicitly introduces a mechanism for state disappearance whose justification rests on observed high-frequency trading behavior.

pith-pipeline@v0.9.0 · 5538 in / 1202 out tokens · 45890 ms · 2026-05-07T17:42:58.051080+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

11 extracted references

[1]

Cambridge University Press, 2016

Fr´ ed´ eric Abergel, Marouane Anane, Anirban Chakraborti, Aymen Jedidi, and Ioane Muni Toke.Limit order books. Cambridge University Press, 2016

2016
[2]

Hawkes processes in finance.Market Microstructure and Liquidity, 1(01):1550005, 2015

Emmanuel Bacry, Iacopo Mastromatteo, and Jean-Fran¸ cois Muzy. Hawkes processes in finance.Market Microstructure and Liquidity, 1(01):1550005, 2015

2015
[3]

Statistical modeling of high-frequency financial data.IEEE Signal Pro- cessing Magazine, 28(5):16–25, 2011

Rama Cont. Statistical modeling of high-frequency financial data.IEEE Signal Pro- cessing Magazine, 28(5):16–25, 2011

2011
[4]

Quantifying reflexivity in financial markets: Toward a prediction of flash crashes.Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, 85(5):056108, 2012

Vladimir Filimonov and Didier Sornette. Quantifying reflexivity in financial markets: Toward a prediction of flash crashes.Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, 85(5):056108, 2012

2012
[5]

Measuring the resiliency of an electronic limit order book.Journal of Financial Markets, 10(1):1–25, 2007

Jeremy Large. Measuring the resiliency of an electronic limit order book.Journal of Financial Markets, 10(1):1–25, 2007

2007
[6]

High-dimensional hawkes processes for limit order books: modelling, empirical analysis and numerical calibration.Quantitative Finance, 18(2):249–264, 2018

Xiaofei Lu and Fr´ ed´ eric Abergel. High-dimensional hawkes processes for limit order books: modelling, empirical analysis and numerical calibration.Quantitative Finance, 18(2):249–264, 2018

2018
[7]

P. A. Meyer. Demonstration simplifiee d’un theoreme de knight. InS´ eminaire de Prob- abilit´ es V Universit´ e de Strasbourg, pages 191–195, Berlin, Heidelberg, 1971. Springer Berlin Heidelberg. ISBN 978-3-540-36517-4

1971
[8]

State-dependent hawkes processes and their application to limit order book modelling.Quantitative Finance, 22(3):563– 583, 2022

Maxime Morariu-Patrichi and Mikko S Pakkanen. State-dependent hawkes processes and their application to limit order book modelling.Quantitative Finance, 22(3):563– 583, 2022

2022
[9]

Masking” Strategy This derivation correctly reflects the logic of the “state-dependent masking

Yosihiko Ogata. On lewis’ simulation method for point processes.IEEE transactions on information theory, 27(1):23–31, 1981. 18 A Proof of Theorem 3.2 In this section, we derive the optimal transition probabilities ˆϕe(x, x′) using the Karush– Kuhn–Tucker (KKT) conditions. To avoid confusion with the intensityλ, we denote the Lagrange multipliers asη e,x. ...

1981
[10]

Consistency: The separability proved in Theorem 3.1 ensures thatϕcan be estimated purely from counts without knowing the Hawkes parameters (ν,θ)
[11]

physical gate

Zero-frequency robustness: IfN e(x) = 0, the KKT condition is satisfied by ˆϕ= 0. This naturally implements the “physical gate” where the intensityλ † becomes zero, preventing unobserved or impossible transitions from biasing the model. B Figures 19 /uni00000013 /uni00000014 /uni00000031/uni00000030/uni00000036/uni00000031/uni00000030/uni00000036 /uni0000...