Content area
Full Text
1. Introduction
Artificial intelligence (AI) has myriad applications in finance [1], [2] (Zheng et al., 2019; Milana and Ashta, 2021; Ashta and Herrmann, 2021; Cao, 2022). For instance, based on the big data accumulated over many years, financial institutions including banks and insurance companies can apply various machine learning approaches to determine whether a loan applicant will pay back the originally borrowed fund and how to customize insurance policies (Abdou and Pointon, 2011). More recently, AI-based software has significantly enhanced the efficiency of financial institutions to identify financial frauds and money laundering activities (Albashrawi, 2004; Couchoro et al., 2021). However, perhaps the most exciting application of AI in finance could be in the field of wealth management and investment (Cao and He, 2009; Ruf and Wang, 2020; Tadapaneni, 2020; Petrelli and Cesarini, 2021; Ferreira et al., 2021).
Academia and financial practitioners have mixed opinions about whether AI can beat the stock market. As AI is more powerful than human brains in both calculations and information processing, especially in dealing with a huge amount of data, many people with a fantasy about AI naturally believe that with significant advancements in IT technology, sooner or later AI will win the game if not now (Giudici, 2018; Chopra and Sharma, 2021). Meanwhile, the efficient market hypothesis (Fama, 1970) implies that there does not exist any chance for an AI-based investment strategy to beat the stock market if the strategy is only focused on technical analysis of historical data, as the stock prices have already fully reflected them [3], [4].
To explore the above question, we should establish a clear boundary of the role played by AI in the stock trading process [5]. In a typical AI-based investment scenario, asset managers may apply natural language processing software to mine tons of newspapers/social media posts to better predict investor sentiments around specific companies called “sentiment analysis” and, thus, may be more likely to create an informed investment strategy than a traditional investor to outperform the stock market (Kearney and Liu, 2013; Nassirtoussi et al., 2014; Day and Lee, 2016; Gurrib and Kamalov, 2022). Furthermore, AI-based investment strategies are not limited to passive sentiment analysis. As the recent presidential election has revealed, even intelligent people can...