Tech Explained: Here’s a simplified explanation of the latest technology update around Tech Explained: Traders Overwhelmed by Iran News Are Turning to AI for Help in Simple Termsand what it means for users..
Photographer: Majid Saeedi/Getty Image
(Bloomberg) — As Maxence Visseau spent the first few days of the Iran war trying to make sense of what the conflict would mean for markets, he put artificial intelligence at the heart of his investment process.
Most Read from Bloomberg
Large-language models enabled Visseau, the founder of investment firm Arkevium, to cut the time he spent on research by about 80%. He used Anthropic’s Claude to stress-test multiple scenarios in parallel, compare historical precedents and map out potential ripple effects across asset classes.
“I was up for almost 48 hours straight, monitoring the interceptions in the United Arab Emirates while simultaneously running scenarios and preparing for the market open,” said Visseau, who’s based in Dubai and specializes in macro trading strategies. “That’s precisely the kind of moment where AI becomes indispensable.”
While Visseau said the technology isn’t a reliable substitute for human judgment, he views the time-saving benefits of AI as increasingly essential for navigating markets whipsawed by a war that has upended energy supplies and left at least 4,000 dead. Interviews with investors and strategists across the globe suggest the conflict has led AI tools to become more entrenched in workflows, even as several flagged pitfalls including sloppy prompts and inaccurate results.
“We are witnessing history — the first major conflict where AI is being used to fight and where traders rely on AI to map out the war in ways that have never been done before,” said Nick Twidale, chief market analyst at AT Global Markets in Sydney and who’s a 25-year veteran of trading markets.
One of the advantages of using AI tools such as OpenAI’s ChatGPT, Google’s Gemini and China’s DeepSeek is a dramatic improvement in time management.
Jian Shi Cortesi says where she previously may have spent half an hour reading different sources to catch up on the news, now the Zurich-based fund manager at GAM Investment Management can get a summary of the latest developments in the war in seconds. Gathering information about a particular company takes a day or less, down from multiple days previously.
“In the past, it’s like digging a hole with a shovel. Now you’re digging dirt with these massive excavators,” Cortesi said. “The speed has probably increased by five times.”
Another is being able to mine history near instantaneously for insights and context on what may happen next, especially given the volatility in markets. Brent crude surged as much as 11% to top $119 a barrel on Thursday on concern over the risk of escalating tit-for-tat attacks on key energy facilities in the Middle East, before settling at about $109.
Deep Search
Anna Wu, a cross asset strategist at Van Eck Associates Corp. in Sydney, used ChatGPT and Claude to go back 100 years to track every war-driven oil breakout, and find out what asset classes outperformed in each occurrence. To improve the utility of the answers, she asked AI to cross reference with other data points such as the median inflation and global economic growth.
“It definitely has brought in more efficiency,” Wu said. “A lot of the historical analysis right now has become a lot less time-consuming because before it would be me searching to the end of Google.”
For Gustavo Pessoa, artificial intelligence tools provide instant access to information that previously might have been hard to get, at a time when calculating the impact of the virtual closure of the Strait of Hormuz is becoming more essential to investing.
“We use it for everything — from understanding the types of ships to analyzing the elasticity of oil demand to prices and even estimating how many barrels will be needed to stabilize flows,” said Pessoa, who’s a founding partner at Sao Paulo–based hedge fund Legacy Capital Gestora de Recursos Ltda.
AI is not perfect, nor is it a replacement for human experience and decision-making. The technology has made mistakes in everything from gaming development to news content representation. A Bank of England policymaker warned that AI adoption in trading may amplify market shocks and herd-like behavior.
Arkevium’s Visseau said the answers needs to be constantly tested for accuracy.
“It’s an iterative process — I’ll challenge the output, refine assumptions, introduce new data points,” he said.
Junior Analysts
Knowledge is vital to being able to make the most out of AI, said Michael Brown, senior research strategist at Pepperstone Group in London, adding that the technology isn’t a ‘silver bullet.’
“Participants still need a deep understanding of the situation themselves in order to make that final trading call and being able to gauge if the AI models are throwing out spurious information, which we all know can sometimes be the case,” Brown said.
Yet the growing effectiveness of the tools means positions like entry-level research analysts could be at risk.
Cortesi said she could do without junior analysts, describing AI as her “best” research assistant.
“I can ask AI: use a Warren Buffett approach to give me the key points of this company, and the AI will do it right away,” she said. “But if you ask a junior, the junior may not know what Warren Buffett approach is. So these types of more complex requests, definitely AI does much better. Also AI does it so much faster.”
How long AI remains in this ancillary role is uncertain. Before the Iran war started, investor paranoia over which industries would be ravaged by the displacing efficiency of artificial intelligence spurred selloffs in the shares of software companies to food delivery services.
John Foo, founder of Valverde Investment Partners in Singapore, said the technology is far from replicating the human thought process — for now.
“AI is an additional resource that complements the human research effort,” he said. “I say complement and not replace as there is still an element of good judgment and experience that goes into decision making that AI at this stage — maybe the next 2-3 years — still cannot fathom. After that it will be anyone’s guess.”
