ENGINE
systems project // current WIP
Order Book (Under Construction)
I am building a high-performance electronic limit order book in modern C++ because apparently my idea of a relaxing side project is asking, "what if a queue had opinions about fairness, latency, and the memory hierarchy?"
This is explicitly a current work in progress. It is not done, not polished, and not pretending to be a production exchange. The goal is to understand the infrastructure underneath modern markets: matching orders, preserving price-time priority, and designing data structures that do not make the CPU file a formal complaint.
As it grows, I am planning to add benchmarking, profiling, unit tests, market data feeds, replay tools, and progressively more performance-focused experiments. Less "hot stock tip," more "let me build the tiny engine room where the market plumbing lives."
[DONE] boot price-time priority core
[DONE] teach limit orders to stand in line politely
[ACTIVE] wire cancel/modify paths without angering the cache
[QUEUED] benchmark the spicy parts
[QUEUED] add replay tools and market data feeds
[QUEUED] stare at perf traces until enlightenment happens
awaiting next commit
| Bid Size | Bid | Ask | Ask Size |
|---|---|---|---|
| 420 | 101.24 | 101.25 | 180 |
| 310 | 101.23 | 101.26 | 260 |
| 780 | 101.22 | 101.27 | 640 |
| 150 | 101.21 | 101.28 | 390 |
| 900 | 101.20 | 101.29 | 710 |
The resting book holds queued buy bids and sell asks. The demo introduces a buy priced high enough to trade with the cheapest asks.
SYSTEMS
Latency-minded C++
This is less get-rich-quick and more why-is-this-pointer-chasing-making-my-CPU-sad: memory layout, cache behavior, and low-latency design are the fun bits.TOOLING
Benchmarks incoming
The project will grow into unit tests, profiling, benchmarks, replay tooling, and performance experiments that make the engine easier to trust.WHY