Risk management can be broadly understood as the systematic identification, measurement, mitigation, and monitoring of potential losses arising from trading activities (Hull, 2018). In traditional, discretionary trading—where human traders analyze market conditions and make judgment calls—risk arises primarily from human error, cognitive biases, and emotional decision-making (Kahneman & Tversky, 1979; Shefrin, 2007). By contrast, algorithmic trading replaces human discretion with automated rule-based execution. While this mitigates certain behavioral risks, it introduces new classes of model, technology, and operational risks (Taleb, 2010; Lopez de Prado, 2018).

The two primary objectives of this chapter are as follows:

• Explain why risk management is essential for algorithmic strategies, given their reliance on mathematical models, data integrity, and continuous operation in volatile markets.

• Highlight the key differences between discretionary and automated risk profiles, illustrating how each approach demands distinct controls, oversight, and mitigation techniques.

By the end of this chapter, readers will understand that while algorithmic trading can enhance efficiency and consistency, it also demands rigorous risk governance frameworks—often more stringent than those used in discretionary trading. Properly managing algorithmic risk is not merely a best practice but a necessity for ensuring long‐term viability, regulatory compliance, and preservation of investor capital.

2. Why Risk Management Is Essential for Algorithmic Strategies

Algorithms, by design, can operate continuously, process vast amounts of data, and execute large volumes of trades within fractions of a second (Aldridge, 2013; Cartea, Jaimungal, & Penalva, 2015). These characteristics confer significant advantages but simultaneously amplify potential losses if left unchecked. In this section, we examine the multiple facets of risk that make robust risk management indispensable for algorithmic trading.

2.1. The Nature of Algorithmic Trading and Its Risks

Definition and Scope. Algorithmic trading refers to any strategy in which orders are generated and executed by computer algorithms—ranging from simple rule‐based systems (e.g., moving average crossovers) to highly sophisticated high‐frequency trading (HFT) models that exploit microsecond price discrepancies (Kirilenko, Kyle, Samadi, & Tuzun, 2017). The capacity to process real-time market data, news sentiment, and technical indicators allows these algorithms to identify and exploit patterns across multiple markets simultaneously (Aldridge, 2013).

Risk Amplification. While humans can place dozens of trades a day at most, a single algorithm can execute thousands or millions of orders in the same timeframe. Consequently, a minor flaw in the algorithm’s logic or unexpected market event can rapidly generate outsized losses before human overseers become aware (Biais, Foucault, & Moinas, 2015). This “amplification effect” underlines why algorithmic systems require continuous health checks, real-time monitoring dashboards, and automated risk controls that halt or throttle trading under predefined conditions (Lopez de Prado, 2018; Hull, 2018).

2.2. Leverage, Speed, and Scale: Amplifiers of Risk

Leverage. Many algorithmic strategies—especially in futures, options, and currency markets—employ significant leverage to magnify returns. While leverage can increase profits during favorable market conditions, it equally magnifies losses. For instance, a 1% adverse price move on a position using 50:1 leverage results in a 50% loss. Without strict position‐sizing rules and automatic deleveraging mechanisms, a small market aberration can wipe out capital rapidly (Thakor & Ramadorai, 2015).

Speed and Scale. Algorithms operate on millisecond or microsecond timescales, far outpacing human reaction times. This speed enables them to exploit fleeting arbitrage opportunities but can also lead to “liquidity vacuums” if many algorithms react similarly (Hendershott, Jones, & Menkveld, 2011). In such scenarios, continuous liquid execution may suddenly evaporate, causing sharp price moves—exemplified by the May 6, 2010 Flash Crash, where algorithmic feedback loops drove the Dow Jones Industrial Average down nearly 1,000 points within minutes (Kirilenko et al., 2017).

Statistical Example: Return Distribution. Consider two trading approaches, each generating a modest 0.1% return per trade. A discretionary trader executing 10 trades per day with $100,000 capital risks a maximum of $500 per trade (0.5% stop‐loss). In contrast, an HFT algorithm executing 1,000 trades per day with $100,000 capital might risk $100 per trade to maintain the same daily risk budget. However, if the algorithm malfunctions or if market conditions radically shift, losses compound almost instantaneously across thousands of positions, outpacing any human intervention (Aldridge, 2013).

