Blog
Insights on software development, high-frequency trading, system architecture, and more.
I Used My Gaming GPU to Revolutionize Trading Strategy
Achieving a 695× throughput boost with an NVIDIA RTX 3090 and CuPy
Letting Strategies Evolve Instead of Writing Them: My Experiments With Evolutionary Search
The strategies that survived weren’t the ones I expected — they were the ones I never would have designed.
The Architecture of a Modern Quant Research Platform
If I were rebuilding my research infra from scratch today, I’d unify three things: reproducibility, latency monitoring, and quant collaboration. Everything else is secondary.
Integrating Reinforcement Learning for Adaptive Market-Making
RL isn’t magic. It’s fragile. But if tuned right, it learns where not to trade, which is where most edges live.
Designing My Own Backtesting Framework for Multi-Asset Portfolios
I realized I didn’t trust third-party backtesters anymore. So I built one. Vectorized, reproducible, and benchmarked against Alpaca-like APIs.
Redis Streams for Real-Time Pipeline Coordination
Kafka was overkill for my real-time pipelines. I needed millisecond coordination between stages without unnecessary operational overhead. Redis Streams hit the sweet spot.
Why I Switched from Python to C++ for My Execution Engine
Python got me started. But for sub-millisecond execution, the interpreter became my enemy. Rewriting my core in C++ dropped 25ms off each roundtrip.
When Network Latency Becomes a Trading Strategy
The fastest code isn’t always the smartest. I learned that anticipating where latency will spike—across gateways, venues, or even inside your own NIC queues.
Experimenting with Simulated Order Books
I wanted to simulate real-time liquidity conditions. I ended up writing my own order-matching engine in 300 lines of Python just to understand the microstructure better
Optimizing Tick Data Storage: From SQLite to Parquet
Migrating my tick data storage from SQLite to Parquet changed everything. What started as a harmless 50 GB SQLite file quickly became slow, unsearchable, and nearly impossible to maintain.
Building My First Latency-Sensitive Trading Bot in Python
When I started experimenting with live market feeds, I didn’t fully appreciate how much latency matters. My first trading bot used asyncio in Python.