Trading at Light Speed: Designing Low Latency Systems in C++ - Lecture

Explore industry insights on designing low latency trading systems from scratch in this conference talk by David Gross, Auto-Trading Tech Lead at Optiver. Delve into the intricacies of translating CPU and hardware knowledge into C++ for high-speed trading. Learn strategies for utilizing multiple CPU cores, managing concurrency issues, and optimizing performance. Discover key concepts such as data modeling for performance, concurrent data handling in trading systems, and proper system tuning. Gain valuable knowledge on topics like C-state and P-state optimization, shared LLC optimization, and essential metrics for evaluating system performance. Understand why making a trading system "fast" must be a fundamental design consideration rather than an afterthought in this comprehensive exploration of low latency programming for high-stakes automated trading.

Syllabus

Intro
AUTOMATED TRADING A HIGH STAKES GAME
AUTOMATED TRADING: THE NEED FOR SPEED
DESIGN FOR PERFORMANCE
STRATEGY & TACTICS
HOW FAST IS FAST?
AN UNDERWHELMING PROFILING RESULT
DATA MODEL FOR PERFORMANCE
DATA MODEL: INSTRUMENT STORE
STABLE VECTOR
WSS ESTIMATION
CONCURRENT DATA IN TRADING SYSTEMS
HOW MUCH DATA?
SEQLOCK PROPERTIES
CONCURRENT DATA: EVENTS
SPMC QUEUE V2
IS YOUR SYSTEM TUNED CORRECTLY?
C-STATE, P-STATE
SHARED LLC OPTIMIZATION
METRICS
CONCLUSION