Quantitative research · Market microstructure · Statistical inference

I build probabilistic models and research systems for decisions under uncertainty: from stochastic neural dynamics to market making and execution. I am a computational neuroscience Ph.D. researcher at Albert Einstein College of Medicine, with a minor in applied mathematics and statistics and a B.S. in mathematics.

My primary focus is quantitative research and trading. I also work on causal experiments in machine learning and practical research tools.

Quant projects · Research · Merged contributions · Email

A Python market maker combining rate-transition models, conditional return distributions, and cross-company residual covariance. Estimated probabilities become quotes and sizes subject to uncertainty, inventory, and capital limits.

Akuna Challenge 2026: 15.70/16 strategy points · 20/20 evaluation cases passed · zero bankruptcies. These are captured simulation results, not a competition ranking or live return. The public case study covers the modeling approach and results without disclosing challenge materials or submission code.

An open-source research scaffold for central limit order books: deterministic Level-2 replay, features available at decision time, future-midpoint labels, purged walk-forward evaluation, and negative controls. A visible-depth cost sweep makes execution assumptions explicit.

What it demonstrates: data causality, reproducible event reconstruction, and cost-aware evaluation. The shipped sample is synthetic; it establishes pipeline behavior rather than trading profitability.

I develop and operate a prediction-market research and execution system, with settlement-based evaluation, inventory-aware market making, and post-trade attribution. My research asks whether an apparent entry signal survives uncertainty analysis, and separates trading outcomes from venue incentives.

I have also traded SK Hynix relative value across Korean shares, USD-settled perpetual futures, and the U.S. ADR, studying funding, hedge construction, convergence, and financing constraints. Trading code and account records remain private.

My research centers on stochastic processes, latent-state inference, and likelihood-based estimation. In Ruben Coen-Cagli's lab, I develop continuous-time normalization models with stochastic input and volatility, derive moment approximations, and fit response dynamics to neural data.

• Probabilistic segmentation & neural dynamics — public research code connecting natural-image structure, inferred uncertainty, and early visual cortical dynamics.
• Bayesian observer models — efficient sensory encoding, Bayesian inference, and perception-to-action mapping; Python demo and original MATLAB code from my perception research with Alan Stocker at UPenn.

Selected publications: co-first author, PLOS Computational Biology (2025), on response-range-dependent heading biases; first author, Journal of Vision (2022), on serial dependence in heading perception.

Four merged external upstream PRs across prediction-market infrastructure and node tooling. Status checked September 7, 2026.

Mechanistic interpretability lab — causal interventions in GPT-2-small, matched controls, and explicit decision rules. The public record includes null and inconclusive results alongside a bounded causal-subspace study; compact evidence is available, while full reruns require additional source artifacts.

Tools I build for research workflows: taskdone-runner for background-task notifications and review tracking; HTML report skill for readable reports with structured evidence; Codex Quota Bar for a native macOS view of subscription quotas (interactive preview).

• September 2026: published taskdone-runner and added Codex Quota Bar to the public tool portfolio.
• August 2026: published the quant trading challenge case study and expanded the causal-subspace research record.
• June–July 2026: four upstream PRs merged into PMXT and Filecoin Lotus.

Core tools: Python · NumPy / SciPy / pandas · SQL · PyTorch · MATLAB / R
Engineering contributions: TypeScript · Go
Contact: linghaoxu11@gmail.com · Ph.D. expected December 2027