How the Speed of a Trade Got Down to Nearly the Speed of Light

What It Covers

Sociologist Donald McKenzie, author of *Trading at the Speed of Light*, traces how electronic equity markets evolved from two-second trade execution in the late 1990s to nanosecond-speed transactions today, explaining the physical infrastructure arms race, maker-taker dynamics, and why banks consistently lost to independent HFT firms.

Key Questions Answered

• •Electronic order book mechanics: Every equity trade routes through a matching engine that pairs bids and offers at identical prices. If no match exists, the order rests in the book until canceled or matched. Island's matching engine in the late 1990s executed trades in two milliseconds — a thousand times faster than its closest competitor, Instinet, opening the door for automated trading strategies.
• •Maker-taker speed race: Market-making HFT firms populate order books with resting bids and offers. When Chicago futures prices shift, those orders become stale within nanoseconds. Taking firms race to execute against stale orders while makers race to cancel them. This structural dynamic — not algorithmic cleverness — is the core engine driving the continuous speed arms race between competing HFT firms.
• •Physical infrastructure determines profit: The Chicago-to-New-Jersey signal path evolved through three phases: stitched fiber routes (Getco's "gold line"), a dedicated drilled cable (Spread Networks, 2010), then microwave links. Microwave transmits at roughly the speed of light in a vacuum versus two-thirds that speed through fiber optic glass, delivering a measurable latency edge worth single-digit billions of dollars annually across the industry.
• •Why HFT firms beat banks structurally: Independent HFT firms average 50–150 employees with flat hierarchies and founder ownership. Early firms could purchase a new server on a personal credit card and deploy it within ten days. Banks required six-month procurement cycles and separate IT approval chains. Founder-controlled risk management also produced fewer catastrophic blowups than expected given the complexity of the systems involved.
• •Diminishing returns govern both HFT and AI investment: Economist Eric Budish quantified exploitable structural arbitrage in equity markets at single-digit billions annually, creating a natural ceiling on speed investment. Sam Altman's own framing of AI intelligence scaling as a logarithmic function of compute resources mathematically confirms diminishing returns — meaning each capability increment requires exponentially greater capital and energy expenditure to achieve.