Selected work

Representative Papers

Temporal Coarse-Graining of Latent Default-Probability Paths Generates Effective Default Correlation

May 30, 2026

Long-term default counts often exhibit overdispersion, autocorrelation, and scale-dependent default correlation, which are frequently misattributed to contagion or asset dependence. This work proposes a temporal coarse-graining approach within an Ornstein–Uhlenbeck-driven latent intensity and binomial observation framework (OU–binomial), wherein aggregating default probability paths yields a scale-consistent effective mixture distribution that naturally accounts for these statistical features. This mechanism serves as a benchmark that mitigates overattribution to effects such as Davis–Lo contagion or Vasicek common factors, thereby enhancing model identifiability and predictive consistency. Under controlled residual covariance contributions, the method significantly outperforms direct scale-wise fitting, achieving markedly higher block-wise expected log predictive density.

1 citationsRead paper

Crude, Commercial, and Self-Referential: Chinese-Language Coordinated Activity in Japanese-Language X

Sep 28, 2026

This study addresses the lack of characterization regarding malicious coordinated behavior and ordinary user responses in existing research. Leveraging a large-scale dataset comprising 730,000 accounts and 495 million posts from platform X, this work integrates large-scale data mining, text duplication detection, and longitudinal quantitative analysis to provide the first quantitative mapping of the distributional characteristics and user response mechanisms underlying crude Chinese-language coordination campaigns within Japanese information streams. The findings reveal that most coordinating accounts retain traces of automation while producing predominantly non-political content. Furthermore, large-scale cascades primarily stem from internal self-referential interactions rather than external diffusion. These results uncover a novel propagation paradigm for cross-lingual coordinated inauthentic behavior, offering critical insights into how such campaigns operate across linguistic boundaries on social media platforms.

0 citationsRead paper

Attentional DoS: Repeat Reposting, Collective Attention, and Information Diffusion on X

Sep 28, 2026

This study addresses how repetitive retweeting on social media monopolizes collective attention and distorts information diffusion. Focusing on the X platform, this work innovatively proposes the concept of “Attention Denial-of-Service (Attention DoS)” and employs an integrated methodology combining large-scale cascade analysis, sequence modeling, lexical matching, and temporal synchronization detection. The results reveal that repetitive retweeting exhibits characteristics of distributed, synchronized amplification, primarily reinforcing reach among existing audiences rather than competing for new traffic. Furthermore, such behavior is strongly associated with Chinese-language commercial spam. By elucidating the propagation dynamics underlying malicious attention manipulation, this research offers a novel perspective and empirical evidence to inform governance strategies for social media platforms.

0 citationsRead paper

Uncovering Non-Normality in Information Flow: Network Structure and Dynamics of Social Media Cascades

Sep 27, 2026

This study addresses the limitation of traditional models in quantifying the non-hierarchical properties of social media information cascades, which frequently deviate from ideal tree structures. By introducing Henrici spectral non-normality into social diffusion research for the first time, this work integrates graph theory with statistical modeling to quantify the directional asymmetry and hierarchical structure of approximately 58,000 cascade networks on platform X. The analysis reveals that spectral non-normality correlates strongly with peak concentration rather than total cascade size, uncovering rapid asymmetric retweeting mechanisms and establishing a quantifiable predictive benchmark for directional diffusion architectures. Furthermore, when 50%–60% of nodes are observed, structural prediction accuracy exceeds 80% across all dynamic clusters, demonstrating the robustness of the proposed framework in characterizing complex cascade dynamics.

0 citationsRead paper

Temporal Coarse-Graining of Multi-Sector Default Count Data Generates Posterior-Implied Copulas

Jun 20, 2026

This study addresses the limitations of existing static correlation or factor models in capturing the distinct dependence structures between monthly and annual credit default data. The authors propose a dynamic low-rank state-space model that integrates monthly multi-sector default counts and employs temporal coarse-graining to derive term-dependent default probability distributions and their implied copula structures. Innovatively combining temporal coarse-graining with posterior-implied copulas, the framework reveals how default dependence evolves across time scales. The model features AR(1) latent factors, a binomial observation layer, principal eigenvector loadings, and survival aggregation. A two-factor specification successfully replicates the empirically observed amplification of the leading eigenvalue at annual horizons and significantly improves interval coverage, CRPS scores, and sector-level calibration accuracy for portfolio default forecasts.

0 citationsRead paper

Recent publications

Latest Papers

Crude, Commercial, and Self-Referential: Chinese-Language Coordinated Activity in Japanese-Language X

Sep 28, 2026

This study addresses the lack of characterization regarding malicious coordinated behavior and ordinary user responses in existing research. Leveraging a large-scale dataset comprising 730,000 accounts and 495 million posts from platform X, this work integrates large-scale data mining, text duplication detection, and longitudinal quantitative analysis to provide the first quantitative mapping of the distributional characteristics and user response mechanisms underlying crude Chinese-language coordination campaigns within Japanese information streams. The findings reveal that most coordinating accounts retain traces of automation while producing predominantly non-political content. Furthermore, large-scale cascades primarily stem from internal self-referential interactions rather than external diffusion. These results uncover a novel propagation paradigm for cross-lingual coordinated inauthentic behavior, offering critical insights into how such campaigns operate across linguistic boundaries on social media platforms.

0 citationsRead paper

Attentional DoS: Repeat Reposting, Collective Attention, and Information Diffusion on X

Sep 28, 2026

This study addresses how repetitive retweeting on social media monopolizes collective attention and distorts information diffusion. Focusing on the X platform, this work innovatively proposes the concept of “Attention Denial-of-Service (Attention DoS)” and employs an integrated methodology combining large-scale cascade analysis, sequence modeling, lexical matching, and temporal synchronization detection. The results reveal that repetitive retweeting exhibits characteristics of distributed, synchronized amplification, primarily reinforcing reach among existing audiences rather than competing for new traffic. Furthermore, such behavior is strongly associated with Chinese-language commercial spam. By elucidating the propagation dynamics underlying malicious attention manipulation, this research offers a novel perspective and empirical evidence to inform governance strategies for social media platforms.

0 citationsRead paper

Uncovering Non-Normality in Information Flow: Network Structure and Dynamics of Social Media Cascades

Sep 27, 2026

This study addresses the limitation of traditional models in quantifying the non-hierarchical properties of social media information cascades, which frequently deviate from ideal tree structures. By introducing Henrici spectral non-normality into social diffusion research for the first time, this work integrates graph theory with statistical modeling to quantify the directional asymmetry and hierarchical structure of approximately 58,000 cascade networks on platform X. The analysis reveals that spectral non-normality correlates strongly with peak concentration rather than total cascade size, uncovering rapid asymmetric retweeting mechanisms and establishing a quantifiable predictive benchmark for directional diffusion architectures. Furthermore, when 50%–60% of nodes are observed, structural prediction accuracy exceeds 80% across all dynamic clusters, demonstrating the robustness of the proposed framework in characterizing complex cascade dynamics.

0 citationsRead paper

Temporal Coarse-Graining of Multi-Sector Default Count Data Generates Posterior-Implied Copulas

Jun 20, 2026

This study addresses the limitations of existing static correlation or factor models in capturing the distinct dependence structures between monthly and annual credit default data. The authors propose a dynamic low-rank state-space model that integrates monthly multi-sector default counts and employs temporal coarse-graining to derive term-dependent default probability distributions and their implied copula structures. Innovatively combining temporal coarse-graining with posterior-implied copulas, the framework reveals how default dependence evolves across time scales. The model features AR(1) latent factors, a binomial observation layer, principal eigenvector loadings, and survival aggregation. A two-factor specification successfully replicates the empirically observed amplification of the leading eigenvalue at annual horizons and significantly improves interval coverage, CRPS scores, and sector-level calibration accuracy for portfolio default forecasts.

0 citationsRead paper

Temporal Coarse-Graining of Latent Default-Probability Paths Generates Effective Default Correlation

May 30, 2026

Long-term default counts often exhibit overdispersion, autocorrelation, and scale-dependent default correlation, which are frequently misattributed to contagion or asset dependence. This work proposes a temporal coarse-graining approach within an Ornstein–Uhlenbeck-driven latent intensity and binomial observation framework (OU–binomial), wherein aggregating default probability paths yields a scale-consistent effective mixture distribution that naturally accounts for these statistical features. This mechanism serves as a benchmark that mitigates overattribution to effects such as Davis–Lo contagion or Vasicek common factors, thereby enhancing model identifiability and predictive consistency. Under controlled residual covariance contributions, the method significantly outperforms direct scale-wise fitting, achieving markedly higher block-wise expected log predictive density.

1 citationsRead paper