There’s a lot of buzz around emerging artificial intelligence tools such as ChatGPT and their ability to absorb vast banks of data and game investment markets. While the use of this technology will become eventually be ubiquitous, it still can’t predict winners and losers in markets or provide specific financial advice.
When ChatGPT is asked about its potential for retail investors, it tells you it’s useful for summarising company financial and identifying industry trends. It can also generate reports on portfolio holdings and provide rebalancing recommendations.
Robots can, and already do, provide simple investment advice. At the other end of the scale, high frequency traders already use sophisticated algorithms to buy and sell big share parcels based on short-term market movements, to get the best price for their trades. This is where the real magic happens; it’s not with ChatGPT. Yet.
Jack Magann, emerging companies portfolio manager with wealth advice firm Oracle Investment Management, thinks over time AI will have a big effect on investing, financial analysis and portfolio management.
He says AI will enhance the information investors have to manage risk in their portfolios. “AI can evaluate an enormous amount of data related to the economy, social trends and market movements, helping investors to make informed decisions about how to position their portfolio.”
Magann also believes AI will be able to predict future outcomes more accurately. At the moment fund managers rely on data tools that have limitations on the number of variables they can analyse. AI has the potential to increase the number of variable inputs to uncover data relationships that were previously unavailable.
He says AI will become an important way of removing cognitive bias. This is a way of thinking that’s essentially incorrect, the most common example is confirmation bias, when we think when something goes our way, it’s attributable to ourselves, but if something bad happens, it’s someone else’s fault.
In investing, we might think if we pick a stock and its value goes up, it’s because we’re an excellent stock picker. But if the share price goes down, it’s the market’s fault, rather than our own fault for not doing enough research about the factors that could cause a decline in the stock’s value.
Magann believes machines can help us remove this bias and make us more successful investors. “In the future we may be able to run an investment idea or decision through a virtual AI assistant to scan for cognitive bias that may be influencing decisions. This is exciting because investors may not be aware of the biases they have developed from past experience.”
But AI can’t remove bias altogether. As Sarah King, head of client care and advice with robo financial advice specialist Stockspot notes, the biggest issue with ChatGPT and AI in general is it’s tainted by the biases of its authors or creators.
“It reflects their views and biases and it can’t take into account complex personal circumstances. If you asked for tax planning, estate planning and ETF investment advice, the responses it generates are likely to be shallow and lacking in depth and insight.”
Another limitation is that ChatGPT’s knowledge is not up-to-date. For example, when you ask ChatGPT about the outlook for bond ETFs in Australia, its knowledge is limited to 2021. It says: “my knowledge cut off in 2021, so I cannot provide up-to-date information on the current market.”
Says King: “much to investors’ annoyance, ChatGPT doesn’t have predictive ability. So, it won’t be able to tell you which investments will outperform.”
The dataset ChatGPT was trained on only goes up to 2021. So anything that happened after that is missing. Another limitation is it isn’t connected to the internet. The software also doesn’t know what a share is, what value is or what risk is.
“If you ask, it will define risk but it isn’t aware of what risk is. This is the biggest problem with it. The question is whether it matters if an algorithm has no concept of money, but can predict stocks that are most likely to appreciate in the next 12 months. This will come with better training and development of the algorithms,” says James Eling, an IT expert with Extreme Networks.
Nevertheless, this technology is likely to shake up the way investment banks do research and make stock recommendations. “We will see a next iteration of the ‘battle of the algorithms’ as each bank looks for an edge in managing their portfolios,” Eling says.
He believes, however, AI has the potential to increases volatility, especially if models are trained on similar datasets with similar parameters and too many algorithms take an analogous approach. “But I think there will be quite a bit of diversity among the algorithms.”
Plus, it may threaten jobs such as quantitative analysts. “There are a lot of roles we thought were safe, that are looking like they do not have strong futures.”
Longer term, there will be finance-specific AI applications, trained on historic finance data and connected to the internet with real time feeds. Eling says this will enable very specific advice to be given to retail investors.
“This will be when AI really comes into its own and I don’t think that time is far off; maybe one to two years. This is because the datasets are readily available and there is a lot of historic data already available. Machine learning will be able to see the patterns between many disparate factors, such as the economy, commodity prices, demographics and the political environment. The exciting thing about ChatGPT is that it gives retail investors the power to ask simple questions and get easily-understood answers.”
This means retail investors will be able to get answers to their investment questions, such as: which commodity will perform the best over the next five years?, followed by, which stock will benefit from a 30 per cent rise in the price of that commodity? Or, which stocks will fall most if interest rates increase by two per cent?
Eventually, this kind of computing power may be an opportunity for disintermediation, to retail investors’ benefit.