Search icon
ReadWrite
see notifications
Notifications
see more
    paint-brush
    How Useful are Computational Models to Traders in Their Decision Making? by@everyone
    458 reads
    458 reads

    How Useful are Computational Models to Traders in Their Decision Making?

    by __nikApril 4th, 2024
    Read on Terminal Reader
    Read this story w/o Javascript
    tldt arrow

    Too Long; Didn't Read

    An article on the history, present and future of Machine Learning in trading.
    featured image - How Useful are Computational Models to Traders in Their Decision Making?
    __nik HackerNoon profile picture

    The world of trading has changed significantly since the creation of Stock Exchanges, and this transformation is due to developments and advancements in technology. Places like the London Stock Exchange and New York Stock Exchange have become concepts, as trading no longer requires people to be physically present - trades went from taking up to a week to being placed in fractions of seconds from all over the world. (bebusinessed, 2015) Real-time information is now available to anyone with a connection to the internet. Breaking news about the economy and how the market is moving can be received as almost instant notifications to the trader's phone. In a system where time is money, seconds and even fractions of seconds can result in huge losses or gains, and computers can eradicate these minor lags in buying or selling stock. In the past traders had to call brokers to place trades, now these can be executed with a few clicks of a button or even automatically. This increase in speed with which traders must make crucial decisions has resulted in a highly competitive environment relying heavily on incredibly fast computers and software.

    The prevalence of algorithmic trading

    Whilst one of the main changes to the environment of trading is the speed and magnitude at which trades can be placed, another is who, or rather what is placing those trades. According to (Quantified Strategies, 2022) around 65-70% of the total trading volume of the US equity market, European financial markets, and major Asian capital markets, are algorithmic trading accounts. The growth of algorithmic trading can be credited to several factors, some of which are advancements in technology, the increasing availability of market data, and the growing competition among financial institutions. Algorithmic trading has also become more accessible to individual traders and investors, thanks to the rise of online trading platforms and the availability of low-cost computing resources.


    Algorithmic trading has also expanded beyond Proprietary trading and the Debt Market; It is now being used in Forex markets, commodity markets, and cryptocurrency markets. This expansion has led to the development of algorithms designed specifically for these markets and resulted in more traders using algorithmic trading. (Analyzing Alpha, 2021)

    To understand the extent of usefulness of computational models to traders, it is important to understand how these computational models work, their role in algorithmic trading and HFT, and the different types of computational models. It is also helpful to know what mathematical models the computational models are based on as this will help in understanding the uses and limitations of these models. This will help when evaluating how useful these models and algorithmic trading are to traders.

    What is Algorithmic Trading?

    Systematic trading and its use of algorithms to make trading decisions

    Trading methods can be split into several types, one of which is systematic trading. Systematic trading is a type of trading that involves using a pre-determined set of rules or algorithms to make trading decisions. This approach is often used by quantitative traders, who use statistical and mathematical models to analyse market data and identify trading opportunities. Traders use a variety of mathematical models (including statistical and machine learning models, and data mining techniques) to analyse large amounts of market data and identify trading opportunities. These models can be categorized into two main types, Fundamental and Technical analysis models.


    Fundamental Analysis Models use financial and economic data to determine the intrinsic values of securities, such as stocks or bonds. Fundamental analysis looks at factors like financial statements, economic indicators, and industry trends so that traders can evaluate a company's performance and financial health. This evaluation can then be used to make investment decisions. (Sahakian, Cuzzolin and Buczynski, 2022)


    Technical Analysis Models predict future price movements by analysing market data and detecting patterns and trends in the data. These can be patterns in price movements, volume, and other market indicators. Based on the patterns they spot; traders can identify trading opportunities and make buy-and-sell decisions to make a profit. (Majaski, 2019)

    How systematic trading can be automated

    Traders can use these fundamental and technical analysis models to develop trading algorithms that automatically perform trades based on market conditions. Systematic trading can be used in a range of asset classes, for example, currencies, fixed income, equities, and commodities. Models used in systematic trading detect market inefficiencies or other anomalies through quantitative analysis and use this to their advantage to generate consistent returns over time.


    Systematic traders use backtesting and simulation to refine their models and evaluate the performance of their strategies under different market conditions. This allows them to check the performance of their models and develop or improve them.

