Skip to content
View donghui-0126's full-sized avatar

Highlights

  • Pro

Block or report donghui-0126

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this userโ€™s behavior. Learn more about reporting abuse.

Report abuse
donghui-0126/README.md

Hi, I'm Donghui ๐Ÿ‘‹

Crypto Quant Researcher based in Seoul, Korea.

Julia๋กœ ์•ŒํŒŒ ๋ฆฌ์„œ์น˜ ํŒŒ์ดํ”„๋ผ์ธ์„ ๋งŒ๋“ค๊ณ , ์‹œ๊ทธ๋„์„ ๋ฐœ๊ตดํ•˜๊ณ , ๋ฐฑํ…Œ์ŠคํŠธํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค. ๋น—์ธ ์ˆ˜์ˆ˜๋ฃŒ 0bp ๊ธฐ๊ฐ„์— ๋ˆ„์  ๊ฑฐ๋ž˜๋Œ€๊ธˆ 10์–ต์› ์ด์ƒ์„ ๋‹ฌ์„ฑํ•œ ๊ฒฝํ—˜์ด ์žˆ์Šต๋‹ˆ๋‹ค.

What I Do

  • Alpha Research - Cross-sectional alpha screening with HAC t-stats, triple-barrier labeling, signal path analysis
  • Backtesting - Vectorized cross-sectional backtester, rolling ML models, decay-based alpha modeling
  • Data Pipeline - Julia-first parallel computation, ClickHouse/Parquet data layer, Streamlit dashboards
  • Market Microstructure - Limit order book modeling, online estimation, HFT research

Highlights

ํ†ต๊ณ„ ๊ธฐ๋ฐ˜ ๋ฆฌ์„œ์น˜

  • Newey-West HAC ๊ธฐ๋ฐ˜ alpha t-stat screening์œผ๋กœ spurious alpha ํ•„ํ„ฐ๋ง
  • Triple-barrier labeling + signal path analysis๋กœ ์‹œ๊ทธ๋„์˜ ์‹œ๊ฐ„์  ๊ตฌ์กฐ ๋ถ„์„
  • Rolling quantile smoothing, cross-sectional rank/percentile normalization ๋“ฑ ๋‹ค์–‘ํ•œ ์•„์ด๋””์–ด ๊ธฐ๋ฐ˜ ๋ถ„์„ ํŒŒ์ดํ”„๋ผ์ธ ๊ตฌ์ถ•

Julia ์—”์ง€๋‹ˆ์–ด๋ง

  • GC๊ฐ€ ์žˆ๋Š” ์–ธ์–ด์—์„œ์˜ ์„ฑ๋Šฅ ์ตœ์ ํ™” ๊ฒฝํ—˜ - @view, @inbounds, pre-allocation, ์‹ฌ๋ณผ๋ณ„ ๋ฐฐ์น˜ ์ฒ˜๋ฆฌ ํ›„ GC.gc() ํ˜ธ์ถœ ๋“ฑ ๋ฉ”๋ชจ๋ฆฌ ๊ด€๋ฆฌ ํŒจํ„ด ์ ์šฉ
  • Julia ํŒจํ‚ค์ง€ ํ•œ๊ณ„๋ฅผ DLL(ccall)๋กœ ์šฐํšŒํ•˜์—ฌ ๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•œ ๊ฒฝํ—˜ โ€” ์ดํ›„ ์ด ๊ฒฝํ—˜์ด OMS-v2 Rust ๋งˆ์ด๊ทธ๋ ˆ์ด์…˜์˜ ๋™๊ธฐ๊ฐ€ ๋จ
  • Julia ๋ฉ€ํ‹ฐ์Šค๋ ˆ๋”ฉ(Threads.@threads, @spawn) ๊ธฐ๋ฐ˜ ๋ณ‘๋ ฌ ์—ฐ์‚ฐ ํŒŒ์ดํ”„๋ผ์ธ ๊ตฌ์ถ•

