Selected work

Representative Papers

The"double"square-root law: Evidence for the mechanical origin of market impact using Tokyo Stock Exchange data

Feb 22, 2025

This paper addresses the long-standing debate on the microfoundations of price impact: whether it arises mechanically from order flow or informationally from informed trading. Using high-frequency, trader-identified order-level data from the Tokyo Stock Exchange (2012–2018), we provide the first empirical evidence of the square-root impact law at the individual order level and discover that its temporal decay follows an inverse square-root pattern—collectively termed the “double square-root law”: impact ∝ √volume × 1/√time. Through meta-order reconstruction, anonymized control experiments, and nonparametric impact curve estimation, we demonstrate the robustness of this law and show that synthetically reconstructed meta-orders replicate observed impact dynamics. Our findings strongly support a purely mechanical origin of price impact, offering the first high-resolution empirical validation for market microstructure theory and challenging the dominant informational paradigm.

2 citationsRead paper

LightSBB-M: Bridging Schr\"odinger and Bass for Generative Diffusion Modeling

Jan 27, 2026

This work addresses the computational bottleneck in jointly controlling drift and diffusion in generative diffusion models by proposing the LightSBB-M algorithm. Within the Schrödinger Bridge and Bass (SBB) joint modeling framework, it achieves the first analytical solution to the SBB problem. By leveraging a dual representation of the objective function, the method explicitly derives the optimal drift and diffusion coefficients and introduces a tunable parameter β to continuously interpolate between them, thereby unifying the Schrödinger bridge and Bass transport mechanisms. The approach significantly enhances both computational efficiency and generation quality: on synthetic data, it reduces the 2-Wasserstein distance by up to 32% compared to existing methods, and demonstrates high-fidelity unpaired image translation in the challenging task of adult-to-child face conversion on FFHQ.

1 citationsRead paper

The Fundamental Structure of Risk: From Characteristics to Covariance

Jul 27, 2026

This study addresses the limitations of traditional risk models, which rely on high-frequency return data and suffer from noise-sensitive covariance estimates that poorly generalize to new assets. The authors propose a Characteristic-Driven Dynamic Factor Model (CD-DFM) that leverages low-frequency, observable firm characteristics—such as fundamental accounting variables—to learn interpretable factor exposures and forward-looking covariances through an end-to-end nonlinear latent factor representation. Notably, CD-DFM enables zero-shot embedding of entirely new assets using only their low-frequency features, while simultaneously preserving factor interpretability, achieving well-calibrated covariance forecasts, and maintaining economic plausibility. Empirical results on S&P 500 equities demonstrate that the model’s latent factors exhibit clear economic meaning, yield highly interpretable factor portfolios, and deliver competitive performance in covariance prediction.

0 citationsRead paper

PALoRA: Projection-Adaptive LoRA for Preserving Reasoning in Large Language Models

May 23, 2026

This work addresses the challenge of balancing plasticity and stability in large language models during knowledge updating. The authors discover that reasoning capabilities rely on the full spectrum of singular directions in multilayer perceptron weight matrices, rather than solely on principal components. Building on this insight, they propose a two-stage parameter-efficient fine-tuning framework: first, Singular Value Fine-tuning (SVF) identifies the critical subspace for reasoning; then, new knowledge is injected via LoRA under orthogonality constraints to minimize interference with existing reasoning abilities. Evaluated on Llama 3.1 8B and Mistral 7B, the method preserves 95% of original reasoning performance on average while maintaining strong factual recall, significantly outperforming existing spectral-domain PEFT approaches with a parameter overhead of less than 0.006%.

0 citationsRead paper

Learning Generative Dynamics with Soft Law Constraints: A McKean-Vlasov FBSDE Approach

May 09, 2026

This work addresses the problem of learning stochastic dynamics from endpoint and time-varying marginal distribution observations to generate continuous random trajectories that conform to prescribed marginal laws. The authors reformulate generative modeling as a McKean–Vlasov control problem, replacing hard interpolation or optimal transport constraints with soft energy penalties to enable learning of globally coupled mean-field dynamics. They establish a theoretical connection between the time derivative of the marginal law and a score-like function. Building on a forward–backward stochastic differential equation (FBSDE) framework and leveraging neural SDE solvers, the method accurately reconstructs target marginal distribution trajectories on low-dimensional benchmarks and generates coherent stochastic paths in high-dimensional settings—such as facial images and SMPL-H human motion data—that faithfully evolve according to observed marginal distributions.