    This method of training can be fully automated or can be implemented to be semi-automated (still needing some human intervention) due to its mathematical nature; this is to remove emotions from the decision-making process and rely on objective data analysis.

    Asset classes where systematic trading can be used


    Systematic trading can be used in various asset groups, including:


    Stocks

    The stock (or equity) market uses strategies such as trend following, mean reversion, and statistical arbitrage can be employed to identify opportunities in the stock market. (Abdelmalek, 2021)


    Futures:

    Futures markets provide liquidity and volatility, which makes them ideal for systematic strategies so trend following, and momentum strategies are commonly used in futures trading. (Hull, 2022)


    Options

    Options trading can also be automated using complex strategies and modelling.


    Forex

    The Forex market is the largest market in the world. Strategies such as trend following (takes advantage of trends (Mitchell, 2021)) and carry trade (which is investing in high-return assets using money borrowed at low interest) could be used. (Toptal Engineering Blog, n.d.)  (Chen, 2020)


    Commodities

    Strategies such as trend following, mean reversion, and seasonality can be employed in commodity markets.


    Cryptocurrencies

    With the increasing popularity of cryptocurrencies, systematic trading has gained traction in this asset. For example, momentum, mean reversion and statistical arbitrage strategies can all be used when trading crypto.


    (Quantitative Finance & Algo Trading Blog by QuantInsti, 2020)

    The use of backtesting and simulation to refine trading models

    Backtesting and simulation are important tools that traders use to test and refine their trading models.


    Backtesting involves testing a trading strategy or model on historical market data to see how it would perform under different market conditions and make adjustments to improve its performance.


    Simulation uses a virtual environment to run trading model. The virtual environment will mimic the behaviour of the actual market. Traders can test their models in a range of scenarios and vary market conditions to identify weaknesses in their strategy and refine their model accordingly. This allows traders to test their strategy in real-time market conditions without risking real funds.


    Both of these methods allow traders see how their models perform under different market conditions; they can then evaluate how well their strategy works and its accuracy in predicting the market's movement and refine them to increase accuracy and profitability. These tests can also allow traders to identify potential risks and adjust their strategies to better manage those risks.

    III. Other Trading Methods

    Discretionary trading, event-driven trading, sentiment trading, and manual trading

    Apart from systematic and algorithmic-based trading, there are several other methods, including:

    • Discretionary trading which is making trading decisions based on the trader's personal judgment and experience, instead of using a systematic approach. Discretionary traders often use technical analysis, fundamental analysis, and market intuition to make decisions.
    • Event-driven trading involves using market inefficiencies that arise due to specific events to the traders' advantage to make a profit. For example: mergers, acquisitions, earnings releases, and economic data releases.
    • Sentiment trading is when traders take positions based on the sentiment of the market (or specific stocks), which can be measured through social media and news sentiment, or other sentiment indicators.
    • Manual trading is placing trades manually, without the use of any automated systems or algorithms. This approach is still used by many traders today.

    IV. Quantitative Trading, HFT (High-Frequency Trading), and Algorithmic Trading

    Quantitative trading and algorithmic trading

    Quantitative trading, HFT and Algorithmic Trading fall under the category of systematic trading. There are key differences between these two methods, from how decisions are made to who executes them.


    Algorithmic trading is when a computer follows a set of instructions or specifications (an algorithm) to place trades or make decisions about trades. The algorithm will do this based on various factors such as timing, price, and quantity. Quantitative trading uses mathematical models to determine what trades to place. In contrast, algorithmic models use technical analysis on chart analysis and data from exchanges to find new positions. When using quantitative trading, the traders will often look at price and volume. Another difference between these two types of trading is that algorithmic trading software will automatically execute trades, however, some quantitative trading methods rely on the trader to manually open positions. (IG, n.d.)


    Quantitative traders use mathematical models such as the Black-Scholes model and the Monte Carlo Simulation to make decisions about trading. These models can be implemented as algorithms and used in algorithmic trading. (Agarwal, 2019)


    The Black-Scholes Model is a differential equation used to price European options contracts (and other derivatives). The model assumes that the price of the underlying asset follows a random walk which, according to Vinayak Wagh, Dattatray Raut and Dilip Bera (2022), is “a random process that describes a path that consists of a succession of random steps on some mathematical space".


