By Nargiz Akhmetova

Since 6000 BC, barter has been the oldest method of a market functioning. Back then, the exchange system was absolutely plain. As time progressed, many negative and positive factors changed the market drastically. Today, foreign exchange (forex), a global marketplace that determines the exchange rate for currencies around the world, made a breakthrough in the 21st century as the most ever-growing barter market, with $5 trillion in virtual trades every day (1). 

        Everything in global business is made simultaneously with technological and scientific advancements to provide the most convenient trading method. “Artificial intelligence is a simulation of human intelligence processes by machines, especially computer systems” (2). Foreign exchange markets are one of the many spheres where AI has penetrated, causing the debate over its benefits versus the advantages of long-established systems. Artificial intelligence proved itself as an instrument to reduce human error in predicting various conditions. From a technological perspective, implementing AI in the forex market is especially beneficial as it provides an opportunity to scrutinize vast amounts of data needed to make a high accuracy prediction of the market changes (3).

        The continuation of artificial technology advancements led to an increased number of techniques applied in automated forex trading. Artificial intelligence algorithms, however, can lead to some pitfalls due to the resources needed to run a modern system. The implementation of advanced trading systems still carries risks that were not considered during technological optimization. One of the drawbacks is the difficulty of finding a high-quality dataset, which is vital for evaluating the predicting model. Another con of using machine learning-driven algorithms is the increased risk of volatilization of the market because of the increased speed with which trades are made(4).

 

Benefits of AI-trading

        Jean Folger, financial writer and co-founder of a trading strategy development company, wrote in her article “Automated Trading Systems: The Pros and Cons” that it is officially acknowledged that more than 80 percent of stocks on the US market are traded using automated trading systems (5). In the same paper, she mentioned the opportunity to remove the human emotion factor from the trading process as the most prominent attractive force that makes contemporary investors and traders implement those AI trading algorithms. 

        Folger continues developing her position by introducing the concept (the “law”) by which the trading algorithm is prospected to function. This set of rules is a selection from commonly available technical indicators. Self-created algorithms built on custom indicators are flexible according to market changes, and therefore more accurate than automated algorithms when it comes to prediction output, which proves that AI algorithms pay off in trading. However, this fact is not the only advantage given to readers in Folger’s article. The given list of benefits of automated trading includes applicability of backtesting (rules for determining the viability of the algorithmic idea), diversification of trading (opportunity of using multiple strategies for trading efficiency improvements simultaneously), and uplifting of order entry speed (immediate response on the market condition’s change).

        The provided benefits enhance the productivity of trading many times, which can be seen from the statistical data reported by Analyzing Alpha article (6). The fact that “around 92% of trading in the Forex market was performed by trading algorithms instead of humans” is a crucial factor that proves the workability of algo-trading in the forex market.” Another fact taken from the same article, which demonstrates the efficiency of AI usage in the barter market is that “72% of institutional investors think that AI and Machine Learning provides deep data analytics” (6). This information gives a comprehensive understanding that many knowledgeable institutional investors approve of the modern way of forex trade. Uplisted advantages are mostly about high efficiency and available opportunities provided by market artificial intelligence algorithms. However, algo-trading also has side benefits, which are no less important in the trading process.

        With algorithmic trading, there is less risk of losses and more opportunity to buy and sell because of the possibility of monitoring real-time prices, uplifting the level of transparency. It is essential as it not only indicates a substantial degree of market security but also maintains the high speed of decision-making (7). In continuation of the discussion about artificial intelligence approval in trading, the key notices are the scientific research conducted by ThuongMai University and Pereira Technological Institute (PTI). Thi Thu and Xuan (8), experts in data mining, in their study “Forex Trading Using Supervised Machine Learning,” presented the empirical qualitative and quantitative data of the successful application of the machine learning neural networks in maintaining the prediction of the fluctuation in the barter market. The researchers classified binary value trends changing up and down (true or false); the support vector machine (SVMs model) will be applied to determine the output.

        Another research paper focused on analyzing the performance of artificial intelligence in trading was made by scientists from Pereira Technological Institute. D F Devia Narvaez et al. (9)  used intelligent algorithms to calculate the exchange rates in the forex market. PTI’s scientists found the behavior of the supporting vector as nonlinear and chaotic, similar to randomization in algorithmic programming. Sequential and substantial analysis of those characteristics would give a chance to apply the predictive algorithm with high efficiency. The combination of recognition neural network alongside prediction technique allows a solid strategy for the forex trader. The study of PTI scientists has another focus in comparison with the scientists at ThuongMai University as they implemented a recurrent neural network for determining the fluctuation product of the market. In contrast, ThuongMai University researchers focused on classifying the result by two direct vectors. The two proposed studies are distinct in approach but similar in a conclusive statement that artificial intelligence has an adept expertise in predicting the direction of the market’s change.