2.3. Model Risk and Overfitting

Model Risk Definition. Model risk arises when an algorithm’s underlying assumptions, mathematical equations, or parameters fail to represent real market behavior (Hull, 2018). In algorithmic trading, models are typically built on historical data. If historical market conditions drastically differ from current or future conditions, models can break down, leading to erroneous signals and large trading losses.

Overfitting. One pervasive form of model risk is overfitting—where a strategy is excessively tuned to historical “noise” rather than robust, persistent patterns. An overfitted model might perform exceptionally well on past data but fail miserably on new, unseen data sets (Bailey, Borwein, López de Prado, & Zhu, 2017). Overfitting can occur when too many parameters or indicators are incorporated without proper constraints or when in-sample optimization is prioritized over rigorous out-of-sample validation (Pardo, 2015).

• Quantitative Illustration. Suppose a strategy incorporates ten technical indicators (moving averages, Bollinger Bands, MACD, RSI, etc.) and optimizes the lookback periods for each parameter over a ten-year data set. If the model simply “fits” the historical data, it may inadvertently encode one‐off market events—such as the 2008 financial crisis—that are not repeatable. When deployed live, such a model may produce false positives or fail to adapt to normal market regimes (Bailey et al., 2017).

Mitigation Techniques.

• Cross‐Validation & Walk‐Forward Testing. Partitioning data into multiple in-sample and out-of-sample segments ensures strategies generalize across different time periods (Westra, 2019).

• Regularization. Incorporating penalties for model complexity (e.g., LASSO or Ridge regression) reduces the likelihood of overfitting (López de Prado, 2018).

• Robustness Metrics. Measuring sensitivity of strategy performance to slight changes in parameters helps identify fragile models (Bailey et al., 2017).

2.4. Technology Risk: Infrastructure, Latency, and Bugs

Infrastructure Dependence. Algorithmic trading demands robust hardware, software, and network connectivity. Any failure—server crashes, network latency spikes, or data feed interruptions—can lead to missed trades, orphaned orders, or unhedged positions (Kissell, 2014). Proprietary trading firms and hedge funds often build dedicated co-location facilities near exchange data centers to minimize latency. However, even microsecond delays can dramatically affect strategy performance, especially in high-frequency contexts (Aldridge, 2013; Kirilenko et al., 2017).

Software Bugs and Code Errors. Unlike discretionary traders, algorithmic systems execute exactly as coded—no more, no less. A seemingly innocuous programming error (e.g., an off-by-one error, improper data filtering, or miscalculated position sizes) can instantly translate into large financial losses (Kestner, 2003). In August 2012, Knight Capital Group suffered a $440 million loss in 45 minutes due to a faulty software deployment that erroneously routed orders (Reuters, 2012). The incident underscores how critical rigorous code testing, version control, and pre‐deployment simulations (“paper trading”) are for mitigating technology risk.

Latency and Slippage.

• Latency refers to the delay between sending an order and its execution. In high‐frequency contexts, even nanoseconds matter. Algorithms that cannot process market data and route orders with minimal latency risk being “picked off” by faster competitors, leading to adverse fills (Brogaard, Hendershott, & Riordan, 2014).

• Slippage occurs when an order is filled at a price different from the intended level. Variable liquidity and rapid price movements mean that algorithms must anticipate slippage, especially for large orders or in less liquid instruments (Kissell, 2014).

Mitigation Techniques.

• Stress Testing. Simulating trading through realistic “market replay” environments helps identify performance under stress.

• Redundant Systems. Implementing failover servers and multiple data feeds reduces the risk of single points of failure.

• Real-Time Monitoring. Automated health checks and “circuit breakers” that pause or shut down trading when key metrics (e.g., message queue lengths, execution delays) exceed thresholds.

2.5. Operational and Systemic Risks

Operational Risk. Encompasses any risk arising from inadequate or failed internal processes, people, or systems (Basel Committee on Banking Supervision, 2011). For algorithmic strategies, operational risk can manifest through:

• Manual Overrides. Human intervention to override algorithmic decisions, which can lead to unintended exposures.

• Miscommunication. Between developers, quants, traders, and IT teams—e.g., deploying code changes without adequate version control or notifying stakeholders.

• Third‐Party Dependencies. Reliance on external data vendors or cloud services can introduce vendor‐specific risks (e.g., price feed outages).

Systemic Risk. When many market participants rely on similar algorithms or models, correlated behaviors can exacerbate market stress (Sornette, 2003). The “Flash Crash” of May 6, 2010, demonstrated how multiple high‐frequency strategies reacting to a sudden price drop created a feedback loop, ultimately plunging indices before circuit breakers temporarily halted trading (Kirilenko et al., 2017). As a result, regulators have called for greater transparency in algorithmic strategies and enhanced market‐wide risk controls (U.S. Securities and Exchange Commission, 2010).

Mitigation Techniques.

• Diversity of Strategies. Incorporating uncorrelated models that react differently under stress.

• Centralized Risk Oversight. Developing enterprise‐level risk committees to oversee algorithmic deployments and enforce stress‐testing protocols.

• Regulatory Safeguards. Adhering to market‐wide circuit breakers, minimum resting times for orders, and kill‐switch mechanisms.

2.6. Regulatory and Compliance Imperatives

Regulatory Landscape. As algorithmic trading proliferated, regulators worldwide instituted rules to mitigate systemic risk, ensure fair markets, and protect investors. Key regulations include the European Union’s Market in Financial Instruments Directive II (MiFID II), which imposes stringent requirements on algorithmic trading firms—such as pre‐trade risk controls, testing protocols, and mandatory reporting of algorithmic strategies (European Securities and Markets Authority [ESMA], 2018). In the United States, the Securities and Exchange Commission (SEC) and Commodity Futures Trading Commission (CFTC) have implemented rules focusing on audit trails, kill switches, and transparency (SEC, 2010).

Compliance Requirements. Algorithmic firms must develop and document comprehensive governance frameworks, including:

• Pre‐Deployment Testing. Stress tests, regressions, and backtests under a range of market conditions.

• Operational Controls. Real‐time surveillance systems that flag abnormal trading patterns or performance metrics.

• Record‐Keeping. Detailed logs of algorithmic code versions, parameter changes, and trading decisions.

• Kill Switches. Procedures for immediate deactivation of algorithms in emergencies.

Failure to comply can result in fines, trading suspensions, or reputational damage. For example, in 2015, the SEC fined a major bank $16.7 million for failing to implement adequate controls over its algorithmic trading (SEC, 2015).

Balance Between Innovation and Regulation. While robust risk controls are critical, overly prescriptive regulations risk stifling innovation. The challenge for both firms and regulators is to create frameworks that allow algorithmic strategies to evolve while maintaining systemic stability and investor protection. Multi‐jurisdictional coordination—especially important for global trading firms—helps harmonize requirements across markets and reduces regulatory arbitrage.

3. Differences Between Discretionary and Automated Risk Profiles

Discretionary and automated (algorithmic) trading represent two fundamentally different approaches to market participation. Each approach carries its own set of risks and risk management requirements. This section systematically compares the risk profiles of discretionary versus automated trading in several dimensions: decision‐making processes, emotional factors, speed of execution, sources of error, and risk controls.

3.1. Decision-Making: Human Judgment vs. Systematic Rules

Discretionary Trading. In discretionary trading, human traders analyze market data—both quantitative (charts, indicators, order book) and qualitative (news, economic outlook, geopolitical events)—before making judgment calls. Humans can flexibly adapt to unforeseen circumstances, interpret ambiguous signals, and exercise contextual reasoning (Loewenstein, Rick, & Cohen, 2008). However, this flexibility introduces variability: two traders given identical data may make opposite decisions based on experience, intuition, or risk tolerance.

• Advantage: Adaptability to unexpected events (e.g., Black Swan events).

• Risk: Inconsistency due to different interpretations, cognitive biases, and fatigue (Kahneman & Tversky, 1979).

Automated Trading. Automated systems rely strictly on preprogrammed rules and parameters. The decision tree is explicit: if Condition A is met (e.g., price crosses moving average), then execute Action B (e.g., buy 1,000 shares). This removes discretionary variability, ensuring reproducibility and consistency (Aldridge, 2013). However, it also means algorithms cannot adapt in real time to novel events outside their programmed scope.

• Advantage: Consistency and elimination of emotional overtrading.

• Risk: Inflexibility and inability to interpret unstructured data unless explicitly programmed (e.g., “If SEC announces rate hike, exit positions”). Overreliance on historical patterns can lead to model breakdown under unprecedented conditions (Taleb, 2010).

Illustrative Example. A discretionary trader might choose to “sit out” during extreme volatility (e.g., Brexit vote uncertainty), while an algorithm without such provisions might continue executing according to technical rules, potentially incurring substantial losses—or, conversely, missing an opportunity to buy during a panic‐driven dip.

• 3.2. Emotional and Behavioral Factors

Discretionary Trading Biases. Human traders are subject to well‐documented cognitive biases, such as:

• Confirmation Bias. Seeking information that supports existing beliefs.

• Overconfidence. Overestimating one’s ability to predict price movements (Barber & Odean, 2001).

• Loss Aversion. Reluctance to realize losses, potentially leading to “holding losers too long” (Kahneman & Tversky, 1979; Shefrin & Statman, 1985).

• Herd Behavior. FOMO (Fear of Missing Out) and following crowd sentiment (Shiller, 2000).

These biases lead to inconsistent position sizing, erratic entries/exits, and a tendency to “chase” prices. Emotional decision‐making can amplify drawdowns and diminish risk‐adjusted returns (Loewenstein et al., 2008).

Automated Trading: Behavioral Elimination & New Risks. By codifying rules, algorithms remove emotional biases entirely—each trade is executed mechanically when criteria are met. However, behavioral risk merely shifts from decision‐making to oversight. Analysts and developers may exhibit:

• Overreliance on Historical Performance. Assuming backtest success guarantees future profitability (Bailey et al., 2017).

• Confirmation Bias in Model Development. Selecting strategies or parameters that confirm preconceived notions (Bailey et al., 2017).

• Illusion of Control. Belief that algorithmic strategies “guarantee” returns, leading to underestimating systemic or technical risks (Langer, 1975).

Ultimately, while algorithms eliminate in‐trade emotional errors, human oversight is still necessary to guard against complacency, overconfidence, and sloppy validation.

3.3. Speed of Execution and Monitoring

Discretionary Trading. Human traders might require seconds or minutes to place orders, especially for large or complex positions. Slower reaction times can be advantageous when avoiding microstructural noise, but can also lead to missed opportunities in fast markets (Hendershott, Jones, & Menkveld, 2011).

• Risk Point: Inability to capture fleeting arbitrage, and vulnerability to slippage when reacting to fast‐moving prices.

Automated Trading. Algorithms can ingest market data, generate signals, and route orders within microseconds or milliseconds. This speed enables:

• High‐Frequency Strategies. Where profitability depends on capturing tiny spreads (<0.01%) (Brogaard, Hendershott, & Riordan, 2014).

• Minimized Slippage. When algorithms use smart order routers and colocation to reduce latency, slippage can be substantially lower than manual orders (Aldridge, 2013).

However, speed also poses risks:

• Amplified Losses. If an algorithm misfires, losses accumulate at high speed (e.g., Knight Capital, 2012).

• Volume‐Based Market Impact. Large algorithms trading quickly can move prices against themselves, increasing implementation shortfall (Kissell, 2014).

Monitoring Considerations. Discretionary traders can visually monitor market screens, news feeds, and execute “gut checks” when unusual events occur. Automated systems require dedicated real‐time monitoring dashboards, automated alerts, and fail‐safe mechanisms to halt trading under suspicious conditions (e.g., excessive drawdown in a single hour).

3.4. Sources of Error: Cognitive vs. Technical

Discretionary Trading Errors.

• Analysis Paralysis. Overthinking leads to indecision.

• Erroneous Execution. Manual order placement mistakes (wrong ticker, wrong quantity).

• Fatigue and Distraction. Mental and physical exhaustion impact judgment.

Automated Trading Errors.

• Code Bugs. Logic errors, data‐type mismatches, rounding errors.

• Data Quality Issues. Incorrect or stale market data feeding the algorithm can cause misfires.

• Parameter Misconfiguration. Mistakes in parameter files (e.g., wrong volatility threshold) can trigger unintended trades.

• Integration Failures. Discrepancies between backtest (simulated environment) and live market interfaces (differences in API responses, connectivity issues).

Risk Management Implication. In discretionary trading, education, experience, and structured workflows help mitigate cognitive errors. In automated trading, rigorous software development life‐cycle (SDLC) practices—unit testing, integration testing, code reviews, continuous integration—are essential to minimize technical errors (Lopez de Prado, 2018; Kissell, 2014).

3.5. Risk Controls and Oversight Mechanisms

Discretionary Trading Controls.

