Trading at the Speed of Light

Preamble

The provided text serves as the bibliographic front matter and table of contents for Trading at the Speed of Light by Donald MacKenzie. It outlines the publication details, copyright information, and structural layout of the book. The contents detail a multi-chapter investigation into the mechanics of high-frequency trading, market infrastructure, algorithmic interaction, and the socio-political complexities governing modern financial exchanges.

The author expresses gratitude to anonymous interviewees, research assistants, and funding bodies (E.S.R.C, E.R.C) for their support. Acknowledgments also extend to Princeton University Press for editorial guidance and colleagues for intellectual contributions. Finally, the author notes the book's accessible style and reliance on previously published academic research.

Trading at the Speed of Light Introduction (Part 1)

The transformation of global finance in the late 20th and early 21st centuries is exemplified by the evolution of 50 Broad Street in Manhattan. Once a neglected office building, it became the home of "Island," an electronic venue launched in 1996 that revolutionized trading. Unlike its predecessors, Island utilized inexpensive, blazingly fast computer hardware in its basement to match bids and offers in milliseconds. Crucially, Island was "machine-friendly," allowing automated systems to interact directly with its electronic order books, effectively ending the era where financial markets were designed exclusively for human participants.

This shift catalyzed the rise of High-Frequency Trading (H.F.T)—proprietary, automated trading where profitability is inherently dependent on speed. As machines became faster, they faced physical limitations, specifically the speed of light. To minimize latency, trading firms began co-locating their servers inside the same buildings as the "matching engines" (the core systems of trading venues). H.F.T today accounts for roughly half of the world's most significant market volume, with firms operating on a massive scale—often trading trillions of dollars in assets despite employing relatively small staffs. While banks once participated in H.F.T, they have largely retreated due to bureaucratic challenges and post-2008 regulatory constraints, leaving the field to specialized, agile firms.

Modern H.F.T is a decentralized "technical system" centered in a few dozen suburban datacenters globally. In the United States, the "equities triangle" of datacenters in New Jersey facilitates the bulk of share trading. Because these datacenters are physically separated, firms compete to master the "geodesics"—the shortest physical paths—between them using fiber-optic cables and wireless links. In this environment, an algorithm's ability to react to "signals" (patterns in data, such as price movements in Chicago-based futures markets) within nanoseconds is essential for profitability.

The author approaches the study of H.F.T through the framework of "material political economy." The "material" aspect acknowledges that H.F.T is not merely abstract information processing; it is defined by the physicality of hardware, electricity, cooling systems, and the immutable constraints of the speed of light. This is an "Einsteinian" materiality, where spatial location and clock synchronization determine success or failure.

The "political" dimension recognizes that the material architecture of markets is not neutral. Following the principles of actor-network theory and field theory, the author argues that the design of trading systems—such as the "last look" procedures that protect incumbents or the allocation of limited bandwidth within a firm—constitutes "material politics." These arrangements create winners and losers, often reflecting incumbent-challenger conflicts.

While "cultural" factors like legitimacy and beliefs remain relevant, the author maintains that the physical infrastructure of H.F.T is the primary driver of market dynamics. Ultimately, H.F.T remains a "material political" arena where technical efficiency, organizational power, and regulatory environments intersect, shaping the fundamental plumbing of the modern global economy.

Trading at the Speed of Light Introduction (part 2)

The author's investigation into High-Frequency Trading (H.F.T) is grounded in the sociological study of finance, emphasizing that financial systems function as "ecologies" where developments in one area—or "field"—often affect others through "hinges." While politics and finance interact, the author notes that political intervention regarding market regulation is often sporadic, despite historical hinges (such as 1970s policy decisions) profoundly shaping contemporary automated trading.

Central to this book is the "mundane political economy" of finance. Moving beyond dramatic narratives of crises or massive fortunes, the author focuses on the everyday, persistent mechanisms of money-making. This entails an analysis of "market structure"—the formal and informal rules governing who trades, how information flows, and who gains access to order books. Historically, information about order books was restricted to N.Y.S.E specialists, granting them informational advantages. H.F.T has largely eroded these traditional structures by making order books electronically visible to all, yet replacing that advantage with a new, material one: speed. In the current regime, financial success is determined by how quickly a firm's systems can receive, process, and respond to order-book updates.

Because H.F.T profit margins per share are exceptionally thin (often fractions of a cent), success hinges on massive volumes, technical efficiency, and navigating high operating expenses, such as exchange fees and specialized infrastructure costs. These costs create a symbiotic but conflictual relationship between H.F.T firms and exchanges, as exchanges benefit from the very speed races they facilitate. Furthermore, the author incorporates Thomas Philippon's findings on the "efficiency" of financial intermediation, which suggest that despite radical technological improvements, the unit cost of finance has barely changed since the 1880s. This may indicate that gains from technological progress are captured by elite participants as "rents"—excessive rewards for occupying favorable market positions—rather than passed on to the wider economy.

To document these dynamics, the author conducted 358 interviews with 337 individuals across key H.F.T hubs (Chicago, New York, London, and Amsterdam). By overcoming "seniority bias" and navigating the sensitivity surrounding proprietary "secret sauce," the author gathered deep insights into common practices, such as the use of market signals and arbitrage strategies. This qualitative research is supplemented by archival work, site visits to datacenters, and analysis of market materiality, such as the physical placement of microwave towers and fiber-optic cables that dictate trading speed.

The study is organized to explore these themes through distinct domains:

• Chapter 2 examines the Chicago Mercantile Exchange's transformation from open-outcry pits to electronic trading, highlighting the conflictual path and political maneuvering that established the C.M.E's central role.

• Chapter 3 traces the rise of "H.F.T-friendly" share-trading venues like Island and the subsequent automation of U.S and European share markets.

• Chapter 4 analyzes the "dealer-client" structure that still dominates sovereign bond and foreign exchange markets, contrasting it with the more automated environments of shares and futures.

• Chapter 5 details the material, technical systems of H.F.T, showing how the macroscopic speed race is mirrored at the microscopic level within the design of computer hardware, such

as field-programmable gate arrays (F.P.G.A's).

