By Paavo DCastro and Owen Jackson
Introduction
The use of AI is becoming increasingly more popular, particularly in the finance industry. The nirvana for would-be traders is the ability to write software that could automatically predict and trade on the financial markets, essentially making money on its own. But if there’s one lesson to learn from the world of finance, it’s that there’s no such thing as a free lunch. In this article, we examine exactly what the claims are about the ability of AI to predict the stock market, and whether they really stand up in practice.
What is Technical Analysis?
The best way to describe technical analysis is using past data to predict future performance of a particular stock, or a stock market more broadly. For example, one might use the term structure on bond yields to predict when there will be a market correction (a fall in the market of more than 20%). There are many self-proclaimed technical analysis experts and magicians who claim to have found the secret formula to predicting the stock market, and that they have made millions in doing so. However, given the sharpest minds are working on these problems in Wall Street and elsewhere, it seems unlikely that any of these claims would actually be true. Moreover, if technical analysis were successful, it would violate a fundamental assumption of financial markets - that of weak form efficiency. Essentially, weak form efficiency requires that the past performance of the market should not influence the future performance of the market. This makes sense, because all traders (or enough traders at least) will have already taken this information into account.
Technical Analysis Using AI
This brings us to the focus of today, the introduction of AI into the field of technical analysis. For the proponents of the field, they claim this has been a great development, because now they won’t have to stick to rudimentary models, instead they will be able to develop systems capable of processing a large amount of data and detecting complex patterns. However, the situation does not change with AI; the above problems we have mentioned still hold in this world as well, but some people seem to conveniently forget that. Furthermore, those who claim that AI and algorithmic trading facilitates unbiased decisions are being misleading. After all, there has to be someone who creates those algorithms, who decides when to crystallize profits and when to cut their losses. In other words, trading is inevitably a human activity.
Empirical Evidence
To be fair to both sides, we should look at the empirical evidence of AIs application in trading. There are many avenues of empirical research into AI and machine learning applications in the financial markets. Some have experimented with various machine learning algorithms and have successfully predicted market sentiment 76% of the time. This success rate has been pushed to 80% over 30 seconds of trading by the Oxford-man Institute. While these success rates are high, researchers find that this is not a high enough level of accuracy given the stock market risks.
A literature review which will be linked below, collated a list of papers that created exceptional returns and prediction accuracies, with some making returns as high as 100%. Highlighting that AI can be successful in predicting the market and generating returns. There have also been several criticisms of empirical research in that some studies don't consider transaction costs, use short back testing periods and have small data sets, which effect the overall effectiveness and viability in predicting the market.
Conclusion
There is much discussion around this topic from both sides and we can’t do it full justice in this short article. Ultimately, what we want to highlight is that nothing in life comes free and people should be wary when others make claims that seem too good to be true, whether they are using AI or not.
If you enjoyed this article, register now for a free membership using this link to receive regular updates about our publications and events.
To sign up: 1. Log into UMSU using the link above. 2. If you are a current University of Melbourne student, add the ‘Finance Students’ Association Student Member’ option to your basket; otherwise, select the ‘Associate Member’ option. 3. Congratulations! You are now a member of FSA!
The views expressed within this article are those of the authors and do not represent the views of the Finance Student's Association. All images and references in this article are for fair and educational purposes only. The content in this article is not intended as legal, financial or investment advice and should not be construed or relied on as such.
References and Further Reading