    The Black-Scholes Model also makes the assumption that options can only be exercised at a certain time, so this model cannot be used for American options, which can be exercised before their expiration date. American Options pricing uses the Bjerksund-Stensland model instead. The Black-Scholes model considers several variables when pricing options. These include the price of the underlying asset, the option's strike price, the time to expiration, the risk-free interest rate, and the volatility of the underlying asset's price. (Kenton, 2019)


    The Black-Scholes model uses the following formula to calculate the theoretical price of a European call or put option:


    (Images from Option Party. (2016). Black Scholes Formula Explained. [online] Available at: https://optionparty.com/black-scholes-formula-explained/.)

    Where:

    • C is the theoretical price of a call option
    • P is the theoretical price of a put option
    • S is the current price of the underlying asset
    • K is the option's strike price
    • r is the risk-free interest rate
    • t is the time to expiration of the option, expressed as a fraction of a year
    • N() is the cumulative normal distribution function


    d1 and d2 are calculated as follows:

    • d1 = (ln(S/K) + (r + σ^2/2)t) / (σsqrt(t))
    • d2 = d1 - σ*sqrt(t)


    Where σ is the volatility of the underlying asset's price. (SOAS, n.d.)


    Like all models, the Black-Scholes model makes assumptions, as not all factors can be included. For example, the Black-Scholes model assumes that the market is efficient meaning the model assumes that the price of an option reflects all available information about the underlying asset; The model also assumes that there are no transaction costs or restrictions on trading. According to Downey, (2021) transaction costs are "expenses incurred when buying or selling a good or service".


    The Black-Scholes model is widely accepted in finance, but it isn't perfect, it does have limitations. One of the limitations is that the model assumes that the distribution of the underlying asset's returns follows normal distribution, which isn't the case every time. Another assumption that the model makes is that the volatility of the underlying asset's price is constant over time, but this isn't always true. However, the Black-Scholes model is an important tool for pricing options and is the foundation for many other option pricing models. (Ganti, 2021)


    The Monte Carlo Method, a multiple probability simulation, is used to price options by generating many random trajectories for the price of an underlying asset. Each trajectory is associated with a certain payoff, which is subsequently discounted to the present; then an average is taken to determine the option's price. Monte Carlo simulation is a tool that can help traders make investment decisions by providing data about the potential risks and returns associated with different financial instruments or investment strategies. (BARILLAS and SHANKEN, 2018)


    In trading, Monte Carlo simulation can be used for a range of purposes, from risk management to options pricing; In Risk Management the simulation can model the potential risk associated with a financial instrument or portfolio. (Agarwal, 2019) Traders can simulate various market scenarios and analyse the probability of the portfolio's value dropping below a certain threshold, allowing them to manage the risk associated with their investments. In Option Pricing, the Monte Carlo simulation can be used to price options (which are financial instruments that give the holder the right to buy or sell an underlying asset at a predetermined price), by simulating different market scenarios, traders can estimate the value of an option and its sensitivity to changes in market conditions. (Ganti, 2021)


    The Monte Carlo simulation can also be used in Portfolio Optimization to optimize a portfolio of financial instruments; traders can identify the optimal portfolio allocation that maximizes expected returns while minimizing risk, by simulating different asset allocations and investment strategies (Kenton, 2019b).

    Algorithmic trading can be considered a subset of quantitative trading

    Quantitative trading refers identifying trading opportunities where a profit can be made, specifically using mathematical models and statistical analysis; It involves the use of data analysis techniques, for example statistical arbitrage, trend analysis, and machine learning, can all be used to analyse market data and help make trading decisions.


    Algorithmic trading can be considered a subset of quantitative trading in cases when the algorithm uses quantitative models and implement them as computer programs that automatically execute trades based on price, timing, and quantity, because it involves the use of mathematical models and algorithms to make trading decisions. Algorithmic trading can execute trades automatically according to predetermined rules, or when certain conditions are met. (Finance Train, n.d.)

    Types of algorithmic trading

    There are several types of algorithmic trading. For example, execution algorithms and Arbitrage algorithms.


