Over the past few years, the finance industry has witnessed a quiet revolution. Artificial intelligence (AI) and machine learning (ML) are no longer buzzwords; they’re tools used to forecast trends, reveal information, and even interpret human emotion. As an academic researching AI in financial markets, I’m often asked: Can a machine really beat the market? The answer isn’t simple - but the early evidence is promising.
At RMIT, my team and I have applied machine learning models to various financial markets with striking results. In China’s corporate bond market – one of the world’s most opaque and fragmented - ML models have outperformed traditional benchmarks in forecasting bond returns. These models capture subtle, nonlinear patterns in issuer and bond characteristics that human analysts often miss.
We’ve also applied similar techniques to the Australian stock market, exploring how ML can uncover predictive signals in company fundamentals, momentum, and risk factors. Here, too, the models proved effective, identifying drivers of return that conventional approaches either overlooked or oversimplified. This reinforces our view that AI tools are not just experimental - they're practical, adaptable, and increasingly essential across diverse market environments.
But machine learning doesn’t stop at crunching numbers. In collaboration with researchers abroad, we’ve also explored how public sentiment - extracted from images rather than words - affects global stock prices. By analysing thousands of news photos using AI vision models, we found that the emotional tone of images has a measurable impact on investor behaviour and, ultimately, stock performance. This taps into a deeper truth: markets are not just about fundamentals - they’re about perception.
Our research also delves into the use of ML in mutual fund management. By analysing fund manager behaviour, past performance, and market responsiveness, we use AI to identify patterns that may signal persistence - or deterioration- in performance. This has important implications for investors seeking better tools for fund selection in an overcrowded marketplace.
Beyond prediction, we’re also investigating how machine learning and natural language processing can promote transparency. One of our current projects uses textual analysis to assess corporate social responsibility (CSR) disclosures and flag potential greenwashing. By training models to detect inconsistencies between companies’ sustainability claims and actual performance indicators, we hope to provide stakeholders - investors, regulators, and the public - with sharper tools to hold firms accountable.
This new frontier in AI-led investing offers opportunities, but also raises concerns. What happens when only large institutions can afford these technologies? Will markets become more efficient - or just more unequal? And can we trust black-box models to guide investment decisions, especially in volatile times?
As educators and researchers, our role is not just to innovate, but to also ask these tough questions. At the College of Business and Law, we’re not only developing next-gen AI tools - we’re also teaching the next generation of finance professionals how to use them ethically, critically, and effectively.
Machine learning won't replace human judgment, but it can certainly augment it. The challenge now is to ensure that the future of finance - powered by AI - is transparent, inclusive, and driven by insight rather than hype.
Author: Xiaolu Hu
