The field known as behavioural finance is not new – insights have been discussed for years. The question, however, has been about how to act on these ideas and use them in portfolios. That requires an ability to crunch vast hauls of data. Well, that's where AI comes in.

When wealth management industry figures ask what use cases exist for AI, the ability to crunch vast amounts of data and avoid human errors in investing seems to rank high on the list.

The chances of human emotions holding sway increase when markets are turbulent and when political dramas intrude. Judgement can be clouded.

There’s plenty of drama right now with the changes to tariffs, geopolitics and interest rates, to name just three.

Understanding biases and emotions affecting investment decision-making is what behavioural finance is about. The term applies to understanding, for example, how people think their cleverness shapes their success, rather than accept that a certain level of luck is also involved. Another error is that people can treat losses more emotionally than they do with gains, leading to excess caution and fear of failure. There is also the danger of crowd behaviour – following the herd. These insights play to the idea that human traits go back to prehistoric times when such qualities ensured that people survived. Today, in an open market economy, such traits can backfire, however.

That’s the theory. A problem that has dogged the field is being able to capture vast amounts of market data and process it quickly enough to prove that controlling biases makes a positive difference. There was until recently a “so-what?” question that was asked – how can the insights of behavioural finance be used in portfolios? Back in 2022 – which is a long time ago in AI terms – Sonia Schulenburg of Level E Research told this publication that without the necessary tech tools, it is hard for investors to use the area’s insights wisely to achieve their goals.

AI is starting to make the field actionable, Greg Davies (pictured below), head of Behavioural Finance at Oxford Risk, told WealthBriefing in a recent call. 
 

Greg Davies

“In the last three to five years, a lot of that 'so what?’ has gone away….it is because we have got to gear our technology to allow us to personalise and communicate ideas and do all that at scale,” Davies said. 

If AI-driven data analysis can demonstrate how certain actions – such as avoiding biases – can produce superior risk-adjusted returns, that is a case of the numbers doing the talking. It takes the investment world closer to a managed experiment, Davies said. 

“When you use a technique to improve the world, they might be able to prove it in a lab but that cannot necessarily be shown in the real world,” he continued. With AI tools, the benefits of behavioural finance can be shown to have real-world effects.

“AI and digital delivery allow controlled experimentation in live environments such as A/B testing of communications, intervention timing, and framing,” Davies said. (A/B testing is employed to compare multiple versions of a single variable.)

Davies added: “This enables firms to observe behavioural changes in engagement, investment follow-through, and persistence over time, rather than relying solely on laboratory results.”

As ever, the cautionary point about AI applies, industry figures say, as summed up by the term "garbage in, garbage out."

“If people were to rely more on AI tools, potentially AI could remove some of these biases. However, as we saw with algorithmic trading platforms, if you just let machines make a decision, it is just based on a load of parameters, with inbuilt biases and assumptions that have been put in by people,” Chris Robinson (pictured below), group technology officer, IQ-EQ, told WealthBriefing. (Robinson is based in Singapore.) Another point to consider is that behavioural finance/economics may not cover all that is known about forces shaping markets and emergent phenomena such as chaos theory, Robinson said.
 

Chris Robinson

“AI is only as good as the data it feeds on. You need to have your data in a very secure, well governed and well organised fashion," Robinson said.

It makes a difference

A 2023 paper by Arthur D Little, a management consultancy, said that people who adopt behavioural finance achieve better financial outcomes. “By reinforcing customers’ commitment to an investment strategy, behavioural finance helps prevent early exits and deviation from an agreed investment plan,” the paper said. “This is the so-called behaviour gap that increases costs and reduces asset performance, which in turn can decrease investor returns by between 3 per cent and 6 per cent.” 

Compounded over time, such a gap adds up.  

Understanding biases come at a price. A report by Capgemini in 2024 said that 65 per cent of high net worth individuals admitted that their biases affected how they invested and 79 per cent said they looked to their relationship managers to mitigate this bias. Some 65 per cent of these wealthy individuals worried about the lack of personalised advice to align with their individual circumstances.