• Position Limits. Human traders are assigned maximum position sizes per instrument.

• Stop‐Loss Orders. Often manually placed or set within trading platforms.

• Trader Surveillance. Supervisors and compliance monitor trading logs and P&L for unusual activity.

• Periodic Reviews. Desk audits and trader performance assessments.

Automated Trading Controls.

• Pre‐Trade Risk Checks. Every order is validated against position limits, maximum notional amounts, and market risk thresholds before routing (ESMA, 2018).

• Kill Switches. Software modules that immediately halt trading if specific conditions are met (e.g., cumulative losses exceed a threshold, connectivity failure).

• Real‐Time Dashboards. Automated alerts for anomalies: order‐to‐trade ratios, latency spikes, or erroneous fills.

• Multi‐Layered Oversight. Involvement of risk managers, quants, developers, and compliance forming “Algo Governance Committees” to review changes and ensure adherence to policies (ESMA, 2018; SEC, 2015).

Example: Pre-Trade Risk Engine. Before each order is sent to the exchange, a risk engine verifies:

• Position Limit Check. Current position + proposed trade ≤ limit.

• Order Rate Check. Orders per second/minute are within allowed thresholds.

• Notional Check. Single‐order notional value ≤ maximum allowed.

• Liquidity Check. Estimated market impact cost falls below predefined benchmark.

If any check fails, the order is either modified (e.g., reduce size) or canceled. Discretionary desks might rely on traders’ judgment for similar controls, but automation is both faster and more consistent, particularly under stress.

4. Conclusion

Algorithmic trading presents unparalleled opportunities to exploit market inefficiencies with speed, precision, and consistency. However, these same characteristics amplify potential risks—making risk management not merely an adjunct but a foundational requirement for sustainable algorithmic operations. Key takeaways from this chapter include:

• Risk Amplification through Speed and Leverage. While discretionary traders can respond to market volatility within seconds or minutes, algorithms execute thousands of orders in microseconds. Even a small programming error or sudden market shift can result in outsized losses before human intervention (Aldridge, 2013; Kirilenko et al., 2017).

• Unique Model and Technology Risks. Unlike discretionary traders who rely on qualitative judgment, algorithmic strategies depend on mathematical models and technological infrastructure. Model risk—stemming from overfitting and historical data limitations—can quickly undermine performance if not rigorously validated via cross‐validation, walk‐forward testing, and stress tests (Bailey et al., 2017; López de Prado, 2018). Simultaneously, technology failures (bugs, latency, connectivity) pose systemic threats requiring redundant infrastructure and real‐time health checks (Knight Capital, 2012; Kissell, 2014).

• Distinct Risk Profiles: Discretionary vs. Automated. Whereas discretionary trading is prone to cognitive and emotional biases, algorithmic trading introduces technical vulnerabilities. Effective oversight in discretionary settings focuses on supervision and compliance checks, while automated environments demand continuous monitoring dashboards, automated “kill switches,” pre-trade risk engines, and multi-layered governance (ESMA, 2018; SEC, 2015).

• Critical Role of Regulatory Compliance. Global regulators have instituted stringent rules—such as MiFID II in Europe and similar SEC/CFTC mandates in the U.S.—to curb systemic risk, increase transparency, and protect market integrity (ESMA, 2018; SEC, 2010). Algorithmic firms must balance innovation with regulatory adherence, ensuring every code release, parameter change, and trading anomaly is documented, reviewed, and reported where necessary.

In sum, rigorous risk management is not optional for algorithmic trading—it is essential. The next chapters will build upon this foundation by exploring advanced risk metrics (e.g., Value at Risk, drawdowns), position sizing methodologies, stop-loss frameworks, budgeting considerations, and automated risk controls. By integrating these disciplines, algorithmic traders can navigate the complexities of modern markets while safeguarding capital and maintaining regulatory compliance.

5. References

Aldridge, I. (2013). High-Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems (2nd ed.). Wiley.

Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2017). The Probability of Backtest Overfitting. Journal of Computational Finance, 20(4), 39–69.

Basel Committee on Banking Supervision. (2011). Principles for the Sound Management of Operational Risk. Bank for International Settlements.

Biais, B., Foucault, T., & Moinas, S. (2015). Equilibrium Fast Trading. Journal of Financial Economics, 116(2), 292–313.

Black, F., & Scholes, M. (1973). The Pricing of Options and Corporate Liabilities. Journal of Political Economy, 81(3), 637–659.

