The engineer's path into high-frequency trading

Hands-on training for the engineers who build and run trading systems. Every lesson is code you write, every drill runs against real order flow, and every concept is measured — in nanoseconds, ticks and fills.

Everything an HFT stack demands of you

Latency18 lessons · live bookp99 4.2 µs
pricesize
4310.5515.2
4310.778.0
4310.9622.8
4311.183.1
4311.4212.4
spread 0.18mid 4310.46tick 47 ms
4310.3718.6
4310.215.4
4310.0227.1
4309.869.8
4309.6114.3
focus: tick-to-tradefeed live

The whole training loop, in one place

Interactive lessons

Every concept lands as a short block with a question on the spot — the lesson does not move on until you have used it.

Tests with instant feedback

Each answer is dissected immediately: why it is right, where the trap was, and what the interview version of it asks.

Drills against real flow

Practice runs on recorded order books and latency budgets — you patch queues, size batches and watch the clock.

Flashcards that stick

Spaced repetition across the whole curriculum, so the vocabulary survives long past the lesson that introduced it.

Progress you can read

Every track, lesson and drill feeds one line: where you stand, what is next and what is still weak.

A diagnostic that places you

Twenty questions map what you know onto the curriculum, so you start where it hurts — not at zero.

0
lessons
0
drills
0
flashcards
0
tracks

The path, in three moves

01

Take the diagnostic

Twenty questions, fifteen minutes. It maps what you know and builds your track from the gaps.

02

Walk your track

Lessons, tests and flashcards in one line — each block unlocks the next, and progress saves itself.

03

Drill and review

Practice runs against recorded order flow; every drill is scored, and every weak spot loops back as a card.

Whole skills, not just notes

Tracks
Latency
Microstructure
Networking
Kernel bypass
Colocation
Systems
Limit order books14 / 18 lessons
✓Tick data: what a feed actually says
✓Queue position and the matching engine
▸Hidden liquidity and iceberg orders
·Adverse selection, measured
·From UTP to your book: the pipeline

Questions, answered

No. The diagnostic places you, and the C++ track starts from the parts that matter for latency — allocation, cache lines, templates — assuming you can read code, not write lock-free queues on day one.

breakHFT

From zero toHFT engineer