Rust ์‹œ์Šคํ…œ ์—”์ง€๋‹ˆ์–ด๋ง

  • Julia + Rust DLL ํ•˜์ด๋ธŒ๋ฆฌ๋“œ OMS๋ฅผ ์ˆœ์ˆ˜ Rust๋กœ ๋งˆ์ด๊ทธ๋ ˆ์ด์…˜ (FFI ๊ฒฝ๊ณ„ ์ œ๊ฑฐ, GC ์••๋ฐ• ํ•ด์†Œ)
  • 27๊ฐœ ํฌ๋ ˆ์ดํŠธ Cargo workspace, 211๊ฐœ ํ…Œ์ŠคํŠธ, 0 warnings
  • 11๊ฐœ ๊ฑฐ๋ž˜์†Œ WebSocket/REST ํ†ตํ•ฉ (๋น—์ธ, ์—…๋น„ํŠธ, ์ฝ”์ธ์›, ๋ฐ”์ด๋‚ธ์Šค, ๋ฐ”์ด๋น„ํŠธ, OKX, ๋น„ํŠธ๊ฒŸ ๋“ฑ)
  • ๊ฑฐ๋ž˜์†Œ๋ณ„ ์ƒ์ดํ•œ ์ธ์ฆ ์ฒด๊ณ„ ๊ตฌํ˜„ (HS256/HS512 JWT, HMAC-SHA256/SHA512 ํ—ค๋”, ๋ฆฌ์Šจ ํ‚ค)
  • Tick-to-trade pipeline: p50 4.23us, 218K tps (117๊ฐœ ์˜จ๋ผ์ธ ํ”ผ์ณ ์—ฐ์‚ฐ ํฌํ•จ)
  • zero-alloc ์ฃผ๋ฌธ ์ƒํƒœ๋จธ์‹  (~50๊ฐœ ์ „์ด), simd-json ๊ณ ์† ํŒŒ์‹ฑ, DoubleBuffer lock-free read

Claude Code ๊ธฐ๋ฐ˜ ๊ฐœ๋ฐœ

  • Claude Code(AI agent)๋ฅผ ํ™œ์šฉํ•œ ๋ฐ์ดํ„ฐ ๋ถ„์„ ๋ฐ ๋ฆฌ์„œ์น˜ ์ธํ”„๋ผ ๊ตฌ์ถ•
  • AI ์—์ด์ „ํŠธ ๊ธฐ๋ฐ˜ EDA ์ž๋™ํ™”, ์•ŒํŒŒ ๋ฆฌ์„œ์น˜ ํŒŒ์ดํ”„๋ผ์ธ ์„ค๊ณ„, Streamlit ๋Œ€์‹œ๋ณด๋“œ ๊ฐœ๋ฐœ
  • ๋ฉ€ํ‹ฐ ์—์ด์ „ํŠธ ์˜ค์ผ€์ŠคํŠธ๋ ˆ์ด์…˜์œผ๋กœ ๋ถ„์„-๊ตฌํ˜„-๊ฒ€์ฆ ์›Œํฌํ”Œ๋กœ์šฐ ์ž๋™ํ™”

์‹ค์ „ ํŠธ๋ ˆ์ด๋”ฉ

  • ๋น—์ธ ์ˆ˜์ˆ˜๋ฃŒ 0bp ๊ธฐ๊ฐ„ ๋ˆ„์  ๊ฑฐ๋ž˜๋Œ€๊ธˆ 10์–ต์›+ ๋‹ฌ์„ฑ
  • Multi-exchange live trading system ์šด์˜ (Binance, Upbit, Bithumb, Coinone)

Tech Stack

Core: Julia (alpha research), Rust (execution infra), Python (visualization)

Data: Polars, Parquet, ClickHouse, Arrow, DataFrames.jl

Analysis: Newey-West HAC, rolling quantiles, EMA, cross-sectional rank/percentile normalization, triple-barrier