Ultimately, the author argues that focusing on the mundane, technical, and historical realities of trading is essential for understanding modern finance, as these "un-dramatic" factors dictate the adoption of technology, the distribution of economic rents, and the structural stability of the financial system.

Trading at the Speed of Light Introduction (part 3)

Chapter six analyzes market-making and liquidity-taking algorithms, while the conclusion explores applying material political economy to broader digital systems.

To the Towers (part 1)

The transformation of the Chicago Mercantile Exchange (C.M.E) and the Chicago Board of Trade from localized agricultural commodity hubs into pillars of global, automated finance is defined by the development of financial futures and the integration of high-speed microwave communication. These microwave towers, visible throughout the Chicago suburbs, symbolize the material infrastructure linking automated trading centers globally, enabling the lightning-fast transmission of financial signals.

Fifty years ago, Chicago's exchanges were synonymous with "open-outcry" trading—a loud, embodied, and face-to-face system located in octagonal pits. Originally, these exchanges traded agricultural goods like grain, butter, and pork bellies. The transition to financial products—such as currency, Treasury securities, and eventually share-price indexes—was driven by the erosion of government control over global finance following the collapse of the Bretton Woods system in 1971.

This expansion faced a significant legal hurdle: the doctrine of "contemplating delivery." To distinguish legitimate trading from illegal gambling (such as "bucket shops"), courts required that traders actually intend to deliver the underlying commodity. Because intangibles like interest rates or index values could not be delivered, they were legally precluded from futures trading.

Leo Melamed, the C.M.E's influential chair, leveraged the early 1970s food-price inflation crisis to navigate this impasse. Recognizing that political anxiety regarding food prices provided an opportunity to modernize regulation, Melamed and his allies pushed for the creation of a new federal regulator, the Commodity Futures Trading Commission (C.F.T.C). By drafting carefully nuanced language into the Commodity Exchange Act—specifically including "all services, rights, and interests"—they created a legal path for cash-settled futures. The Securities and Exchange Commission's (S.E.C) "snooty" refusal to oversee "pork bellies" further consolidated the C.F.T.C's jurisdictional power, allowing the C.M.E to launch innovative contracts like Eurodollar interest-rate futures and S&P 500 index futures.

The S&P 500 futures became a massive success due to their inherent structural advantages over traditional stock trading: higher leverage, lower costs, and easier short-selling. These futures began to "lead" the market, with prices often moving minutes before the underlying shares. Consequently, these signals became the foundational inputs for early high-frequency trading (H.F.T) algorithms.

Despite the shift toward digital signals, the exchange floors remained intensely human environments characterized by physical crowding, specific hand-signal codes, and rigid social hierarchies based on seniority and reputation. The transition to electronic trading was met with "mortal conflict" by floor traders, who viewed the pits as the source of their professional identity and economic survival. Resistance was profound, often resulting in verbal abuse or even death threats against those pushing for automation.

Melamed himself, originally enamored with the "magical" life force of the pits, eventually concluded that technology was an unavoidable imperative. His conversion, coupled with the competitive threat posed by time-zone advantages in London's liffe exchange, cemented the move toward electronic systems. This evolution bridged the gap between the analog, embodied past of Chicago's trading pits and the digitized, high-frequency future of global finance.

To the Towers (part 2)

To counter competitive threats from London, the Chicago Mercantile Exchange (C.M.E) collaborated with Reuters to develop Globex, an electronic trading system launched in 1992. Designed to facilitate trading outside of Chicago's pit hours, Globex struggled initially, hampered by internal C.M.E resistance—members mandated that the system not compete with pit trading—and the withdrawal of other major exchanges like the Chicago Board of Trade (C.B.O.T) and liffe.

Concurrent with the development of Globex, the C.B.O.T pursued a rival project called "Aurora." Unlike Globex, which utilized an anonymous, list-based price interface, Aurora aimed to simulate the trading floor visually using icons on Apple Macintosh computers. However, Aurora was ultimately abandoned due to technical bandwidth limitations, which prevented it from realistically scaling for global use. Consequently, Globex's "first come, first served" algorithmic matching—an anonymous, abstract model—became the standard for automated futures trading.

The turning point for electronic trading at the C.M.E occurred in 1997, prompted by the threat of the C.B.O.T obtaining a license for a Dow Jones index future. To defend its market position, the C.M.E launched the E-Mini S&P 500 contract, a smaller, electronically traded product accessible to retail investors. Despite fierce opposition from pit traders who viewed it as a violation of the 1987 referendum restricting daytime electronic trading, the E-Mini succeeded through the "bigs and the littles" strategy.

Traders leveraged arbitrage opportunities between the pit-traded S&P 500 future and the electronic E-Mini, using headsets and specialized platforms to coordinate between the physical floor and digital screens. This innovation created a single, unified market and laid the groundwork for high-frequency trading (H.F.T) by establishing the necessity for speed, eventually leading firms to replace human manual entry with algorithm-driven "screen-scrape" processes.

The late 1990s and early 2000s saw a broader shift in the global landscape. European exchanges like the Deutsche Terminbörse successfully leveraged electronic systems to overcome decentralized political obstacles, eventually toppling liffe's dominance in Bund futures. This international success put intense pressure on Chicago. The imminent entry of the electronic exchange Eurex into the U.S market in 2003 to 2004 finally broke the hold of traditional pit trading. The C.B.O.T, lacking a clearinghouse of its own, was forced to modernize and eventually merged with the C.M.E. Simultaneously, the C.M.E's leadership pressured Eurodollar pit traders to embrace electronic trading. As traders realized the speed and efficiency of Globex, they migrated from the pits to screens, leading to the rapid decline of face-to-face trading by the mid-2000s.

Despite the move toward automation, the pattern of "futures lead"—where price movements in C.M.E futures precede those in underlying shares—has persisted. This phenomenon, while central to H.F.T, cannot be explained solely by the existence of E.T.F's, the technological sophistication of the trading systems, or competitive latency pressure, as the C.M.E's infrastructure was historically slower and less advanced than the equity markets it led. The endurance of this lead remains a significant, if technically complex, feature of the modern electronic financial ecosystem.

