John Maynard Keynes once compared the competition for capital to a beauty contest, declaring that ‘successful investing is anticipating the anticipations of others’. In the capital markets, judging corporate performance isn’t absolute; it’s about how that performance is perceived relative to others. Trying to figure out what to invest in and what to avoid is what makes a market work.
No wonder, then, that investors are taking to using AI models to help them do this faster and more thoroughly than was ever possible before. They have access to more data, more computing power and more storage. Using AI doesn’t change investors’ views about what drives a company’s share price performance or its ability to repay debt, but it is starting to fundamentally change how they evaluate those drivers and evaluate risk.
‘What used to take an analyst three weeks now takes a couple of days’
Quality information is the lifeblood of capital markets, to paraphrase Arthur Levitt, former chair of the US Securities and Exchange Commission. And investors’ quest for alpha (excess risk-adjusted returns) has never been more informed. Large volumes of data – much of it unstructured – used to be a problem, creating noise when investors are looking for a signal. But the newest AI models are changing that – extremely quickly. Concerns about the length of annual reports and the volume of other data will soon be a thing of the past.
An investor told me: ‘There used to be a cap on the number of data points we could practically use. Now complexity is a benefit, not a challenge, and the thirst for anything a company puts out is high.’ Investors’ requests for ‘more’ are only set to increase. And they have new tools to help them sift through it all – and to do so in minutes or hours, not weeks.
Game changer
Of course, AI is neither new nor a single technology. Investors have been using machine learning and natural language processing for years to uncover trends, flag risks and compare companies. The game changer is generative AI. It saves investors time, finds things they might have otherwise overlooked and highlights where they should focus their analysis. Although adoption is still uneven, its capabilities are improving as datasets grow and models get better trained.
‘AI digests decades of annual reports in less than two hours,’ one fund manager told me. ‘What used to take an analyst three weeks of research to develop an investment thesis, using far less information, now takes a couple of days, with far more reports and data points.’
The machine’s assessment will filter what gets to a human decision-maker
This means investors can consider more stocks, including those from companies in foreign markets or with limited analyst coverage, and in more detail. But interpretability and explainability are still critical when it comes to making an investment recommendation. Critical thinking skills, alongside investment judgment, remain essential when using generative AI. New skills such as prompt engineering and AI supervision can help ensure a model’s output is defensible, understandable and appropriate.
This all points in one direction: a machine will read corporate reports before they ever get to a person. The machine’s assessment will be the filter that determines what gets through to a human decision-maker.
Reporting refocus
What should companies know about today’s competition for capital when the contest is being judged in new ways using vast amounts of information that the company may not even be aware of and cannot control?
First of all, accept that a machine will scrutinise your annual report, AGM materials and other communications before a person will. Many businesses will find this unpalatable, but as entrepreneur Peter Diamandis puts it: ‘The future doesn’t care if you believe in it. It’s coming anyway.’
What investors are looking for – how you’ve performed, where you’re going and what the risks are – is the same as it always was. AI models are being trained to pick these up, so presenting these aspects clearly matters more than ever.
AI sees patterns people can overlook, and can unearth what’s missing
Second, know that AI will corroborate what you say, in detail, against a lot of other information. AI sees patterns that people can overlook, and can unearth what’s missing. It will check what you’ve said before and assess how well you’ve honoured your commitments, progress made against targets and whether those targets have moved. It will check the plausibility of your claims. And it will compare all this to your peers’ performance.
It all helps investors assess the risk that a company is, at best, hiding something and, at worst, lying about it. More positively, it also helps them see whether the market is missing something that they want to get in on, now.
Third, expect investor meetings with the executive team and the board to be more informed. With AI doing much of the data gathering and analysis, investors may arrive with complex questions drawn from a detailed review of your reports and public data. This can help identify the issues they want to engage on and guide which way to vote. It can lead to better investor engagement, but it also means you may have to clarify or correct the conclusions that have been reached.
Clear and truthful
Investor scrutiny, armed with AI, is increasing, and companies need to focus on what they can control: the clarity and accuracy of their reports. To give the machine the best chance of understanding what you intend, it is more important than ever for your reports to be clearly structured and well articulated.
One investor I spoke with said: ‘The quality of reporting will still be indicative of the quality of a company. If anything, AI raises the bar for reporting quality even higher.’ Doing it well can keep you in the contest.