    Execution Algorithms execute trades in the most efficient and cost-effective manner possible - they break up large orders into smaller ones and execute them over time. This reduces market impact and minimizes transaction costs. Arbitrage Algorithms seek to take advantage of price discrepancies between different markets or securities. For example, an arbitrage algorithm may identify a price difference between a stock in the UK, and the same stock trading on a foreign exchange. It will then execute trades to profit from the price difference.


    Other examples include market making, trend following, statistical arbitrage and sentiment analysis algorithms. (Droid, 2018)


    Market Making Algorithms constantly place buy and sell orders to provide liquidity to the market, facilitating trading. These algorithms are often used by market makers, who earn profits from the bid-ask spread. Trend Following Algorithms are algorithms that seek to identify and profit from market trends; they may use technical indicators such as moving averages or chart patterns to identify trends and execute trades in the direction of the trend. Statistical Arbitrage Algorithms use statistical models to identify mispricing in related securities, such as two stocks in the same industry or a stock and its futures contract. The algorithm will buy the underpriced security and sell the overpriced security, seeking to profit from the convergence of their prices. Sentiment Analysis Algorithms use natural language processing and machine learning techniques to analyse news articles, social media posts, and other sources of information to gauge market sentiment. Based on this information, they make trading decisions. (Forex Broker Review, 2022)

    HFT and its use of electronic trading tools and high-frequency financial data to execute trades

    A subset of algorithmic trading is High Frequency Trading (HFT). HFT is characterized by rapid speed, high turnover rates, and a high ratio of orders to trades. It requires electronic trading tools and high-frequency financial data to execute trades, so powerful computer algorithms and mathematical models to analyse market data and identify trading opportunities are used; these algorithms can execute trades within milliseconds or even microseconds. (Aquilina, 2017)


    HFT traders rely on incredibly fast networks and advanced data centre infrastructure to achieve high speeds that are designed to reduce latency and to ensure that trades are executed as efficiently as possible; they use advanced data analytics techniques to process extensive amounts of financial data in real-time and to recognise patterns or trends. (CHABOUD et al., 2014)


    HFT traders often focus on arbitrage opportunities. Arbitrage opportunities are when traders buy and s sell securities in different markets to profit from price discrepancies. HFT traders also engage in market-making, which involves providing liquidity to markets by continuously buying and selling securities, thereby reducing bid-ask spreads, and improving market efficiency (Harford, 2012). HFT has become an important part of modern financial markets, providing liquidity and efficiency to markets while also generating significant profits for traders and firms that are able to leverage its advanced technology and sophisticated trading strategies. High Frequency Trading has been the subject of controversy and criticism, with concerns about their impact on market stability and fairness. (Breckenfelder, 2020) (CHABOUD et al., 2014)


    HFT is increasing in popularity, making up to about “50% of trading volume in US equity markets and between 24% and 43% in European equity markets” (Breckenfelder, 2017).

    Types of HFT

    There are several types of high-frequency trading (HFT) strategies, for example market making; this strategy involves quoting buy and sell prices for a particular security on an exchange. Market makers earn a profit by buying low and selling high, capturing the spread between the bid and ask price.


    Another type is Statistical Arbitrage, a strategy involving analysing statistical relationships between different securities to identify opportunities for profit (for example, if two highly correlated stocks diverge in price, a statistical arbitrageur may short the overvalued stock and buy the undervalued stock, hoping to capture the price difference when the stocks converge again).


    Event-Driven Trading is also a HFT strategy. Event-Driven traders use news and other market data to anticipate and profit from specific events (such as earnings announcements, mergers and acquisitions, or regulatory changes). (BAJPAI, 2022)


    Trend Following is when traders analyse price trends and buy or sell a security based on the trend's direction. If a stock has been rising steadily over the past few days, a trend-following algorithm may buy the stock with the expectation that it will continue to rise.


