AI can absorb and process more information than any investor possibly could.

An AI model can gather data faster and analyze patterns in that data with ruthless speed. As a result, “having the data” is no longer the edge it was once propped up to be.

According to famed investor Howard Marks, the true edge for any investor has now shifted from processing to judgment.

Investor Essentials Daily:
The Monday Macro Report
Powered by Valens Research

A little over a year ago, Howard Marks, the famed co-founder of Oaktree Capital Management published a letter saying he didn’t think we were in an AI bubble yet. And last September, he doubled down saying that, while prices for some of these stocks are high, they deserve those high valuations.

In the wake of AI disrupting the software industry, Marks is back, yet again. This time, though, he talks as much about how AI can be used in investing as he does about investing in the age of AI. That’s why his latest thoughts on artificial intelligence land a little differently.

He isn’t trying to guess which model wins, or whether today’s spending is rational. He’s looking at what happens to the craft of investing when the smartest tool in the room is a machine that can read faster than investors can think.

Marks starts with a blunt reality: AI can absorb more information than any investor.

A language model can gather data faster, remember it better, and analyze patterns in that data with ruthless speed.

It can also do all of those things without getting tired or emotional.

That matters because “having the data” has been a pretend edge for a long time.

Back before most information was available online as soon as it was announced, information was a competitive advantage.

In a digital world, everyone gets the same headlines and data at the same time. And AI won’t just give everyone the same information, it can process it better than most people can.

That’s great because it levels the playing field. On the flipside, it confirms that simply having information is not an advantage anymore.

Marks’ point isn’t that human investors become obsolete. It’s that a big chunk of what used to pass for “skill” has been commoditized.

When everyone can replicate the work of an investment research associate in seconds, investors need to find an edge elsewhere.

The advantage then shifts from processing to judgment.

Marks narrows the human edge to three jobs that still matter, because they live outside clean pattern matching.

First, judging the “import and implications” of new information. AI can tell you what happened, what usually happens next, and which past situations look similar.

The tricky part is deciding what matters when the situation is genuinely new, or when the market is reacting to the wrong detail.

There’s little edge in finding the numbers—the true edge lies in understanding what those numbers mean about the future.

AI models are trained on historical data and inherently, that makes them better at repeatable tasks like pattern recognition. But in investing, investors need to have an eye for the future which is (at least partly) uncharted.

Second, models still struggle assessing qualitative factors. For example, an AI model can give an opinion on whether or not a company’s management team is paid to do the right things and it can wax poetic about a company’s latest innovations.

AI is better than humans at gathering facts, not at forming opinions based on those facts.

Third and finally, great investors will still have an advantage at foreseeing a company’s future. Once again, this is non-quantitative. It requires a bit of “futurism” that a model trained on past data cannot replicate right now.

This is where Marks sneaks in a subtle warning. Investing isn’t built entirely on facts. It also runs on informed extrapolation and, at times, opinion. That last piece is where AI can become dangerous, because it can sound confident while inheriting the biases of its training data and the framing of a prompt.

Investors should treat AI like an investment turbocharger as it’s going to make it easier for investors to get up to speed. It should be used to gather context and let it surface patterns that might have been missed.

Once all that is done, investors should do the part Marks says is protected: Decide on what matters.

Decide which management teams to trust, map out what the future might possibly look like, and identify which risks the market is ignoring.

In a world where everyone has access to the same machine intelligence, the edge comes down to who knows when to outsource work to a model and when to keep the reins.

Best regards,

Joel Litman & Rob Spivey
Chief Investment Officer &
Director of Research
at Valens Research