To the Towers (part 3)

The "futures lead" phenomenon—the tendency for futures prices to precede stock market price discovery—is not an intrinsic quality of futures but a result of market liquidity and leverage. Liquidity is path-dependent, concentrating where large trades are easiest to execute. Furthermore, futures offer superior structural leverage compared to shares, where obtaining similar leverage requires dealer risk and negotiation. This divide is sustained by the jurisdictional split between the S.E.C and C.F.T.C, which prevents regulatory harmonization.

Efforts to merge these agencies fail because the Senate Agriculture Committee, bolstered by industry lobbying and campaign contributions, refuses to relinquish control over financial futures. Consequently, automated trading patterns are deeply embedded in historical political conflicts rather than mere economic or technological necessity.

“We'll show you our book. Why won't they?” (part 1)

The origins of modern high-frequency trading (H.F.T) are rooted in Automated Trading Desk (A.T.D), a firm established in 1989 in Charleston, South Carolina. Founded by academic economist David Whitcomb and statistician James Hawkes, A.T.D represented an experimental bridge between traditional manual trading and the automated, electronic future. Their work, initiated in a modest office, evolved from early statistical modeling of horse racing to the algorithmic prediction of stock price movements.

Whitcomb's approach applied market microstructure theory, specifically employing linear-regression equations to predict a stock's "adjusted theoretical value" (A.T.V) 30 seconds into the future. Because early computing power and data access were limited, the system relied on public information—primarily satellite-fed data from Standard & Poor's ComStock—and key variables like S&P 500 futures prices, which acted as a leading indicator for stock market movements. A.T.D's system was a complex automated setup requiring roughly 80,000 lines of code, managing everything from market data ingestion to order execution and real-time profit calculation.

During the late 1980s and early 1990s, A.T.D navigated a landscape dominated by the New York Stock Exchange (N.Y.S.E), Nasdaq, and the electronic system Instinet. The N.Y.S.E remained a bastion of manual trading, governed by "specialists" who maintained private order books. Although federal legislation—specifically the 1975 Securities Acts Amendments—sought to create a centralized national market system (such as the proposed Consolidated Limit Order Book or C.L.O.B), political pushback from the N.Y.S.E and regional exchanges prevented this. Instead, the Intermarket Trading System (I.T.S) was established. I.T.S functioned at a human pace, often allowing specialists to decline trades, thereby preserving the dominance of the N.Y.S.E and maintaining the opacity of its order books.

Nasdaq provided an alternative, entirely electronic venue without a physical trading floor, though it was dominated by market-makers who enforced strict norms, including a minimum price spread of an eighth of a dollar. These market-makers fiercely guarded their spreads, often harassing any dealer who attempted to lower them. Instinet offered more transparency but struggled to gain widespread adoption among institutional investors, who preferred the "soft dollar" benefits—such as subsidized research and I.P.O access—provided by traditional Wall Street dealers.

A.T.D's initial live trading attempts highlighted the severe financial friction inherent in this era. Trading on Instinet, the firm was often "bagged" by Nasdaq dealers who possessed superior information. Trading on the N.Y.S.E via a major investment bank offered potential, but the standard commission fees were prohibitively high.

Even after negotiating lower rates, A.T.D's survival was only secured by a profit-sharing arrangement with their sponsoring bank, where the bank covered the costs of trading in exchange for a majority share of the profits. This formative period reflects a time when algorithms were nascent, manual intermediaries held significant informational advantages, and high transaction costs served as the primary obstacle to the efficiency that H.F.T would later claim to provide.

“We'll show you our book. Why won't they?” (part 2)

The section explores the emergence of "soes banditry" and the subsequent rise of high-frequency trading (H.F.T) venues like Island, which fundamentally challenged the rigid, dealer-centric structure of the Nasdaq market.

The Breach in the System

The transition began in the aftermath of the 1987 stock market crash. Following dealer misconduct, the S.E.C mandated that Nasdaq dealers honor orders placed through the automated Small Order Execution System (soes). This ruling created a vulnerability: if dealers failed to update their quotes promptly in changing market conditions, observant traders could use soes to "pick off" these stale prices. These "soes bandits"—often amateurish traders operating from makeshift rooms—exploited this latency, forcing their way into a market previously dominated by institutional insiders.

Automated Banditry

As early profitability waned, Automated Trading Desk (A.T.D) sought to professionalize this practice. By partnering with experienced traders, A.T.D automated the predictive logic of the bandits. They discovered that Nasdaq's non-anonymous, limited display of bids and offers functioned as a broadcast of dealers' private information.

By monitoring these arrays, A.T.D's algorithms could anticipate price movements and execute trades faster than human competitors. Despite the social stigma and intense hostility from established dealers, these automated methods allowed firms like A.T.D to survive and scale.

The Rise of Island

To solve the difficulty of closing trades profitably, entrepreneurs Josh Levine and Jeffrey Citron developed "Island," an electronic communications network (E.C.N). Built by programmers steeped in a "hacker ee-thoss"—valuing efficiency, succinct code, and a disdain for bureaucracy—Island was designed for radical speed and direct interaction. Its matching engine, which avoided the slow, "transactionally integral" databases used by traditional institutions in favor of in-memory processing, could execute trades in milliseconds.

Island also introduced structural innovations that would define modern market architecture:

• Low Fees and Rebates: By charging minimal fees and offering liquidity rebates, Island attracted H.F.T firms to act as automated market-makers.

• Technological Advantage: Using fine-grained price increments, Island and other E.C.N's undercut traditional dealers, rapidly capturing a significant share of Nasdaq trading volume.

• The H.F.T "Hinge": A mutually reinforcing cycle emerged. Profits from H.F.T enabled firms to expand, while the liquidity provided by H.F.T firms made Island an increasingly attractive venue.

Market Transformation

This development created a "hinge" that linked the evolution of trading venues to the growth of H.F.T. As H.F.T firms coalesced around a common set of requirements—low latency, competitive pricing, and advanced functionality—they began to exert pressure on global exchanges to adopt similar technological standards. By the early 2000s, this cycle of technological innovation and market structure reform began to dismantle the influence of traditional, slower-moving exchanges like the N.Y.S.E, signaling a broader transformation of global financial markets toward the automated, high-frequency landscape known today.