Discussions about AI’s use in the investment field continue. For example, in a report from Amundi’s research unit last year, entitled Artificial Intelligence for Behavioral Finance, the authors, Marie Briere and Karen Huynh struck a cautious note: “AI has the potential to deliver substantial value to both financial institutions and their clients by deepening the understanding of clients’ needs and using these insights to improve products and services. In particular, AI can provide customers with fast, personalised information and tailored advice. However, leveraging AI to accurately capture client preferences and develop behaviour prediction models or recommender systems is a complex and challenging endeavour, accompanied by inherent risks and significant costs.”

IQ-EQ's Robinson said it is important for those understanding the value of AI in its connection to behavioural finance, or indeed anything else, to be clear about definitions. The acronym “AI” is a catch-all term covering a variety of different type of technology from simple ML to advanced Agentic AI.

For example, there are AI extraction tools, which obtain data from emails, pdfs and other sources and render them coherent and usable, and save people from having to do this. “That is the biggest win in terms of efficiency, and has been so for the past two or three years," Robinson said. 

With agentic AI, “operationally, this is super-useful; these end-to-end process tools are a big focus. With family offices and smaller wealth firms, they are generally quite small in budget terms so they have an issue in developing scalable AI solutions, so it makes more sense for them to explore `out-of-the-box' solutions," he continued. Generative AI is not, yet, as interesting for wealth managers. On the client-facing side, IQ-EQ is not recommending GenAI, Robinson said. (GenAI creates new, original content – including text, images, code, audio, and video – by learning patterns from existing, large-scale data.)

Fear factor

Oxford Risk's Davies said that one insight from behavioural finance is the “fear of getting it [investments] wrong." “This leads to people doing nothing and leaving money in a bank account year after year," he said.

If people leave money in cash – and that is eroded by inflation – this also has macroeconomic implications. Regulators such as the Financial Conduct Authority in the UK and policymakers in the European Union want to encourage citizens to hold more risk assets (within certain boundaries of suitability). So behavioural biases have a public dimension for those trying to reduce pressure on tax-funded pay-as-you-go pensions, for example. 

Nudging towards good habits        

The idea of cultivating certain habits and learning techniques to save and invest seems to be gaining ground. PIMCO, the US fixed income investment firm, issued a paper about six years ago entitled Nudging yourself to better investment decisions. Taking its cue from the ideas expressed in the book Nudge, by Richard Thaler and Cass Sunstein, PIMCO said investors can adopt a framework that involves asking questions. “For example, inserting a step each quarter that asks investors to reiterate their longer-term investment goals and objectives, as well as their short-term needs, can help keep them on track,” it said. (PIMCO, incidentally, has partnered with the Center for Decision Research at The University of Chicago Booth School of Business as part of a drive to improve understanding how people make decisions on topics such as investment.)

Davies said anxieties about the state of the world – evidenced by the gyrations in the gold price for example – certainly make behavioural finance ideas an easy sell. 

“It’s never been easier than now to get people talking about the solutions that it provides,” he said. “The legitimacy of the field is no longer in question.”

One challenge, Davies said, is that investors can try to find a perfect opportunity to invest rather than get on with putting money to work. Perfectionism is a problem: “Don’t let the perfect be the enemy of the good.”

Allied to this is an industry that devotes resources to new products and perhaps not enough on getting people used to the routines of saving and building up investment. 

“Good investing outcomes depend less on having the optimal product and more on having an investment approach that people can emotionally stick with through uncertainty,” Davies added. 

Ironically, while AI and behavioural finance insights might help counteract human emotions, it appears that AI can trigger emotions – or hard rationality – as well. As reported a few days ago, what media reports called the artificial intelligence “scare trade” hit markets earlier this week amid concerns about the disruptive power of AI on the impact on delivery, payments and software companies. For example, shares in IBM fell. A bearish report was issued last weekend by a firm called Citrini Research (Bloomberg, 24 February.)

AI has roiled stock markets because people are concerned that this technology could put certain business models out of business, such as brokerages and online retail. That said, the jury is out on how profitable AI will be, or at least how quickly. Last August, a study from the Massachusetts Institute of Technology said most AI organisations don't make a profit. A report from MIT’s NANDA (Networked Agents and Decentralised AI) initiative found that 95 per cent of enterprise organisations have seen no return at all from their AI efforts.