    Scalping involves making very short-term trades, often holding a position for just a few seconds or less and aiming to profit from small price movements. These algorithms use various signals to make trading decisions, like order flow or price momentum. (BAJPAI, 2022)

    V. Advantages of Algorithmic Trading

    Algorithmic trading removes human emotions and biases

    One advantage of algorithmic trading is that it removes the involvement of human emotions and the biases that come with it. (Seth, 2019a) Algorithms can make decisions quicker and with more accuracy by removing human emotion and error from the decision-making process through using complex algorithms that process vast amounts of market data in real-time. Algorithmic trading follows a set of rules and doesn’t hesitate based on fear. Computers can respond rapidly to changes in the market; the automated systems can generate new trades as soon as the algorithms criteria is met. This is crucial as trades are made in milliseconds and any latency can cost huge capital. This also prevents traders from either backing out due to fear of an unstable market or from becoming overconfident and trying to get a larger profit. Algorithmic trading ensures the traders follow the trade model. This can lead to more objective and consistent trading decisions. (Seth, 2019a)

    Backtesting trading models and the use of fixed rules

    Other advantages include the ability to back test. Whilst the models that traders form could be back tested it is harder to manually test on large quantities of data, computers can do this very quickly as they have a large processing power. The automated models have fixed rules, it is easy to check against historical data compared to traders who make decisions based on hunches or experience.


    Backtesting is applying a trading strategy to historical market data to evaluate how the strategy would have performed in the past; this enables traders to assess potential profitability of their strategies and identify weaknesses to refine the model. (Dave, 2022)


    Simulation is when a model is tested on simulated market data in order to assess its performance under a variety of scenarios. This way traders can ensure that they are not overfitting to historical data. (Solberg, 2021)


    Backtesting and simulation are crucial components in refining trading models for systematic trading. These methods allow traders to test the effectiveness of their trading strategies before putting real capital at risk. By testing and refining their strategies in this way, traders can increase their chances of success and reduce their exposure to risk. Another way traders can reduce risk is by programming algorithms to manage risk more effectively by automatically executing stop-loss orders, limiting exposure to certain assets, and managing position sizes based on market volatility. (Deloitte United States, 2017) (Picardo, 2019)

    Algorithmic trading can make trading decisions faster and more accurately

    Algorithmic trading systems can analyse market data in real-time, identify patterns, and execute trades in milliseconds, faster than human traders can react. This allows algorithmic traders to make more profitable trades by taking advantage of market movements before other traders can react. Algorithms can carry out more accurate trading decisions based on data rather than intuition. (Corporate Finance Institute, n.d.) (Nomad, 2023)


    Algorithms can be used to help traders diversify; since they are automated, they can be implemented in multiple accounts to increase how many trades executed per day. Another advantage of the use of multiple accounts is traders can test several different strategies simultaneously.

    VI. Disadvantages of Algorithmic Trading

    Market Volatility

    What is market volatility

    The frequency and magnitude of variation in the price of a financial instruments like stocks, bonds, commodities, and currencies over a certain period of time is known as market volatility. (Abdelmalek, 2021)


    There are a range of factors that can increase volatility; high volatility can increase the risk of significant losses, so investors and traders use risk management strategies to limit exposure to volatility; Examples of risk management strategies include diversification, stop-loss orders, and hedging with options or futures contracts. Volatility could also be used to a trader's advantage. (Angel One, 2023)


    Market volatility can be measured using statistical measures such as standard deviation or beta. Standard deviation is used to measure how much the price of an asset varies from its average price over a length of time, and beta measures how much an asset's price moves in comparison to benchmark indexes (like the S&P 500 index). (Ashford, 2021)

    How algorithmic trading can contribute to market volatility

    Algorithmic trading can contribute to market volatility by amplifying market moves, encouraging herding behaviour, impacting illiquid markets, and contributing to flash crashes. While algorithmic trading poses risks to market stability, this means careful risk management is needed to minimize these risks. (Fries, 2023)


    Algorithmic trading systems amplify market moves by automatically executing large volumes of trades based on predetermined rules or conditions. When a significant price movement occurs, algorithms can quickly detect it and execute trades, which can lead to a cascade of further trades and exacerbate the initial move. (Arumugam, Prasanna and Marathe, 2023)


    This type of trading can also contribute to herding behaviour; herding behaviour is where multiple traders all make the same trading decisions simultaneously due to using similar algorithms, leading to a sudden surge in buying or selling activity and increased volatility.