“We'll show you our book. Why won't they?” (part 3)

The evolution of U.S share trading was driven by a regulatory and structural "hinge" that favored electronic competition over traditional dealer-based models. In the 1990s, scandals—specifically allegations of "front-running" and the systematic avoidance of odd-eighths pricing—eroded trust in Nasdaq dealers, sparking S.E.C intervention. The 1996 order-handling rules and 1998 Regulation A.T.S allowed new Electronic Communication Networks (E.C.N's) to compete directly with incumbent exchanges by posting prices on public screens. This shift was bolstered by the centralization of clearing and settlement, a "dreary" but vital back-office function that, unlike trading, was integrated into a single, efficient system, thereby lowering barriers for new venues like Island.

Regulation National Market System (Reg N.M.S) in 2005 further catalyzed this transformation. By stripping the N.Y.S.E of its protected status and the requirement to pause for manual, floor-based trades, the S.E.C effectively dismantled the duopoly of traditional exchanges. As market share migrated to E.C.N's, Nasdaq and the N.Y.S.E were forced to acquire their rivals—such as Island and Archipelago—to survive. Consequently, the established venues pivoted from resisting high-frequency trading (H.F.T) to courting it, recognizing that H.F.T firms provided the liquidity and volume necessary to compete. This environment solidified the role of electronic order books, where H.F.T algorithms, utilizing signals like order-book dynamics and market fragmentation, became the dominant market participants.

European markets followed a similar, albeit more complex, path. Initially, high exchange and clearing costs stifled H.F.T until firms like the American-backed Chi-X and bats Europe successfully pushed for lower-cost alternatives and clearing interoperability. The E.U's Markets in Financial Instruments Directive (mifid 1) eventually shattered national concentration rules, allowing these new, fast-matching M.T.F's to thrive. By 2011, Chi-X had become Europe's largest trading venue, illustrating how regulatory changes and technological speed created a nearly universal landscape for algorithmic competition.

The experience of the firm Automated Trading Desk (A.T.D) highlights the high-stakes, shifting nature of this era. Once a pioneer of H.F.T, A.T.D thrived by exploiting pricing inefficiencies and slower human competitors. However, the S.E.C's "decimalization" in 2001 compressed spreads, reducing the profitability of market-making and eliminating arbitrage opportunities between pricing grids.

As the "speed race" intensified, A.T.D struggled with the technical and operational risks of extreme low-latency requirements, evidenced by a $3 million loss caused by a programming error during a desperate push for speed. A.T.D ultimately pivoted to become a "wholesaler"—executing retail orders with lower latency demands—before being sold to Citigroup and eventually absorbed by Citadel. Despite A.T.D's demise, the algorithmic, highly fragmented market structure it helped pioneer remains the standard for modern share trading.

Dealers, Clients, and the Politics of Market Structure (part 1)

The U.S Treasurys market represents a complex hybrid of traditional dealer-client relationships and modern, algorithm-driven electronic trading. Historically, this market functioned through direct, voice-mediated interactions, where "primary dealers"—banks officially designated by the Federal Reserve—acted as essential intermediaries. Clients were "price-takers," relying on dealers to facilitate trades, while dealers managed their own risk by unwinding positions via interdealer brokers.

While electronic systems like BrokerTec (for interdealer trading) and Bloomberg F.I.T (for dealer-client trading) have superseded much of the vocal shouting of the 1980s, the market remains bifurcated. Unlike the "all-to-all" equity markets, where the dealer-client distinction has largely collapsed, the Treasurys market preserves a structure that separates anonymous interdealer order books from non-anonymous dealer-client channels.

Technological shifts, specifically the entry of High-Frequency Trading (H.F.T) firms, have transformed the interdealer segment. In the late 1990s and early 2000s, interdealer platforms Cantor Fitzgerald (eSpeed) and BrokerTec emerged as electronic order books. Initially, these platforms mirrored the traditional "workup protocol" used by human brokers to allow for larger, negotiated trades.

However, a competitive struggle for market share eventually compelled these platforms to grant H.F.T firms access to gain volume. Once H.F.T algorithms entered the Treasurys triangle—a network of interconnected datacenters—they became dominant, executing trades at speeds impossible for human traders. Because H.F.T's are faster at updating prices, they frequently "pick off" dealers who have not yet updated their quotes, forcing dealers to adapt.

Despite this transformation in interdealer trading, a fully anonymous, all-to-all Treasurys market remains elusive. Efforts to launch such platforms, most notably "Direct Match," were thwarted by regulatory and structural hurdles, specifically the requirement for central clearing through the Fixed Income Clearing Corporation (F.I.C.C). Accessing F.I.C.C is capital-intensive, and established banks maintain significant leverage in navigating these requirements. Consequently, H.F.T firms have largely bypassed the creation of new all-to-all venues, opting instead for private, bilateral arrangements with dealers to stream prices directly. These arrangements allow dealers to maintain control over whom they trade with, protecting them from the aggressive speed of H.F.T algorithms, while providing H.F.T's a profitable, less competitive venue.

The persistence of the dealer-client structure is significantly bolstered by the unique role of government policy. Unlike the Securities and Exchange Commission's aggressive reform of equity markets, the Department of the Treasury and the Federal Reserve maintain a conservative stance toward Treasurys. They view the primary-dealer system as a vital component of sovereign debt management, ensuring that government auctions have guaranteed participants. Dealers enjoy "infrastructural power" due to this state-sanctioned role; policymakers are wary of reforms that might threaten bank participation, as seen in the lack of dramatic regulatory interventions over the past decades.

This structural stability is even more pronounced in Europe. European sovereign bond markets rely on platforms like M.T.S, which were integrated into the "liquidity pacts" between governments and primary dealers. Because M.T.S holds a dominant, non-competitive position in the Eurozone, it has been far more successful than U.S platforms in restricting H.F.T entry, effectively blocking the "hinge" between H.F.T and market structure that defined the U.S Treasurys transition. Thus, while technology has introduced H.F.T, the sovereign bond market's structure is ultimately dictated by the symbiotic relationship between states and the banks that finance them.