0 citationsRead paper

Recent publications

Latest Papers

The Fundamental Structure of Risk: From Characteristics to Covariance

Jul 27, 2026

This study addresses the limitations of traditional risk models, which rely on high-frequency return data and suffer from noise-sensitive covariance estimates that poorly generalize to new assets. The authors propose a Characteristic-Driven Dynamic Factor Model (CD-DFM) that leverages low-frequency, observable firm characteristics—such as fundamental accounting variables—to learn interpretable factor exposures and forward-looking covariances through an end-to-end nonlinear latent factor representation. Notably, CD-DFM enables zero-shot embedding of entirely new assets using only their low-frequency features, while simultaneously preserving factor interpretability, achieving well-calibrated covariance forecasts, and maintaining economic plausibility. Empirical results on S&P 500 equities demonstrate that the model’s latent factors exhibit clear economic meaning, yield highly interpretable factor portfolios, and deliver competitive performance in covariance prediction.

0 citationsRead paper

PALoRA: Projection-Adaptive LoRA for Preserving Reasoning in Large Language Models

May 23, 2026

This work addresses the challenge of balancing plasticity and stability in large language models during knowledge updating. The authors discover that reasoning capabilities rely on the full spectrum of singular directions in multilayer perceptron weight matrices, rather than solely on principal components. Building on this insight, they propose a two-stage parameter-efficient fine-tuning framework: first, Singular Value Fine-tuning (SVF) identifies the critical subspace for reasoning; then, new knowledge is injected via LoRA under orthogonality constraints to minimize interference with existing reasoning abilities. Evaluated on Llama 3.1 8B and Mistral 7B, the method preserves 95% of original reasoning performance on average while maintaining strong factual recall, significantly outperforming existing spectral-domain PEFT approaches with a parameter overhead of less than 0.006%.

0 citationsRead paper

Learning Generative Dynamics with Soft Law Constraints: A McKean-Vlasov FBSDE Approach

May 09, 2026

This work addresses the problem of learning stochastic dynamics from endpoint and time-varying marginal distribution observations to generate continuous random trajectories that conform to prescribed marginal laws. The authors reformulate generative modeling as a McKean–Vlasov control problem, replacing hard interpolation or optimal transport constraints with soft energy penalties to enable learning of globally coupled mean-field dynamics. They establish a theoretical connection between the time derivative of the marginal law and a score-like function. Building on a forward–backward stochastic differential equation (FBSDE) framework and leveraging neural SDE solvers, the method accurately reconstructs target marginal distribution trajectories on low-dimensional benchmarks and generates coherent stochastic paths in high-dimensional settings—such as facial images and SMPL-H human motion data—that faithfully evolve according to observed marginal distributions.

0 citationsRead paper

Modeling dependency between operational risk losses and macroeconomic variables using Hidden Markov Models

Apr 23, 2026

This study addresses the challenges posed by the heterogeneity and time-varying structure of operational risk losses by proposing a multivariate hidden Markov model (HMM) that incorporates macroeconomic covariates. The approach extends the traditional HMM framework to integrate auxiliary variables and employs the EM algorithm to jointly model time series of multiple types of operational risk events. This formulation effectively captures the dynamic dependence between operational risk and the macroeconomic environment. Empirical results demonstrate that incorporating macroeconomic variables significantly enhances the model’s relevance and predictive accuracy under stress-testing scenarios, offering a more interpretable and practically useful modeling framework for operational risk management.

0 citationsRead paper

Empirical Evaluation of PDF Parsing and Chunking for Financial Question Answering with RAG

Apr 13, 2026

This study addresses the challenges posed by heterogeneous content—comprising text, tables, and images—in financial PDF documents for retrieval-augmented generation (RAG) systems, an area lacking systematic evaluation of parsing and chunking strategies. The work presents the first end-to-end empirical investigation of PDF processing pipelines in the context of financial document question answering. It introduces a new benchmark, TableQuest, integrates existing datasets, and systematically evaluates diverse PDF parsers and text chunking methods—including overlap mechanisms—on downstream QA performance. The findings reveal a strong correlation between preservation of document structure and answer accuracy. To support reproducibility and future research, the authors release their benchmark publicly and provide actionable, empirically grounded guidelines for designing effective RAG pipelines tailored to complex financial documents.

0 citationsRead paper