    Algorithmic trading systems can also cause increased volatility in illiquid markets; these markets have fewer buyers and sellers, and small trades can have significant impacts on the prices of assets. Algorithmic trading may be more likely to trigger stop-loss orders or sudden price swings in illiquid markets too. They can also contribute to flash crashes, where the market experiences a rapid and severe drop in price followed by a rapid recovery. These events can occur when algorithms detect a sudden change in market conditions and respond by selling large volumes of assets, which can trigger further selling and lead to a downward spiral. (Fries, 2023)

    Examples of market crashes caused by algorithmic trading

    There have been several flash crashes and market crashes that have been connected to algorithmic trading; some of these crashes were the May the 6th 2010 crash, the August 24th, 2015, crash, the February 5th, 2018, crash (Dow Jones Industrial Average Sell-Off), and September 3rd, 2020, crash (S&P 500 Futures Sell-Off). Algorithmic high-frequency trading poses a significant risk to the financial system as algorithms operating across markets can quickly transmit shocks from one market to the next, amplifying systemic risk. The Flash Crash of May 2010 is an example of this risk.


    The use of spoofing contributed to the crash on May 6th, 2010; HFT can exacerbate volatility, create ripple effects, and stoke investor uncertainty, which can affect consumer confidence. Spoofing is a type of market manipulation in which traders place orders that they plan on cancelling before they are executed. Their goal is to create the illusion of demand or supply for a particular security to influence the price in their favour. Using HFT, spoofing can be done by quickly placing and cancelling huge orders to create a false impression of market activity. The HFT algorithm can use this tactic to manipulate the market and generate profits by buying or selling securities at favourable prices. Spoofing has been made illegal and can result in fines or legal action by regulatory authorities, but it can be difficult to discern when it was done intentionally or if it’s simply a cancelled trade. (Corporate Finance Institute, 2023)


    The other flash crashes mentioned also show how High Frequency Trading can have negative effects on the market. On the August the 24th 2015, the Chinese stock market experienced a sharp sell-off. This triggered a global market sell-off, which was blamed on algorithmic trading systems that automatically sold off assets in response to a sudden drop in Chinese stocks. Another flash crash occurred on February 5th, 2018, when the Dow Jones Industrial Average (GANTI, 2019) experienced a drop of over 1,100 points - its biggest single-day point drop in history. One if the factors was the HFT algorithms which, in response to rising interest rates and inflation concerns, sold off assets rapidly. A third crash happened on September 3rd in 2020 when the S&P 500 futures market experienced a sudden and severe drop in prices (falling by nearly 4% in just minutes). This crash was a result of a combination of factors, including a large sell order from an algorithmic trading system and concerns about market volatility.

    Technical Glitches

    Technical glitches are errors or malfunctions that occur in algorithmic trading systems, which can cause unexpected or unintended consequences.


    One common technical glitch is incorrect data feeds, where an algorithmic trading system receives incorrect or incomplete data, which can lead to incorrect trading decisions. (Angel One, 2023) For example, if an algorithmic trading system receives incorrect pricing data, it may place trades at incorrect prices, resulting in significant losses. Another common technical glitch are software bugs, where algorithms have programming errors that cause them to work incorrectly. Software bugs could cause trading computer systems to place trades in the wrong direction or at the incorrect time, which could result in significant loss in capital. Algorithmic trading systems can also experience system overloads. This could happen if the system attempts to process too many trades or data entries at the same time. During times of high market volatility, computers may receive large volumes of trading data, which can overload the system and cause it to malfunction. Technical glitches can also occur due to connectivity issues (such as network or internet outages) which can disrupt the transmission of trading data. If a computer loses connection to the trading platform, it won't be able execute trades.


    Technical glitches can cause in significant losses through erroneous trades, delayed trades, loss of control, and amplification of market volatility. It is important for firms using algorithmic trading systems to have robust risk management and monitoring processes in place to minimize the impact of technical glitches.


    On August 1st of 2012, Knight Capital Group activated new software that contained a major flaw. The software was activated when the NYSE (New York Stock Exchange) opened, causing the firm to lose around $10 million a minute according to Popper (2012).

    What caused the knight flash crash?