Dealers, Clients, and the Politics of Market Structure (part 2)

The struggle between high-frequency trading (H.F.T) firms and established dealer-client market structures highlights deep-seated resistance to institutional change. In Europe, efforts by M.T.S to facilitate H.F.T access to sovereign-bond markets met with extreme hostility from the dealer community, leading to the dismissal of senior executives and the indefinite preservation of an exclusive, bank-centric "club." Regulatory and political spheres largely support this status quo, prioritizing the control of primary debt issuance and the profitability of primary dealers over the algorithmic "efficiency" argued for by H.F.T proponents. Consequently, H.F.T firms remain largely excluded from European sovereign-bond trading.

In the foreign exchange (F.X) market, H.F.T firms faced similar barriers. Traditionally, F.X was a telephone-based, bilateral dealer-client market. When Reuters introduced electronic "conversational dealing" in the 1980s, it succeeded by mimicking personal interaction and avoiding automated matching.

Later, when Reuters attempted to introduce a full-fledged interdealer matching system, incumbent banks—fearing the erosion of their franchise—countered by forming the Electronic Broking Services (E.B.S) consortium. E.B.S was a complex, global technological endeavor that accommodated the realities of F.X trading: the absence of a central clearinghouse (necessitating bilateral credit checks) and the "sun-following" nature of the market. While H.F.T firms eventually gained access to F.X venues like Hotspot and later E.B.S, their influence remains constrained by "material politics," such as last-look mechanisms and the ability of dealers to filter or exclude specific participants based on credit exposure or predatory behavior.

These distinct historical and institutional paths have created variations in the "signals" available to H.F.T algorithms. In futures, high market concentration (dominated by the C.M.E) prevents fragmentation, leaving H.F.T firms reliant almost exclusively on order-book dynamics. In contrast, U.S share trading is characterized by significant venue fragmentation, providing rich signals for H.F.T. Sovereign bonds and F.X occupy middle ground; while they feature fragmented trading environments, H.F.T access is limited by the prevalence of dealer-client structures and the inability to fully exploit price differences across venues due to exclusionary practices and non-standardized clearing.

Ultimately, these differences illustrate that the politics of market structure is inextricably linked to materiality. The varying effectiveness of H.F.T across these asset classes is determined by whether the underlying market infrastructure permits the exploitation of fragmentation, orderbook signals, or lead-lag relationships. Where dealer-client systems remain robust, H.F.T firms are forced to make substantial compromises, adapting their strategies to the existing material and political order rather than transforming it. The resulting landscape is not uniform but shaped by the specific technological, regulatory, and institutional constraints of each asset class.

“Not only would I lose my job, I might lose my legs too!” (part 1)

This section details the physical and technological evolution of high-frequency trading (H.F.T) through the lens of datacenters and long-distance connectivity.

Datacenters as Material Infrastructures

The Chicago datacenter "Cermak" serves as a primary case study. Formerly a massive printing plant, its structural legacy—specifically its reinforced floors and vertical shafts—proved ideal for hosting modern server racks. Datacenters function as the nexus of the Internet, housing "meet-me rooms" where private networks converge.

These facilities are primarily constrained by power density and thermal management. Because nearly all electricity input eventually converts to heat, high-capacity cooling systems using massive quantities of refrigerant are essential to ensure equipment reliability. Trading firms operate within these spaces using leased, secured, and often "caged" hardware. Internally, data transmission occurs via optical fibers; orders move from the firm's system to an exchange's matching engine through an "order gateway."

The exchange then publishes market data back to firms, allowing them to mirror the order book. The physical configuration of these internal paths is highly consequential; even minor differences in cable length or electronic routing within a datacenter can yield critical speed advantages.

The Materiality of Distance: Fiber and the Geodesic

Transmission between Chicago and New Jersey is the most vital route for H.F.T. Initially, telecom providers ignored speed, prioritizing low-cost, hub-and-spoke networks optimized for ease of repair. H.F.T firms, however, necessitated the "shortest path" (the geodesic).

The industry responded by developing the "gold line"—a dedicated "dark-fiber" connection that bypassed standard telecom routing. Because segments were leased from existing infrastructure, it could not perfectly follow the geodesic. This limitation prompted the creation of Spread Networks, which invested over $300 million to build a direct, private fiber line.

The project underscored the "tyranny of location," where physical obstacles—such as rivers or parking lots—transformed into "pinch points" or "obligatory passage points." These sites allowed owners to charge significant economic rents for access. Spread Networks enforced extreme technical uniformity to ensure fairness among subscribers, even compensating for microscopic inhomogeneities in glass fibers to guarantee identical transmission times.

The Rise of Microwave Transmission

While Spread Networks optimized fiber, it remained constrained by the refractive index of glass, which slows light to two-thirds of its vacuum speed. This led to the adoption of microwave transmission, which travels through the atmosphere at near-vacuum light speeds.

Alex Pilosov, a pioneer in this field, built a clandestine microwave link using modest resources and ingenious, "hands-on" engineering. Unlike fiber, which requires fixed, buried lines, microwave is a line-of-sight technology that relies on towers. Pilosov's success relied on bypassing standard, slow contracting processes and performing site analysis himself. He managed to secure a faster path than fiber, despite not following the geodesic, by finding efficient tower alignments.

The subsequent proliferation of microwave links intensified the speed race, proving that in H.F.T, market dominance often favors the firm that is "first 99 percent of the time." The transition from fiber to microwave also highlighted the physical contingencies of the "material world": high-altitude antenna maintenance, susceptibility to weather, and the persistent need for proprietary teams to monitor infrastructure. Ultimately, these developments reveal that H.F.T is not an abstract digital activity but a practice deeply bound to geography, engineering, and the physical constraints of light transmission.

“Not only would I lose my job, I might lose my legs too!” (part 2)

The pursuit of speed in high-frequency trading (H.F.T) has transformed microwave and millimeter-wave network design from a conservative telecommunications practice into a ruthless optimization effort. The competition between firms like Vigilant (D.R.W), New Line (Getco/Virtu and Jump), and McKay Brothers (founded by physicists Robert Meade and steh-fawn Týč) has forced a departure from traditional reliability-focused engineering. These firms prioritize total path length above all else, often employing risky "long hops" over physical barriers like Lake Michigan or Lake Erie to minimize repeater-induced delays.

