Real-time crypto market observation engine focused on liquidity, orderflow, and derivatives-based market behavior. / 기관의 유동성 행동을 관측하는 실시간 마켓 마이크로스트럭처 엔진. 가격 예측이 아닌, 시장 구조를 읽기 위한 도구.

Low latency broker-agnostic trade execution engine with modular adapters and retry handling (FastAPI, Python)

VPIN — Volume-Synchronized Probability of Informed Trading (Easley, Lopez de Prado, O'Hara 2012)

High-performance C++ limit order book engine built for quantitative research, with microsecond-level matching, realistic order types, and strict price-time priority.

Real-time quantitative engine for L2 Limit Order Book reconstruction and liquidity analysis. Implements non-blocking async pipelines to synchronize REST snapshots with high-frequency WebSocket deltas. Engineered to detect hidden iceberg liquidity and spoofing patterns, utilizing 1D-CNNs for real-time stress forecasting in fragmented crypto markets.

Code, thesis source, and visual summary for Price Impact on Uniswap V3

Detecting structural changes in market microstructure around the 2008 TARP intervention using intraday ETF data. XGBoost classifier achieves 0.952 AUC-ROC distinguishing pre/post-TARP trading days via engineered volatility, volume profile, and cross-asset correlation features.

Formally verified comparison of exchange designs: who gets exploited and why

Python client library for Aperiodic.io — institutional-grade market microstructure, liquidity and order flow metrics with full exchange universe coverage. Turn flow dynamics into alpha in hours, not months. No tick infrastructure to build or maintain.

Autonomous research instrument for cross-silo financial inference. Pre-registered corpus (12,030 facts, 171 causal edges), public prediction ledger, 12-week empirical study Jun–Aug 2026.

📈 Build and analyze exchange-grade order book matching, market data replay, and microstructure analytics with modern C++20.

Latency-aware limit order book simulator with Avellaneda–Stoikov market making strategy and experiments on latency vs profitability.

GRU neural network for predicting short-term price direction from high-frequency order book data (Argentine bonds). Full Python pipeline + C++ inference engine via libtorch. 63.8% directional accuracy.

Python-based paper trading research framework for systematic strategies, market making, and risk-controlled execution using live Binance data.

BTC Perpetual Futures Microstructure Analysis — Order flow modeling, domain scoring system, and Walk-Forward Validation

Early warning for BTC/ETH flash crashes using trade-only features (Binance 2021–2024). XGBoost; strict temporal splits; zero median false alerts/day on quiet periods.

Jane Street portfolio: OCaml, Coq, boundary-aware modelling, formal verification

PhD researcher — ML for financial market security | GNNs, LLMs, options microstructure | IEEE BigData 2025 | AIAI 2026 accepted | JRFM under review

Central Limit Order Book (CLOB) matching engine with price-time priority, performance benchmarking, market simulation, and REST API. Java-based HFT system for quantitative finance.