4. Algorithmic Trading#
4.1. Introduction#
Algorithmic trading refers to the use of computer programs to automate the process of making trading decisions and executing orders in financial markets. Although the term is widely used, its precise meaning can vary depending on the context and the regulatory framework. Two influential definitions are the following:
Bank for International Settlements (BIS, 2011)[2]: Trading technology in which order and trade decisions are made electronically and autonomously.
MiFID II (Directive 2014/65/EU, Article 4, Definition 39)[3]: Trading in financial instruments where a computer algorithm automatically determines individual parameters of orders such as whether to initiate the order, the timing, price or quantity of the order or how to manage the order after its submission, with limited or no human intervention.
Notice that we use the terms algorithmic trading and automated trading as synonyms. In some references, algorithmic trading is used more narrowly to denote algorithmic execution, a specific form of automated trading focused on the optimal execution of predefined orders.
Algorithmic trading can be viewed as a subset of systematic trading, which refers to any trading strategy defined in a rule-based, methodical manner. It is also a subset of quantitative trading, where trading decisions follow the principles of the scientific method. In this framework, we first construct a scientific model of the trading environment—for example, a stochastic process such as a random walk—to represent market dynamics. This model is then used to derive inferences about quantities of interest, such as the likely range or direction of future prices. These inferences serve as inputs to mathematical optimization procedures, which determine the optimal trading actions—such as when and at what levels to trade—under given objectives and constraints.
The MiFID II definition is intentionally broad. It encompasses both complex, fully automated trading strategies and simpler automated rules, such as dynamic stop-loss orders or RfQ auto-negotiation rules. The exception being complex orders that are directly implemented within a exchange, which are considered part of the market infrastructure. In the following discussion, we will adopt the first, more restrictive definition, focusing on sophisticated algorithmic strategies where automation plays a central role in decision-making and execution.
4.2. High-Frequency Trading#
High-Frequency Trading (HFT) is a subset of algorithmic trading distinguished by extremely short holding periods, rapid order submission and cancellation, and technological infrastructures designed to minimize latency. According to the Bank for International Settlements [2], HFT is a “subset of automated trading in which orders are submitted and trades executed at high speed, usually measured in microseconds, and a very tight intraday inventory position is maintained.” Such strategies seek to gain advantage from the ability to process information on market conditions and react almost instantaneously, typically resulting in a very large number of small trades, held for short periods, and generating substantial message traffic. To achieve this, HFT firms tend to place their trading servers physically close to the electronic market’s matching engines—a practice known as co-location—to minimize transmission delays or latency.
At the European regulatory level, MiFID II (Directive 2014/65/EU, Article 4(1)(40)) [3] defines HFT as an algorithmic trading technique that relies on infrastructure designed to minimize latency, such as co-location, proximity hosting, or high-speed direct electronic access, and in which individual trades or orders are initiated, generated, routed, or executed by systems without human intervention. It also specifies that such activity typically involves high intraday message rates consisting of orders, quotes, or cancellations.
In practice, low-latency infrastructures are at the core of HFT. Co-location refers to hosting trading servers directly within or adjacent to an exchange’s data center to reduce round-trip latency to microseconds. Examples include Equinix LD4 in Slough (London) and NY4 in Secaucus (New Jersey), which host a large portion of global financial trading infrastructure. Proximity hosting involves maintaining servers in nearby facilities linked via dedicated fiber, while high-speed network access often relies on optimized fiber or microwave connections between major trading hubs (for instance, London–Frankfurt or New York–Chicago routes). Even marginal reductions in latency—on the order of microseconds—can yield significant competitive advantages in markets where prices evolve continuously and across fragmented venues.
HFT strategies themselves vary in scope and complexity. Common approaches include electronic market making, where firms continuously quote bid and ask prices and manage risk within very short horizons; statistical and cross-venue arbitrage, which exploit fleeting price discrepancies between related instruments or exchanges; latency arbitrage, which reacts faster than competitors to public information or order book events; and smart order routing, which optimizes execution quality across multiple venues, sometimes also to capture fee rebates or queue priority. These strategies generate immense volumes of order messages and depend critically on precise market data, optimized software, and sophisticated risk controls.
From a regulatory perspective, MiFID II and MiFIR introduced a harmonized framework for algorithmic and high-frequency trading, elaborated through Regulatory Technical Standard 6 (RTS-6) [4], which defines organizational, risk-control, and testing requirements. As with all European directives, MiFID II provisions must be transposed into national law, and implementation may differ across jurisdictions. For example, Germany’s High-Frequency Trading Act [5] introduced an explicit authorization regime for HFT firms and added an additional criterion based on the speed of the connection, ensuring that firms not genuinely engaged in high-frequency activity would not fall under the same regulatory burden. These frameworks impose significant obligations on HFT firms, including mandatory system testing, kill switches, message-rate controls, record-keeping, and stringent resilience standards—all of which contribute to higher compliance costs and operational complexity.
Despite their efficiency benefits, HFT firms have attracted public and regulatory scrutiny. Critics argue that practices such as latency arbitrage or preferential access to dark pools create an uneven playing field, eroding confidence in market fairness. Michael Lewis’s Flash Boys [6] popularized this perception by portraying HFT as exploiting microscopic speed advantages to anticipate and “front-run” slower participants, contributing to the view that HFT is inherently predatory. Although many of these practices are fully compliant with market rules, they raise ethical and transparency questions, particularly in fragmented markets where access to speed and information is uneven.
Regulators are also concerned about the systemic implications of HFT. The Flash Crash of May 6, 2010 [7] illustrated how rapid, automated interactions among algorithms could amplify volatility and cause temporary dislocations. Two main risks are frequently cited [8] [9]: false liquidity—where HFT appears to provide deep liquidity in normal conditions but withdraws it abruptly during stress—and algorithmic herding, when many algorithms respond similarly to the same signals, creating self-reinforcing price dynamics. To mitigate these risks, regulators have implemented safeguards such as circuit breakers, message-rate limits, and system testing obligations.
Yet, it is important to balance these concerns against the structural role HFT plays in modern markets. By continuously arbitraging prices across venues, narrowing spreads, and enhancing price consistency, HFT contributes to market efficiency and liquidity provision under most conditions. The regulatory challenge is therefore not to constrain speed itself, but to ensure that technological advantages do not compromise fairness or stability. In this sense, HFT embodies both the promise and the tension of modern market microstructure: the pursuit of efficiency through automation, bounded by the need to maintain integrity and resilience in the face of ever-faster financial systems.