To maintain geodesic-hugging routes, firms have abandoned the reliable 6 G.H.z frequency in favor of higher bands (11, 18, and 23 G.H.z), which are more susceptible to "rain fade." This trade-off—prioritizing being fast 99% of the time over being reliable 100% of the time—highlights the industry's single-minded pursuit of latency reduction. The infrastructure has become increasingly specialized; repeaters are now weatherproofed and placed directly on towers near antennas to eliminate signal travel time within cable racks.

Moreover, firms face the "tyranny of exact location," where even a few meters of distance significantly impacts speed. This has led to intense battles over tower siting, land easements, and the physical distance of "fiber tails" connecting antennas to datacenters. These networks are further constrained by atmospheric phenomena like ducting and rain, which can cause intermittent signal failure, sometimes affecting market prices during adverse weather.

In the "New Jersey metro" area, where multiple datacenters house the matching engines for share trading, fiber-optic networks were initially built without prioritizing speed. As firms recognized the competitive necessity of latency reduction, they engaged in "microbuilds"—securing private rights-of-way to optimize routes. Because microwave transmission lacks the capacity for the high-volume data streams required in share trading, firms transitioned to millimeter-wave wireless (70 to 80 G.H.z). This offered higher capacity but introduced further vulnerability to rain. To combat this, Anova Financial Networks developed hybrid links that seamlessly switch between millimeter-wave and laser transmission, depending on weather conditions. Firms even spent significant resources engineering anti-fouling coatings to prevent bird droppings from obstructing laser optics.

Capacity remains a central hurdle, as bursts of market data can exceed the bandwidth of wireless links, leading to "buffering delay." Firms circumvent this by "editing down" datafeeds to transmit only critical information. Newer technologies like Local Multipoint Distribution Service (L.M.D.S) are being deployed to increase capacity further, potentially nullifying the structural advantages held by those with private, exclusive links.

Finally, the "materiality of the small" extends into the datacenter itself. Beyond network geography, H.F.T firms exploit internal physical advantages, such as receiving proprietary "fill" messages regarding their own trade executions before those trades are published in the exchange's public datafeed. This "loophole," involving millisecond-scale differences, allows firms to deploy "scout" or "canary" orders to gain early intelligence on price movements, a practice once so sensitive that market participants feared professional and even physical retaliation for discussing it.

“Not only would I lose my job, I might lose my legs too!” (part 3)

This section explores the "materiality of the small" in High-Frequency Trading (H.F.T), where minute technological advantages translate into significant economic outcomes. A central controversy involves whether private "fill" messages regarding trade execution should arrive before public datafeed reports. Proponents argue this rewards the risk-taking inherent in market-making, while critics view it as an inequitable loophole. Exchanges have faced pressure to standardize these timings to balance market incentives.

Beyond messaging order, H.F.T is profoundly shaped by physical infrastructure. Because light signals take nanoseconds to travel mere centimeters, the precise cabling within datacenters—and the location of servers relative to matching engines—are critical. While exchanges now often mandate equal-length "cross-connects" to prevent unfair advantages based on physical proximity, firms continue to pursue speed through higher-capacity, lower-latency connections that bypass standard network switches.

The pursuit of speed extends into the hardware itself. To bypass the limitations of general- purpose CPUs, firms employ Field-Programmable Gate Arrays (F.P.G.A's). These chips allow for "close-to-the-metal" control, enabling hardware-level execution of trading algorithms. Programming F.P.G.A's is a labor-intensive, material process, as engineers must navigate physical constraints like limited chip "real estate," signal propagation delays, and the stability of internal clock cycles.

Furthermore, H.F.T firms utilize "speculative triggering"—a technique where systems begin transmitting order packets before a signal is fully confirmed. If the anticipated market move does not occur, the system kills the packet (e.g., by scrambling the checksum) to ensure no invalid order is placed. This practice has become so widespread that major exchanges like the C.M.E and Eurex have had to implement mitigation strategies to prevent system overload. Ultimately, the H.F.T industry is characterized by a relentless, resource-heavy "arms race." Experts within the field acknowledge the profound opportunity cost, as vast amounts of specialized brainpower are dedicated to shaving nanoseconds off execution times, a pursuit driven by the competitive dynamics of modern financial markets.

How H.F.T Algorithms Interact, and How Exchanges Seek to Influence It (Part 1)

High-frequency trading (H.F.T) is fundamentally structured around the interaction of "making" and "taking" algorithms. Making orders (liquidity-providing) post bids or offers at prices that do not immediately execute, thereby populating the order book. Taking orders (aggressive) execute against these existing entries immediately.

While this distinction may seem technical, it represents a significant divide in H.F.T, often carrying profound moral weight; market-making is frequently framed as a socially legitimate service, while taking is sometimes viewed with suspicion. This legitimacy is often used by firms to defend their practices against regulators or skeptical observers.

In practice, market-making is economically precarious. Algorithms aim to profit from the spread between bids and offers, but they risk being "run over"—purchasing assets just before prices fall or selling before they rise. To mitigate this, firms employ predictive models, typically using linear-regression to calculate a "theoretical value" or "microprice," allowing them to adjust their inventory and bids dynamically.

Because many markets are "tick-constrained"—meaning prices can only move in set increments—H.F.T success often hinges on "getting good places in the queue." Time-priority rules ensure that earlier orders are executed first, making the speed of order placement critical.

In the U.S. equities market, this dynamic is further complicated by Regulation N.M.S (Reg N.M.S) and its "order protection" rules, which prevent trading through superior prices on other exchanges. Because information transmission is limited by the speed of light—a phenomenon described as "Einsteinian" rather than "Newtonian"—algorithms often struggle to see price changes across geographically separated datacenters simultaneously. To navigate this, firms use Intermarket Sweep Orders (I.S.O's), which allow an algorithm to clear protected orders on other venues while simultaneously placing their own. Using I.S.O's requires sophisticated "compliance infrastructure" to prove adherence to regulations, creating a barrier to entry for smaller firms.

"Taking" strategies, by contrast, are more heterogeneous. Highly quantitative taking involves processing vast, complex data streams from global markets to predict price movements across assets. Unlike market-making, which prioritizes speed to maintain queue position, these sophisticated taking algorithms sometimes prioritize analytical depth over raw speed, as they are not competing for a position in the queue.

