Adam Smith Business School

      

AI for Financial Technology is a long-running collaboration led by Professor Bowei Chen at Adam Smith Business School. It brings together academic researchers and two fintech companies, ArrayStream Technologies and Inboc Technologies, to create artificial intelligence tools that support risk management, trading and investment decisions in financial markets.

The team works with large datasets from markets such as the S&P 500 index, using them to design models that can be used in practice by financial firms, while still behaving in ways that finance professionals recognise and can trust.

Making AI that “thinks” like finance experts

One strand of the work looks at what option prices can tell us about risk. Option markets contain an important signal called implied volatility, which reflects how likely the market thinks large price moves are for a particular asset.

Professor Chen and collaborators Professor Yu Zheng and Dr Yongxin Yang have developed an AI model that learns to read this risk signal from 20 years of option data on the S&P 500 index. The key idea is that the model is built so it automatically respects well-known financial “rules of the game”, such as the absence of obvious arbitrage opportunities and the typical “smile” shape that traders see when they plot implied volatility against option prices.

Because these rules are built into the design and training of the model, its behaviour can be checked against established financial theory, rather than treated as a mysterious “black box”. In tests, the model not only follows these rules in practice, but also predicts implied volatility more accurately than widely used finance models and other AI approaches.

Designing simple funds that follow the market

Another strand of the project focuses on index tracking, a common “passive” strategy in which an investment fund aims to match, as closely as possible, the performance of a market index such as the S&P 500, rather than trying to beat it.

In many cases, it is too costly and complex to hold every single share in a large index. Instead, funds look for a smaller set of shares that together behave much like the full index. This is known as partial replication.

Working with Professor Yu Zheng, Professor Timothy Hospedales, and Dr Yongxin Yang, Professor Chen has helped design an AI-based method that selects a limited number of shares while closely tracking the index. The model is tested on more than ten years of S&P 500 data and is compared with standard approaches that are widely used in the industry.

Across a range of portfolio sizes, the AI approach delivers the lowest tracking error, the gap between the index and the fund, while achieving similar overall risk and return measures, such as volatility, Sharpe ratio and maximum drawdown, even once transaction costs are taken into account.

Partnership, funding and recognition

This work has run from 2009 to 2023. It is supported by the ESRC Business Boost Fund and the Shenzhen University WeBank Institute of FinTech Grant, reflecting interest from both UK and international partners in the use of AI in finance.

The collaboration with ArrayStream Technologies, funded through the ESRC Business Boost Fund, was a finalist for Best Collaboration in Business at the University of Glasgow Knowledge and Public Engagement Awards in 2020.

The wider research team spans several institutions, including the Southwestern University of Finance and Economics, Queen Mary University of London and the University of Edinburgh, alongside the fintech partners ArrayStream Technologies and Inboc Technologies.

 

---

For further information, please contact business-school-research@glasgow.ac.uk 

First published: 18 November 2025