HFT Quant Developer
Builds the market-data handlers, order-book engines and simulators a high-frequency team runs on.
The job
An HFT quant developer builds the software a high-frequency, market-making or execution team runs on: market-data handlers, order-book engines, execution schedulers, and the simulators researchers use to decide whether a strategy works. A simulator that fills orders more often than the exchange would makes a losing strategy look profitable, and slow or non-deterministic code makes results impossible to trust. Things go wrong with optimistic fill models, latency, and replays that do not match live trading.
- Supports
- High-frequency trading, market-making and execution teams
- Stages
- 5 to learn, then proof of work
- Two routes
- Free resources, or SelfMadeQuant programmes and projects
The roadmap
The same stages either way. The free route uses university courses, lectures, books and papers that are free online. The SelfMadeQuant route maps each stage to our programmes and ends in projects on real market data.
Stage 01 · Foundations
Programming foundations
Clean, tested code and the data structures interviews ask about.
- Python
- Data structures and algorithms
- Testing and design
- 6.006 Introduction to Algorithms (opens in a new tab)
Erik Demaine, Jason Ku and Justin Solomon / MIT OpenCourseWare
Sorting, hashing, heaps and graph algorithms with complexity analysis.
- 6.0001 Introduction to Computer Science and Programming in Python (opens in a new tab)
MIT OpenCourseWare
Python programming from scratch, with computational thinking exercises.
- Algorithms, Part I (opens in a new tab)
Robert Sedgewick and Kevin Wayne / Princeton University on Coursera
Priority queues, symbol tables and balanced trees, the structures behind order books; free to audit.
Stage 02 · Core
Modern C++ and performance
The language of the hot path, and how to measure what is slow.
- Modern C++
- Memory and cache
- Profiling
- When a Microsecond Is an Eternity: High Performance Trading Systems in C++ (opens in a new tab)
Carl Cook / CppCon 2017
A trading firm's view of hot paths, cache warming and measuring latency in C++.
- Trading at Light Speed: Designing Low Latency Systems in C++ (opens in a new tab)
David Gross / Meeting C++ 2022
Data structure and memory layout choices for low-latency trading code.
- C++ Core Guidelines (opens in a new tab)
Bjarne Stroustrup and Herb Sutter (eds.) / Standard C++ Foundation
Rules for modern C++ ownership, resource management and safe performance.
- 6.172 Performance Engineering of Software Systems (opens in a new tab)
Charles Leiserson and Julian Shun / MIT OpenCourseWare
Profiling, cache behaviour, vectorisation and parallelism, with hands-on projects.
Stage 03 · Core
Systems and networking
What happens between the network card and your code.
- Operating systems
- Concurrency
- Networking
- Operating Systems: Three Easy Pieces (opens in a new tab)
Remzi and Andrea Arpaci-Dusseau / University of Wisconsin-Madison
Processes, virtual memory, locks and concurrency explained in short free chapters.
- Computer Networks: A Systems Approach (opens in a new tab)
Larry Peterson and Bruce Davie
How TCP, UDP, multicast and switching work, free from the authors.
- Beej's Guide to Network Programming (opens in a new tab)
Brian 'Beej' Hall
Practical TCP and UDP socket programming in C.
- 15-213 Introduction to Computer Systems (opens in a new tab)
Carnegie Mellon University
Machine-level code, the memory hierarchy and concurrency from the CS:APP course.
Stage 04 · Specialist
Market data and order books
Exchange feeds, book building and matching rules.
- Exchange feeds
- Order-book engines
- Queue position
- Nasdaq TotalView-ITCH 5.0 Specification (opens in a new tab)
Nasdaq
The add, execute and cancel message formats you parse to rebuild a full book.
- Limit Order Books (opens in a new tab)
Martin D. Gould, Mason A. Porter, Stacy Williams and others
Review of order-book mechanics, priority rules and empirical regularities.
- LOBSTER sample files (opens in a new tab)
LOBSTER / Humboldt University Berlin
Free reconstructed message and book files to test your matching and book engine against.
Stage 05 · Specialist
Execution and simulation
Schedules, replays and fill models a researcher can trust.
- Execution schedules
- Replay simulators
- Fill models
- Optimal Execution of Portfolio Transactions (opens in a new tab)
Robert Almgren and Neil Chriss
The classic trade-off between impact cost and timing risk that execution schedules encode.
- HftBacktest documentation (opens in a new tab)
nkaz001 (open source)
Tick-level backtester modelling queue position, latency and partial fills.
- ABIDES: Towards High-Fidelity Market Simulation for AI Research (opens in a new tab)
David Byrd, Maria Hybinette and Tucker Hybinette Balch
Design of an agent-based exchange simulator for testing strategies against reactive markets.
Stage 06 · Proof of work
An ITCH order-book engine and replay benchmark
Write a C++ parser for Nasdaq TotalView-ITCH 5.0 and an order-book engine, replay a full sample trading day and record per-message latency and throughput. It answers how fast a single-threaded book can process a day of market data and which data structures dominate the cost.
Stage 01 · Foundations
Programming foundations
Clean, tested code and the data structures interviews ask about.
- Python
- Data structures and algorithms
- Testing and design
- Python and Data Primer
- Python Foundations
- Professional Workflow
- Python for Quantitative Finance
- Object-Oriented Programming
- Testing, Debugging, and Style
- Model and software design
Stage 02 · Core
Modern C++ and performance
The language of the hot path, and how to measure what is slow.
- Modern C++
- Memory and cache
- Profiling
- Modern C++ for Quantitative Finance
Lessons are not published yet. Its programme page lists intended outcomes and free preparation resources.
Stage 03 · Core
Systems and networking
What happens between the network card and your code.
- Operating systems
- Concurrency
- Networking
- Trading Systems and Low-Latency Infrastructure
Lessons are not published yet. Its programme page lists intended outcomes and free preparation resources.
- Quantitative Data Engineering
Lessons are not published yet. Its programme page lists intended outcomes and free preparation resources.
Stage 04 · Specialist
Market data and order books
Exchange feeds, book building and matching rules.
- Exchange feeds
- Order-book engines
- Queue position
- Market Microstructure and Execution
- Market Data and Trading State
- Orders and Execution Schedules
Stage 05 · Specialist
Execution and simulation
Schedules, replays and fill models a researcher can trust.
- Execution schedules
- Replay simulators
- Fill models
- Market Microstructure and Execution
- Post-Trade Analysis and Monitoring
- Quantitative Research, Data and Backtesting
- Economic Backtest Simulation
- Linear Algebra, Optimisation and Numerical Methods
- Optimisation
Stage 06 · Proof of work
Build the projects
Each project is built on real market data and reports what the data said, including when the idea failed.
- Q8Order-Book and Slippage CalculatorA pre-trade slippage calculator on Coinbase BTC-USD level-2 snapshots: a 0.5 BTC market order walks 12 levels for 0.66 bp.
- EXExecution replay and transaction-cost diagnosticsFor orders up to 10 BTC, sending at once cost 0.56–3.36 bp with far less variance than TWAP.
- LSQueue-Position and Latency-Aware Order-Book Simulator (C++)A C++ order-level book engine that scores 14 passive-fill models against exact queue position on 1.87 million virtual orders.
Practise the interview
Questions this role gets asked, each with a hint and a worked solution.
Start the SelfMadeQuant route free: a free account opens one complete topic from each primer and the interview question bank. No card needed.
Start free