A second, more common form of taking is "picking off" stale quotes. When significant market events occur, such as a shift in a share-index future, existing market-making quotes in related instruments may become mispriced or "stale" before the maker can cancel them. Taking algorithms profit by executing against these stale quotes. This creates a relentless, nanosecond-speed race between makers trying to cancel outdated orders and takers trying to capture them.

This interaction is so sensitive that it is influenced by physical factors, such as the reliability of wireless microwave links. For instance, heavy rain can disrupt microwave signals, forcing U.S. market-makers to widen their spreads to account for the loss of a protective data feed. The material reality of these links and the environmental conditions affecting them demonstrate that the interaction between algorithms is as much a product of physical and geographical constraints as it is of financial logic.

How H.F.T Algorithms Interact, and How Exchanges Seek to Influence It (Part 2)

The interaction between "taking" and "making" algorithms in high-frequency trading (H.F.T) is defined by a sophisticated, ongoing game of prediction and reactive adaptation. Taking algorithms profit by identifying predictable patterns in the behavior of market-making algorithms—such as their use of linear regression signals, reliance on inventory management thresholds, or delayed reaction times when markets shift. To exploit these patterns, advanced takers utilize large-scale grid computing to run complex machine-learning models, such as "random forests," offline. These models identify the signatures of specific competitors, allowing takers to anticipate when a market-maker must "shade" its prices to reduce inventory or when it will be forced to liquidate a position.

This "cat and mouse" game is increasingly enabled by greater technological determinism in exchange systems, such as precise atomic-clock synchronization and improved C++ programming, which remove the "jitter" that once masked algorithmic behavior. While price prediction and picking off stale quotes are standard, more controversial tactics include market-impact trading. In this practice, an algorithm executes trades specifically designed to trigger reactions in other systems, such as forcing "weak-handed" firms to liquidate positions (swiping the book). Such behavior is often viewed by market participants and regulators as potentially manipulative, though it is difficult to prosecute, unlike "spoofing," which involves placing and canceling large orders to mislead other algorithms and has faced increased legal and criminal scrutiny.

The industry displays a functional specialization between "making" and "taking." Market-making is primarily an engineering-heavy discipline focused on robustness, speed, and continuous quoting, whereas taking is a quantitatively focused discipline centered on alpha signals and statistical analysis. This specialization is often "hardwired" into firm architecture; for instance, some firms implement internal self-trading prevention software. While designed to avoid internal conflicts, such systems can unintentionally protect a firm's own market-making algorithms from being "run over" by their own taking algorithms, thereby reinforcing institutional biases toward one strategy over the other.

In less regulated environments like foreign exchange (F.X), the friction between incumbent dealer banks and H.F.T firms has led to deliberate, material interventions. Facing challenges from H.F.T "takers" that exploited slow, manual, or fallible bank systems, F.X venues introduced "last look" protocols. This allows a dealer to receive a transaction request and reject it within a set timeframe, effectively neutralizing predatory taking strategies like triangular arbitrage. Though "last look" is a subject of intense ethical debate and recent regulatory scrutiny, its implementation contributed to a notable shift: many H.F.T firms moved away from aggressive taking toward market-making.

Other interventions, such as Minimum Quote Lifespans (M.Q.L) and partial reversals of decimalization, aimed to protect manual traders from being outpaced by automated "quote-matching" algorithms. These measures reflect a broader trend where venues, pressured by incumbent dealers, have modified market rules to curb perceived unfair advantages, sometimes creating new venues—like ParFX—that mandate transparency and higher price increments to foster a "cleaner" market environment.

How H.F.T Algorithms Interact, and How Exchanges Seek to Influence It (Part 3)

To maintain liquid markets and avoid "empty screens," trading platforms have increasingly employed "material politics"—physical or architectural interventions—to protect market-making algorithms from being "picked off" by faster, predatory "taking" algorithms.

A primary strategy is randomization, used by platforms like ParFX, E.B.S, and Reuters. By adding variable delays or batching orders, these venues prevent the fastest algorithms from consistently winning the "race to the front" of the queue. Reuters evolved this approach into an "asymmetric speed bump," which delays incoming "taking" orders while allowing "cancellations" to pass through instantly. This gives market-makers a crucial window to retract stale quotes when market prices shift, without granting them the unfair informational advantages associated with the controversial "last look" practice.

In U.S share trading, this has sparked significant regulatory conflict. While I.E.X's symmetric speed bump (slowing all traffic) gained approval, attempts to implement asymmetric bumps by exchanges like edga have faced rejection by the S.E.C, which scrutinizes whether such designs unfairly discriminate against specific participants.

Beyond speed bumps, options exchanges use specialized infrastructure to protect makers, such as "purge ports" that facilitate "mass cancels" to instantly invalidate hundreds of quotes simultaneously. Furthermore, platforms may employ less visible, structural interventions, such as channeling makers and takers through different hardware or A.P.I architectures, where the maker's path is significantly faster. These maneuvers demonstrate that the "material politics" of finance extends beyond public debates, residing deep within the "black boxes" of exchange systems, where architectural choices fundamentally reshape the economic interaction between market participants.

Conclusion (part 1)

The conclusion to the text, framing the study through the lens of "material political economy," begins with a historical analogy: the 1381 storming of St Albans Abbey, where townspeople destroyed a floor paved with confiscated millstones. This act highlights how feudal lords used their control over milling infrastructure to force tenants into using their mills for a fee, a direct intervention in market structure. Similarly, the modern financial sector exhibits comparable material interventions. The text cites the 2019 controversy over wireless antennas at the New York Stock Exchange's datacenter, where a marginal speed advantage for some firms sparked conflict over fairness, highlighting how physical infrastructure—antennas, fiber-optic routes, and pole placements—directly determines market participants' competitive capabilities and profits.

The author argues that market structure is fundamentally "material political economy." This encompasses three core elements: the physical reordering of the world, political acts facilitated by legal and power structures, and economic consequences that shape competition. In modern trading, this materiality extends beyond hardware to include systems for clearing and settlement. These mundane back-office processes are crucial; they bind markets together, and access—or the lack thereof—can effectively dictate a firm's ability to participate. The contrast between the fragmented, competitive landscape of U.S shares and the concentrated, centralized nature of futures trading is shown to be a product of distinct historical interactions between political systems and financial exchanges in the 1970s.

