Brian Ingram’s startup, Lodebar, feeds high-frequency market data into a machine-learning algorithm that turns it into jazz.Photo Illustration by The Globe and Mail. Sources: Getty Images
Where most people see market noise, Brian Ingram hears jazz.
When the New York-based hedge-fund-manager-turned-entrepreneur was trying to find new ways to unlock alpha – trading jargon for investment returns above the market itself – he decided there was a better way than watching traders. He wanted to capture the intuitions of people who didn’t even realize they were expressing a view on the markets: jazz musicians.
Mr. Ingram’s startup, Lodebar, feeds high-frequency market data into a machine-learning algorithm that turns it into jazz. The company hires accomplished musicians to play repeated solos over generated backing tracks, feeds the solos back into its algorithms and deconstructs them into trading strategies.
In one case, granular data on foreign-exchange trades between the U.S. dollar and British pounds became a keyboard and bass track over a drum loop, and a saxophonist’s solo was used to forecast volatility patterns in the currency pair.
The musicians “believe that they’re doing something completely unrelated,” he said. ”When you do that, you’re able to capture ... a core, visceral signal that still is processing the same high amount of information that a trader at a screen might, but able to do it faster and in more replete fashion.”
Mr. Ingram’s approach may be unorthodox, but it illustrates the lengths to which investors are going to find investment opportunities with the help of machine learning and artificial intelligence. Those technologies have made it possible for people to sift through vast quantities of unstructured information, such as market data, company releases and even jazz solos, for trading insights. Planners and even some data providers say retail investors should still approach AI analysis with caution.
“AI can generate a lot of feasible outcomes,” said Sandi Martin, an advice-only financial planner in Gravenhurst, Ont. “But can it pick the one that’s actually going to happen in a timeline that makes sense for you to be able to enter and potentially exit to profit consistently?”
Using AI to analyze companies isn’t just for bleeding-edge startups. In a presentation for the CFA Institute late last month, Malcolm White, a director and portfolio manager at BMO Global Asset Management, said that the extensive use of AI tools had been “truly inspirational” and helped to make successful investments that went against the market consensus.
In one case, by running different scenarios through AI models, BMO Global Asset Management came to a conclusion in August, 2025, that tariffs imposed by President Donald Trump under the International Emergency Economic Powers Act would be thrown out by the U.S. Supreme Court, Mr. White said. At the time, six months before the tariffs were ruled illegal, markets had priced in only a 40-per-cent chance that they would be scrapped, he said.
Today, Mr. White and his team use AI extensively to research companies, including in the AI market itself. He said they use it to find satellite images, track data centre construction projects, create charts to help visualize project timelines and generate interactive maps. AI coding tools help them to analyze correlations between companies.
Some firms are making AI analysis tools available to regular investors. Miramichi, N.B.-based fintech company Stockcalc is building a natural language processing model to extract financial data and sentiment indicators from media releases and corporate disclosures. Brian Donovan, Stockcalc’s president, said that by feeding that data into its valuation models, the company is able to provide “almost real-time” results for retail investors.
“If Shopify put out a news release this morning and we grab that news release this morning, we can parse that through, extract the data that we want, put it into our APIs and come up with an updated valuation almost immediately. A retail investor cannot do that,” he said. (APIs, or application programming interfaces, allow different software applications to communicate.)
A separate machine-learning project aims to use historical data and the company’s valuation models to construct portfolios, which Mr. Donovan said had consistently outperformed the S&P/TSX Composite Index when back-tested against it.
For many Canadian investors, the possibilities posed by new methods of market analysis may nevertheless be a double-edged sword. Mr. Donovan said that while new tools can provide valuable analysis, they can also make investing more difficult by forcing investors to consider a larger set of data inputs.
“It becomes almost chaos ... how do I get my arms around all of this information?” he said, adding that AI tools could help to create screens and filters.
Ms. Martin said that despite access to new tools to understand a greater volume of information, people are “still prone to making poorly informed, context-less decisions about their investments.”
The gap between retail investors and professionals with the ability to access or create better models also leads to a “tremendous information imbalance,” she said.
“Those are the people you’re trading against. You’re not trading against somebody else who’s also using a paid version of Claude.”
And sometimes the problem is not a glut of information, but the lack of it.
One client who had used AI tools to create a plan for restructuring their portfolio had been given incorrect guidance because of the model’s unfamiliarity with the internal workings of banks and brokerage firms, Ms. Martin said.
“It’s not out there, so [AI] doesn’t have access to it and it hasn’t been programmed with it.”