    When Knight replaced its old software, the new RLP (Retail Liquidity Program) code repurposed one of the flags that initially activated a program called “Power Peg”, this was intended to change the code so that when the flag was activated the new component of the RLP would be activated (Dolfing (2019)). Many large software companies such as Knight often keep “dead code”, which can result in flaws like this when these systems are maintained for years by several people. Power Peg was an algorithm designed to be used to check the behaviour of proprietary trading algorithms in a controlled algorithm. (Google Docs, 2013)


    Proprietary trading is “trading in financial instruments or commodities as principal. It requires the use of a firm’s own capital, or liquidity, or both.” - (Prudential Regulation Authority, n.d.)


    The test program bought high and sold low to move stock prices higher or lower.


    Whilst algorithmic trading can execute trades at an increased speed and therefore quantity, this can be negative as shown by the Knight Capital software error on August 1, 2012. The algorithm’s speed of trading resulted in Knight buying 150 different stocks, at a price of around $7 billion, in just under the first hour of trading. While this was a huge issue, it highlights two disadvantages of using algorithmic trading – the speed and malfunctions in the code that can result in catastrophic decisions. Algorithms don’t have the capability to realize when there is something going wrong, they have set instructions that they follow until they are shut down. In the time it takes for a human to shut down the program, the algorithm will keep running, exacerbating the situation. (Google Docs, 2013)


    Algorithmic trading will require human monitoring in case of malfunction. Automated trading system cannot substitute humans, because they need constant monitoring to prevent mechanic failures and data of system.

    VII. Counter Arguments

    The benefits of algorithmic trading outweigh the disadvantages

    Algorithmic trading improves efficiency by reducing the time and resources required to execute trades. Algorithmic trading can also increase trading accuracy (by analysing large amounts of data), reducing the impact of emotion-based trading, and improving risk management (Senior Supervisors Group Algorithmic Trading Briefing Note, 2015). It can also contribute to increased market efficiency by facilitating faster and more accurate price discovery. The European Central Bank conducted a study that showed algorithmic trading can improve market liquidity and reduce bid-ask spreads. (Brogaard, Hendershott and Riordan, 2014)(HENDERSHOTT, JONES and MENKVELD, 2011)


    The risks associated with algorithmic trading - technical glitches and the potential for increased market volatility - can be reduced through careful risk management and monitoring; algorithmic trading and computational models can help improve risk management and reduce the likelihood of market disruptions. (Spotlight Review Emerging themes and challenges in algorithmic trading and machine learning, 2020)

    The disadvantages can be mitigated through better regulation and oversight

    Technical Glitches:

    Technical glitches are a major risk associated with algorithmic trading, but this risk can be reduced through better regulation and oversight. Regulators can enforce rules that require firms to implement adequate safeguards (such as circuit breakers and kill switches) to prevent technical glitches from causing serious market disruptions. Following the 2010 Flash Crash, regulators in the USA and Europe introduced circuit breakers and kill switches to prevent technical glitches from causing significant market disruptions. These have been effective in preventing similar incidents from occurring. (U.S. securities regulator says more to be done on ‘kill switches’, 2013)

    Managing Market Volatility:

    Algorithmic trading can contribute to market volatility, but better regulation and oversight can help manage this risk, for example by requiring firms to follow guidelines on market-making and limit order placement to prevent excessive volatility. Regulators have introduced rules to manage market volatility caused by algorithmic trading: the Securities and Exchange Commission in US  introduced a limit-up/limit-down mechanism to prevent excessive price movements, and the European Securities and Markets Authority introduced guidelines on market-making and limiting order placement to prevent excessive volatility.

    Improving Transparency:

    Algorithmic trading can be quite opaque, but better regulation and oversight can improve transparency by creating regulations that firms must disclose their algorithmic trading strategies and make trading data more accessible to the public. In 2018, the European Securities and Markets Authority introduced regulations for firms that use algorithmic trading. These rules forced them to disclose their trading strategies and algorithms to regulators; this has improved market transparency and regulators are aware of the risks associated with algorithmic trading. (MiFID II Review Report MiFID II/MiFIR review report on Algorithmic Trading, n.d.)