Information itself is treated as a material asset. Algorithms operate on "signals"—datafeed attributes—that are not natural givens but the outcome of past regulatory and structural conflicts. Hardwired into modern technological systems, these signals are essentially materialized political economy. Furthermore, the debate surrounding order book anonymity and the dissemination of "smart money" data (like Nasdaq's retracted Pathfinders feed) demonstrates that the materiality of markets is inseparable from cultural and normative preferences regarding transparency and fairness.

Constraints on algorithmic action are equally political. The crackdown on "spoofing" reflects varying degrees of regulatory enforcement across asset classes, with authorities often hampered by jurisdictional splits. Similarly, the pivotal role of "intermarket sweep orders" in U.S share trading exists only because of the 2005 Regulation National Market System (Reg N.M.S), a regulation born of intense lobbying and party-line splits among S.E.C commissioners.

Ultimately, the text contends that government regulation is neither autonomous nor purely external. It is a variable force, often buffeted by finance-sector lobbying and tactical maneuvering. Regulators frequently propose ambitious, controversial reforms (like the "consolidated limit order book") to secure more modest, incremental gains—a strategy exemplified by the "battle of the asterisk," where subtle technical changes were used to force market transparency. In this view, markets are not natural phenomena but are actively "produced" by governmentality, lobbying, and the material infrastructures of trading, ensuring that the winners and losers of financial competition are continuously redefined by both human intent and technical design.

Conclusion (part 2)

This concluding section synthesizes the complex relationship between the political system, market economics, and the material realities of high-frequency trading (H.F.T).

The Political Economy of Market Structure

The author argues that political intervention in financial regulation generally follows two paths. The first is pervasive lobbying by specific interest groups to influence regulators. The second—rarer and more consequential—occurs during crises or scandals, such as the late-1960s "paperwork crisis," where political actors seek reforms to appease constituents. The post-2008 Dodd-Frank Act illustrates how regulators can utilize broader reform mandates to address specific issues, such as the ban on "spoofing," which, while unrelated to the banking crisis itself, was nonetheless enacted during that window of legislative opportunity.

Structurally, the U.S. financial system is defined by a deep, persistent divide between the S.E.C and the C.F.T.C, underpinned by the jurisdictions of powerful congressional committees, most notably the Senate Agriculture Committee. This separation has prevented the consolidation of financial regulation, thereby sustaining the distinct regulatory environments that allow H.F.T arbitrage strategies, such as "futures lead," to flourish.

Mundane Economics and the Battle of the Asterisk

The author posits that the politics of market structure are often "local," driven by participants within the market jockeying for advantage. The "battle of the asterisk"—a dispute over whether to label rounded price quotes—is presented as a parable for this mundane politics. Though the price difference was negligible, incumbents fought the asterisk because it threatened the routine, undramatic money-making that defines H.F.T.

H.F.T is characterized by narrow profit margins per trade, necessitating massive volume and rigorous cost management. Because revenue is easily eroded by expenses—such as exchange fees, data costs, and technology investments—H.F.T firms are obsessed with efficiency. Tiered fee structures often solidify the power of incumbents, creating barriers to entry for smaller firms. Furthermore, while individual trade profits are small, the aggregate "prize" of global speed races is estimated to be well in excess of $5 billion annually, creating a massive, collective economic incentive that fuels continuous technological spending.

Technological Hinges and Future Evolution

The author explores the potential for new "hinges" between trading strategies and market designs. While the current system incentivizes extreme speed, there is emerging evidence that firms are exploring alternatives, such as venues that mitigate speed races through asymmetric speed bumps or frequent batch auctions. While exchanges have historically resisted these changes to protect revenues from selling fast data, the rise of firms that prefer to compete on grounds other than latency suggests that the current market environment may not be permanent.

Unruly Materiality and Systemic Risk

Finance's large technical systems are not fully under human control; they are "unruly." Events like the 2010 "flash crash" and the 2014 Treasurys price spike highlight how automated, rule-bound systems—designed to be prudent—can produce collective, disorderly outcomes. Technical malfunctions, such as the catastrophic 2012 Knight Capital error, underscore the risks of software-driven trading.

Interestingly, some technical "flaws" may have unintended benefits. "Jitter" (random latency in exchange systems) can serve as an equalizer, preventing the fastest firms from dominating and allowing slower competitors to remain viable. Ultimately, the author concludes that while H.F.T has reduced transaction costs, the systemic pursuit of speed may lead to the waste of social resources, necessitating caution, experimentation, and potentially a shift toward more robust market designs.

Conclusion (part 3)

The conclusion explores the "material political economy" of high-frequency trading (H.F.T) and extends this analytical framework to other digital domains. In H.F.T, the shift toward deterministic systems—exemplified by field-programmable gate arrays (F.P.G.A's) at the C.M.E—has largely resolved issues of "jitter," yet this has exacerbated the trend toward market consolidation. Because speed races now consistently reward a small number of firms capable of near-instantaneous responses, the H.F.T landscape has shrunk, with giants like Virtu and Citadel dominating U.S share trading. This material reality, where speed is a binding, Einsteinian constraint, necessitates critical scrutiny of the current market infrastructure.

The author argues that this approach is broadly applicable to other sectors. In cryptocurrencies, the struggle over "asic resistance"—the attempt to prevent specialized mining hardware from centralizing control—is a clear manifestation of material politics. Similarly, the mainstream digital economy relies on vast "material megasystems" managed by automated, high-scale infrastructure.

The advertising ecosystem, in particular, exhibits high levels of intermediation and opacity. Real-time bidding systems, which require automated auctions within milliseconds, create an environment where intermediaries capture significant portions of revenue, often at the expense of publishers. Because these systems are increasingly opaque and concentrated, the author advocates for a rigorous, "black box" analysis of their mechanisms, asserting that understanding the material foundations of these digital spaces is essential to navigating the political and economic conflicts embedded within them.

You have reached the end of the document.