 

Pitfalls of AI-trading

Despite the numerous positive sides stated earlier, there are some pitfalls of machine learning usage in the trading forex market. Folger (5) provided the list of the algo-trading limitations: 

  • Mechanical failures (simple technology malfunctioning: power loss, connection problems, or computer crush). 
  • Monitoring issues (many traders implementing AI-based algorithms do not have enough expertise in the way of algo-trading functionality as it requires knowledge of deep learning systems, this fact makes monitoring the efficiency of trading difficult). 
  • Over-optimization (excessive curve-fitting that makes the algorithm inapplicable in real market’s flow).
  • Scam risks (as the novice technological systems put the ground onto the new ways of fraud)

The given information is reinforced by the fact that “only 37% of the buy-side forex traders In the US and Europe use algorithmic trading.” (6). By-side forex traders are financial institutes operating with investment security (10). If less than 40% of the buy securities rely on algo-trading, some positive inferences on the reliability of AI trading should be made.

 

Conclusion

        Based on the articles and studies, it is reasonable to assume that artificial intelligence has proven to be a powerful way to trade in the forex market. In the research, algo-trading has limitations and weaknesses, so there can be no uniformly correct algorithm. However, it is a collection of directions that might be efficiently implemented. It is important to note that those algorithms still required human management and launch, so “full automation” of the trading process cannot be claimed. 

 

Sources:

[1] Bradfield D. The history of forex [Internet]. DailyFX. 2018 [accessed 2022Aug 2]. Available from: https://www.dailyfx.com/education/beginner/history-of-forex.html

[2] Burns E, Laskowski N, Tucci L. What is Artificial Intelligence (AI)? Definition, benefits and use cases. [Internet]. SearchEnterpriseAI. TechTarget; 2022 [cited 2022Aug 2]. Available from: https://www.techtarget.com/searchenterpriseai/definition/AI-Artificial-Intelligence

[3] Wilson J. How much artificial intelligence has changed the Forex Trade [Internet]. AiThority. 2021 [cited 2022 Aug 2]. Available from: https://aithority.com/guest-authors/how-much-artificial-intelligence-has-changed-the-forex-trade/

[4] A-Team. Potential and pitfalls of Artificial Intelligence in the trading environment [Internet]. TradingTech Insight; 2022 [cited 2022 Aug 2]. Available from: https://a-teaminsight.com/potential-and-pitfalls-of-artificial-intelligence-in-the-trading-environment/?brand=tti

[5] Folger J. Pros and cons of Automated Trading Systems [Internet]. Investopedia. 2022 [cited 2022 Aug 2]. Available from: https://www.investopedia.com/articles/trading/11/automated-trading-systems.asp

[6] Smigel L. 79+ amazing algorithmic trading statistics (2022) [Internet]. Analyzing Alpha. 2022 [cited 2022 Aug 2]. Available from: https://analyzingalpha.com/algorithmic-trading-statistics

[7] Trends M. Machine Learning in forex trading [Internet]. Analytics Insight. 2021 [cited 2022Aug 2]. Available from: https://www.analyticsinsight.net/machine-learning-in-forex-trading/

[8] Nguyen Thi Thu T, Dang Xuan V. Forex trading using supervised machine learning. International Journal of Engineering & Technology. 2018;7(4.15):400. DOI: https://doi.org/10.14419/ijet.v7i4.15.23024

[9] Devia Narvaez DF, Ospina R, Mesa F. Using implementation of artificial intelligence in estimating the exchange rate in the foreign exchange market. Journal of Physics: Conference Series. 2021;2118(1):012017. DOI: https://doi.org/10.1088/1742-6596/2118/1/012017

[10] Young J. Buy-side [Internet]. Investopedia. 2022 [cited 2022Aug 2]. Available from: https://www.investopedia.com/terms/b/buyside.asp

 

About The Author:

 

 

 

 

 

 

 

 

 

 

 

 

 

Nargiz is a grade 12 student at Nazarbayev Intellectual School in Nur-Sultan, Kazakhstan. Aspiring Data Scientist and AI researcher, Nargiz is exceptionally interested in implementing artificial intelligence in different life science fields. Amidst AI, she is also interested in the questions of human rights and politics.

Contact Nargiz:

Linkedin: Nargiz Akhmetova

Published On: 23 October 2022 / Categories: STEM Fellowship Journal / Tags: , , , , , , , /