Infra: tokio, WSL, tmux, Streamlit, Claude Code

Current Projects

Project Description Stack
amuredo-EDA Alpha research & backtest tooling - signal screening, path analysis, Streamlit dashboard Julia, Python
amuredo-alphago Alpha factory platform - 1D/2D alpha screening, cross-sectional analysis, rolling ML Julia, Python
amuredo-StrategyStore Strategy storage & management Julia
amuredo-OMS-v2 Pure Rust OMS - 27 crates, 11 exchanges, p50 4.23us tick-to-trade Rust, Private
AMuReDoTrade Live trading system - multi-exchange execution (Binance, Upbit, Bithumb, Coinone) Private

Research Projects

Project Topic
ML-HFT High-frequency trading framework with machine learning for futures
deepOBs Short-term predictability in order book markets via deep learning
lob-world-models Model-based RL for LOB prediction
lob-deep-learning Deep learning models for limit order book
online_estimation Online rolling estimation (mean, weighted mean, std, skew)
JaneStreetKaggle Jane Street real-time market data forecasting
Machine-Learning-for-Factor-Investing ML techniques for factor investing
crypto-scalping-RL-Agent RL-based crypto scalping agent with custom chart env

Awards

Year Competition Result
2025 Jane Street - Real-Time Market Data Forecasting (Kaggle) Bronze Medal
2024 BDA X ASCEND - Bitcoin Volatility Prediction ์šฐ์ˆ˜์ƒ
2023 WorldQuant Alphaton Korea 3rd Place
2023 FSI Data Challenge - EV Customer Prediction ์šฐ์ˆ˜์ƒ

Research & Experience

Publication

  • Hyeong-jin Son, Lim Donhui, & Young-woo Han. (2023). Reinforcement learning portfolio optimization based on portfolio theory. ํ•œ๊ตญ์ •๋ณด์ฒ˜๋ฆฌํ•™ํšŒ ํ•™์ˆ ๋Œ€ํšŒ๋…ผ๋ฌธ์ง‘, 30(2), 961-962.

Experience

  • KHU AIMS LABS - Undergraduate Researcher (2024) : Combinatorial Optimization with RL (TSP, VRP)
  • ๋ชจ๋‘์˜ ์—ฐ๊ตฌ์†Œ PISTAR LAB (2024) : Algorithm Trading with AI
  • KHUDA 3-4๊ธฐ Financial Track (2023) : ๊ฒฝํฌ๋Œ€ ๋ฐ์ดํ„ฐ๋ถ„์„/AI ๋™์•„๋ฆฌ
  • ๊ธˆ์œต๊ณตํ•™ ํ•™ํšŒ UFEA 36๊ธฐ

Past Projects

2022-2024 ํ”„๋กœ์ ํŠธ ์•„์นด์ด๋ธŒ
  • Causal Inference: Factor of Happiness (2024)
  • Sentiment Analysis by BERT (2023)
  • Virtual Trading Based on TA (2023)
  • Predict Resell Price of Shoes (2023)
  • Dacon: Lettuce Growth Forecast AI (2022)
  • What is important Stats in NBA? (2022)

Code Archive: Algorithm | ML/DL | RL | Data Structure | Rust | Julia


๐Ÿ“ซ lukedonghui@gmail.com

Pinned Loading

  1. crypto-scalping-RL-Agent crypto-scalping-RL-Agent Public

    Jupyter Notebook 1

  2. mini-project mini-project Public

    ๋Œ€๋‹จํ•˜์ง„ ์•Š์ง€๋งŒ ์†Œ์†Œํ•˜๊ณ  ์žฌ๋ฐŒ๋Š” ๊ฐœ์ธ/ํŒ€ ํ”„๋กœ์ ํŠธ

    Jupyter Notebook 2

  3. Reinforce-Learning Reinforce-Learning Public

    Jupyter Notebook

  4. machine-learning machine-learning Public

    ๋จธ์‹ ๋Ÿฌ๋‹ ๊ณต๋ถ€ ๊ธฐ๋ก์žฅ

    Jupyter Notebook

  5. Machine-Learning-for-Factor-Investing Machine-Learning-for-Factor-Investing Public

    Jupyter Notebook 1

  6. Gym-Trading-Env Gym-Trading-Env Public

    Forked from ClementPerroud/Gym-Trading-Env

    A simple, easy, customizable Gymnasium environment for trading.

    Python 1