Brogaard, J., Hendershott, T., & Riordan, R. (2014). High-Frequency Trading and Price Discovery. Review of Financial Studies, 27(8), 2267–2306.

Cartea, Á., Jaimungal, S., & Penalva, J. (2015). Algorithmic and High-Frequency Trading. Cambridge University Press.

Chan, E. (2013). Algorithmic Trading: Winning Strategies and Their Rationale. Wiley.

European Securities and Markets Authority (ESMA). (2018). MiFID II / MiFIR: Final Report on Implementing Technical Standards.

Hendershott, T., Jones, C. M., & Menkveld, A. J. (2011). Does Algorithmic Trading Improve Liquidity? Journal of Finance, 66(1), 1–33.

Hull, J. C. (2018). Risk Management and Financial Institutions (5th ed.). Wiley.

Kahneman, D., & Tversky, A. (1979). Prospect Theory: An Analysis of Decision under Risk. Econometrica, 47(2), 263–291.

Kirilenko, A. A., Kyle, A. S., Samadi, M., & Tuzun, T. (2017). The Flash Crash: The Impact of High-Frequency Trading on an Electronic Market. The Journal of Finance, 72(3), 967–998.

Kestner, L. N. (2003). Quantitative Trading Strategies: Harnessing the Power of Quantitative Techniques to Create a Winning Trading Program. McGraw-Hill.

Kessler, R. (2014). The Science of Algorithmic Trading and Portfolio Management. Academic Press.

Kissell, R. (2014). The Science of Algorithmic Trading and Portfolio Management. Academic Press.

Knight Capital Group. (2012). Internal Incident Report (public summaries).

Kuhnen, C. M., & Knutson, B. (2005). The Neural Basis of Financial Risk Taking. Neuron, 47(5), 763–770.

Kuhnen, C. M., & Knutson, B. (2005). The Neural Basis of Financial Risk Taking. Neuron, 47(5), 763–770.

Langer, E. J. (1975). The Illusion of Control. Journal of Personality and Social Psychology, 32(2), 311–328.

Lo, A. W., Repin, D. V., & Steenbarger, B. N. (2005). Fear and Greed in Financial Markets: A Clinical Study of Day-Traders. American Economic Review, 95(2), 352–359.

Lopez de Prado, M. (2018). Advances in Financial Machine Learning. Wiley.

Loewenstein, G., Rick, S., & Cohen, J. D. (2008). Neuroeconomics. Annual Review of Psychology, 59, 647–672.

Pardo, R. (2015). The Evaluation and Optimization of Trading Strategies (2nd ed.). Wiley.

Reuters. (2012, August 1). Knight Trading’s $440 Million Software Glitch. Reuters. Retrieved from https://www.reuters.com

Shiller, R. J. (2000). Irrational Exuberance (2nd ed.). Princeton University Press.

Shefrin, H. (2007). Beyond Greed and Fear: Understanding Behavioral Finance and the Psychology of Investing (Revised ed.). Oxford University Press.

Shefrin, H., & Statman, M. (1985). The Disposition to Sell Winners Too Early and Ride Losers Too Long: Theory and Evidence. The Journal of Finance, 40(3), 777–790.

Sornette, D. (2003). Why Stock Markets Crash: Critical Events in Complex Financial Systems. Princeton University Press.

Taleb, N. N. (2010). The Black Swan: The Impact of the Highly Improbable (2nd ed.). Random House.

Thakor, A. V., & Ramadorai, T. (2015). Micro‐Market Architecture and Trade Execution: Evidence from “Trade‐At” Orders. Journal of Financial and Quantitative Analysis, 50(6), 1235–1267.

Tversky, A., & Kahneman, D. (1974). Judgment under Uncertainty: Heuristics and Biases. Science, 185(4157), 1124–1131.

U.S. Securities and Exchange Commission (SEC). (2010). Findings Regarding the Market Events of May 6, 2010. Retrieved from https://www.sec.gov

U.S. Securities and Exchange Commission (SEC). (2015). SEC Charges Major Bank with Algorithmic Trading Controls Failures. Retrieved from https://www.sec.gov

Westra, A. (2019). Building Winning Algorithmic Trading Systems: A Trader’s Journey from Data Mining to Monte Carlo Simulation to Live Trading (2nd ed.). Wiley.