DEVOURED
Software engineering at a proprietary trading company: Optiver
Proprietary trading firms like Optiver operate in a unique engineering niche where nanosecond latency constraints necessitate custom hardware and full-stack ownership.
What: Optiver employs nearly 1,000 engineers to maintain high-frequency trading (HFT) systems. The firm uses a three-layer stack—signals, strategy, and execution—and customizes everything from Linux kernels and Postgres to proprietary FPGAs to shave nanoseconds off trade processing.
Why it matters: Proprietary trading firms are increasingly shifting focus from raw speed to AI-driven model efficiency, making them major employers of high-end infrastructure and machine learning talent.
Deep dive
- Trading firms treat the platform as the product, spending 30-40% of engineering headcount on platform work.
- Latency is the primary engineering constraint, often requiring custom hardware and kernel-level optimizations.
- Optiver uses a 'retreat' system to update thousands of option prices in nanoseconds after a trade occurs.
- Software engineers at these firms manage full-stack responsibilities, from FPGA RTL design to high-level trading strategy code.
- The business is shifting from pure low-latency arbitrage to AI-based quantitative modeling.
Decoder
- HFT (High-Frequency Trading): Trading strategies that use powerful hardware and algorithms to transact a large number of orders at extremely high speeds.
- FPGA (Field-Programmable Gate Array): Integrated circuits designed to be configured by a customer or designer after manufacturing, often used for hardware-level acceleration.
- Market Maker: A firm that provides liquidity to markets by constantly offering to buy and sell securities at specified prices.
Original article
Full article content is not available for inline reading.
Explore all 58 articles in today's digest
Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi