Odd Lots

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

March 2, 2026

Key Takeaways Copied to clipboard!

  • The evolution of trading speed has moved from human-perceptible timeframes (tenths of a second) to machine-centered trading in the nanosecond regime, fundamentally changing trading strategies and required skills. 
  • High-frequency trading (HFT) firms often outperform large banks due to their flatter organizational structures, which allow for faster adoption of new technology, such as acquiring faster servers immediately. 
  • The core of the HFT speed race involves exploiting structural market features, such as the relationship between futures and underlying equities, leading to a continuous competition between 'liquidity makers' and 'liquidity takers' within the electronic order book. 

Segments

Introduction and Guest Background
Copied to clipboard!
(00:01:54)
  • Key Takeaway: Professor Donald Mackenzie studies the intersection of finance and technology, including quantitative finance models and high-frequency trading.
  • Summary: The episode introduces Professor Donald Mackenzie, a sociologist of technology whose work spans from nuclear missile guidance to digital advertising and AI. His book, Trading at the Speed of Light, focuses on high-frequency trading. Mackenzie employs a qualitative, historical approach by interviewing industry participants rather than using quantitative social science methods.
HFT Aesthetics and Culture
Copied to clipboard!
(00:10:36)
  • Key Takeaway: HFT firms adopt a Silicon Valley aesthetic, contrasting sharply with the traditional investment bank or trading floor image, signaling a cultural shift toward technology-centric operations.
  • Summary: The modern HFT office environment, characterized by casual dress and modern tech setups, reflects a cultural alignment with tech startups rather than traditional finance. This contrasts with the formal attire required at investment banks where Mackenzie previously conducted research. The shift indicates that coding and technical expertise now define the core identity of these trading operations.
The Island ECN and HFT Genesis
Copied to clipboard!
(00:13:05)
  • Key Takeaway: The electronic communications network (ECN) Island was crucial for HFT because its matching engine executed trades in two milliseconds, a thousand-fold improvement over previous systems.
  • Summary: Island operated using an electronic order book managed by a matching engine, eliminating direct human negotiation. Its speed advantage over predecessors like Instanet created the perfect environment for automated trading systems to thrive. This technological shift spurred established exchanges to modernize to remain competitive, creating today’s electronic market structure.
Quantifying Speed: Milliseconds to Nanoseconds
Copied to clipboard!
(00:19:07)
  • Key Takeaway: The speed of trading has progressed from milliseconds to nanoseconds, where light traveling one foot (30 centimeters) takes one nanosecond, marking the physical limits of speed competition.
  • Summary: The human perception threshold for time is about one-tenth of a second, meaning HFT operates entirely beyond human reaction time. The race progressed from milliseconds (thousands of a second) to microseconds (millions of a second) and eventually to nanoseconds (billionths of a second) by the late 2010s. This progression signifies a shift from human-centered to machine-centered trading decisions.
Bank vs. HFT Organizational Speed
Copied to clipboard!
(00:24:18)
  • Key Takeaway: HFT firms’ small, founder-run structures allow them to implement technological upgrades, like buying new servers, in days, whereas large banks face months of bureaucratic delays.
  • Summary: Banks struggle to compete in speed races because their IT departments are separate from trading functions, requiring extensive management sign-off for hardware and software changes. HFT firms, often privately owned by founders, can bypass this bureaucracy, sometimes even using personal credit cards for immediate purchases to gain a competitive edge.
The Physical Speed Arms Race
Copied to clipboard!
(00:29:18)
  • Key Takeaway: The speed race evolved from co-locating servers within the same data center (enforced by equal cable length rules) to optimizing cross-city links using microwave technology.
  • Summary: After exchanges like NASDAQ and NYSE adopted electronic systems around 2005, HFT firms began physically moving machines closer to matching engines. Exchanges now enforce equal cable lengths via coiled fiber optics to neutralize proximity advantages within a single data center. The race then shifted to inter-data center links, where microwave transmission became superior to fiber optics because light travels faster through the atmosphere than through glass cable.
HFT Firm Structure and Competition
Copied to clipboard!
(00:37:41)
  • Key Takeaway: HFT firms often operate using their own capital, reinvesting profits, and are internally structured either as competing desks or as unified entities sharing infrastructure.
  • Summary: HFT firms typically trade proprietary capital, often started by successful floor traders using their accumulated wealth, avoiding the limited partner structure of hedge funds. Competition exists between firms, and sometimes internally, where trading desks compete for resources like bandwidth based on profit and loss performance. The dynamic between liquidity-providing market makers and liquidity-taking firms drives the need for speed, especially when reacting to price discrepancies between related instruments like futures and equities.
Market Efficiency and Financial Sector Pay
Copied to clipboard!
(00:48:57)
  • Key Takeaway: Despite decades of technological advancement, the unit cost of financial intermediation has not significantly declined since the 1880s, largely because efficiency gains are captured as high compensation within the financial sector.
  • Summary: Research by Thomas Philippon indicates that the cost of financial intermediation has remained relatively flat over a century, despite technological progress. This is attributed to the massive increase in compensation for finance professionals since the 1970s, offsetting potential cost savings for investors. While retail fund fees have dropped, high fees in areas like private equity balance the overall cost metric.
AI Scaling and Diminishing Returns
Copied to clipboard!
(00:52:03)
  • Key Takeaway: The massive investment in AI infrastructure follows a diminishing returns curve, raising the critical question of when the enormous resource expenditure will cease to yield proportional intelligence gains.
  • Summary: AI effectiveness is currently tied to scaling laws—increasing network size, data, and parameters—but this relationship follows a logarithmic function demonstrating diminishing returns. The core sociological question is determining the economic and environmental stopping point on this curve, especially when each incremental improvement requires trillions of dollars and significant carbon emissions. This mirrors the HFT dynamic where firms must continue investing to avoid falling behind, even if the marginal profit decreases.