    Reducing Market Manipulation:

    Algorithmic trading can be used for market manipulation, but this can be prevented by better regulation and oversight. (Shearer, Rauterberg and Wellman, 2019) Regulators enforce rules against market manipulation and require firms to implement monitoring systems to detect and prevent manipulative trading practices. In USA and Europe, rules to prevent algorithmic traders using market manipulation have been created. The European Securities and Markets Authority has introduced guidelines for algorithmic trading that make manipulative trading strategies prohibited; the US Commodity Futures Trading Commission implemented rules requiring firms to have monitoring systems to detect and prevent manipulative trading practices.


    (ALESSIO AZZUTT, WOLF-GEORG RINGE, H. SIEGFRIED STIEHL, n.d.)

    Protecting Investors:

    Algorithmic trading can pose risks to investors, but better regulation and oversight can protect them. Regulators check that firms have adequate risk management systems in place and require firms to disclose risks associated with algorithmic trading to investors. Rules protecting investors from these risks have been introduced - the US Securities and Exchange Commission created rules requiring firms to provide risk disclosures and the European Securities and Markets Authority have rules requiring firms to implement risk management systems. (Algorithmic Trading Compliance in Wholesale Markets 2 Financial Conduct Authority Algorithmic Trading Compliance in Wholesale Markets, 2018)(Algorithmic trading, 2018)(www.finra.org, n.d.)(Bank, 2019)(Lee and Schu, 2021).


    Better regulations and oversight (like the ones mentioned above) will ensure that algorithmic trading will contribute to a fair and efficient market by reducing risks. They will also make sure computational models and algorithmic trading are used in a responsible and sustainable way; this is important because the benefits are significant enough to outweigh the potential disadvantages.

    VIII. Conclusion

    Key points discussed in the essay

    This EPQ explores various methods of trading, some using computational models and some excluding them, to evaluate to what extent traders use computational models when making decisions.


    Algorithmic trading has advantages, such as increased efficiency and improved risk management, but it also poses risks to market stability and so requires careful risk management. Although technical glitches and market manipulation are potential risks associated with algorithmic trading, the benefits of algorithmic trading outweigh the potential disadvantages. These risks can be mitigated through better regulation and oversight; Developments in machine learning will allow traders to develop even more complex automated trading systems. The various types of computational models in trading show how they play a large part in trading today, and help traders make decisions on a daily basis. The extent that which computational models are used has increased significantly since their introduction and will continue to increase.

    The continued growth and evolution of algorithmic trading in the trading world.

    Algorithmic trading has grown and evolved over the past decade. With increased usage, advancements in technology, expansion into new markets, and increased competition, regulatory scrutiny has also increased. Whilst these developments led to new opportunities for traders and investors, they have created new challenges for regulators and market participants. Computational models are predicted to continue playing a significant role in the trading world as technology continues to advance and the markets evolve.






    (4997 words excluding references, titles, subheadings and pictures)

    Miro-Leaders
    L O A D I N G
    . . . comments & more!

    About Author

    __nik HackerNoon profile picture
    __nik@everyone
    Inika Prakash [biography]
    Read my storiesAbout @everyone

    TOPICS

    purcat-imgfinance #trading #future-of-trading #ai #options-trading #market #traders #future #machine-learning

    THIS ARTICLE WAS FEATURED IN...

    Permanent on Arweave
    Read on Terminal Reader Terminal
    Read this story w/o Javascript Lite
    Coffee-web

    RELATED STORIES

    Article Thumbnail
    Artificial Neural Networks (ANNs) Have Transformed the Way We Interact With Computers
    by everyone
    Jan 17, 2023
    #artificial-intelligence
    Article Thumbnail
    The Noonification: How Often Do NFTs Pass The Howey Test? (1/13/2023)
    by noonification
    Jan 13, 2023
    #noonification
    Article Thumbnail
    Darwin's Hybrid Intelligence to Align AI & Human Goals for Startups & VCs
    by natasha
    Jun 25, 2019
    #hackernoon-shareholder-series
    Article Thumbnail
    The Noonification: White Man (11/26/2022)
    by noonification
    Nov 26, 2022
    #noonification
    Article Thumbnail
    The Noonification: The Metaverse is a Sh*tshow (11/2/2022)
    by noonification
    Nov 02, 2022
    #noonification
    Join HackerNoonloading
    Latest technology trends. Customized Experience. Curated Stories. Publish Your Ideas