While the advent of artificial intelligence (AI) has begun to have a significant impact on some industries, it’s yet to have a big impact on the world of investing although it is starting to creep into asset managers’ investment processes, according to Orbis Investments’ Clair Gallagher (pictured). 
Gallagher, a quantitative analyst, said AI is helping firms to identify patterns, sift through data and help to make processes more efficient.
“We know that machines have the ability to process in large amounts of data and they do it in a very disciplined way,” she explained.
“They don’t need coffee to function, they’re pretty consistently most of the time.”
But this system only works based on the quality of the data fed into the machine.
“They’re only as good as the data you put into them,” said Gallagher. “So, if you put rubbish in you get rubbish back.”
However, while AI is much faster at reading and interpreting market data and coming up with stock picks, stock selection is best left to humans, said the analyst.
“On average per screening they’re going to throw away 90- 95 per cent of the names and say okay that’s not for today,” she explained. “And we believe that the best controlled investment decisions are made by individuals.
“But we also recognise that we need to use technology to execute on our investment philosophy.”
She added: “Computers are good at solving some problems - something like chess, where the rules are constant and there’s a fixed number of outcomes.
“[But] in financial markets, the rules can change through time, so you need to be more flexible. And here, this is where computers are still struggling.
“This is what humans are good at – being flexible – because they know when there [are] exceptions, they know when things have changed.
“People are creative. They’ve got good general knowledge and abstract thinking, so they’re better at broader problems and that for us – as long-term investors – is very relevant because when you have a long investment horizon, the number of outcomes is infinite.”
Being flexible and putting things into context are important traits that AI cannot yet replicate.
Indeed, Gallagher explained that, as the market is currently in the late-cycle stage, traditionally investors would be “piling into your McDonald’s and your Coca-Cola stock, because in the past value and high-beta stocks would have led the market”. But, in fact, the opposite has been true, she said, a distinction that AI might not have made.
While AI processes are not yet perfect, Gallagher said, they can help act as a ‘bedrock’ for active management as they are less vulnerable to human biases.
“Humans are emotional and that can be good because you have your gut feeling,” the Orbis analyst said. “But you can also introduce biases which hamper your decision making.
“I don’t say machines are not biased because at the end of the day it’s human programming. So, there will be some biases in it, but they’ll be consistent biases.”
Whilst the technology is used to provide idea generation for potential holdings, it can also be used to check a fund’s investment thesis, said Gallagher.
“One of the ways you can really do damage long term for clients is by clinging on to losing positions and suffering from sort of confirmation bias with thinking that your thesis is just about to be proven right,” she explained.
“This is a bit of transparency into the underlying reality to force you to try and spot the stocks that aren’t working out as quickly as possible.”
An example of this can be seen between two very different stock options – global sports fashion retailer Nike and technology giant Intel. Both companies were highlighted as opportunities for the £74.5m Orbis Global Equity fund, but only the former was added to the portfolio.
Nike stock performance over the past 5yrs
Source: Google Finance
Looking at Nike, Gallagher said, “It had underperformed because there was a lot of headwinds in the physical retail industry in the US and Adidas were coming up with a more fashionable clothing and it was taking some of their market share. The strong dollar was a headwind too.”
But through pooling this analytical data on Nike following Twitter trends during the Colin Kaepernick campaign and customer visit to the website and app they still saw the quality of the business, according to Gallagher, an opinion informed by AI.
Intel stock performance over the past 5yrs
Source: Google Finance
With Intel – a potential beneficiary of the strong appetite for US tech stocks – the company was hit by a drop-off in demand from China for its microchips as it aims to become self-sufficient.
“What we found with Intel was that it’s a high-quality business that generates long profits, but the future prospects are quite low,” she said, noting that it was not added to the portfolio despite its own AI system flagging it up as a good potential holding.
“We are using the machines,” she concluded. “It’s the human telling the machine what’s important, it’s not the machine telling the human.”