Marc-André Lewis sees information value in prediction markets but says structural barriers keep institutions on the sidelines
While prediction markets - online platforms where participants trade contracts on the outcome of real-world events - have captured retail attention and media headlines, institutional investors in Canada just aren't buying, in any sense of the word, according to one chief investment officer.
Marc-André Lewis, president and CIO at Toronto-based CI Global Asset Management believes these platforms remain too small, too ambiguous, and too close to gambling to warrant a serious allocation among the country's largest plan sponsors and asset managers.
Where prediction markets fall short
"Investing is not the same as trading and it's not the same as gambling. Prediction markets are pretty still close to gambling. My business is mostly about investing. Investing is medium to long-term, and therefore predictions on short-term events are not necessarily appealing," said Lewis.
He draws a firm distinction between investing and prediction markets’ lookalikes. Traditional asset classes, like equities and bonds, reward holders with a structural risk premium over time, he noted. Meanwhile, stock indices slope upward across decades as a government bond held to maturity delivers a positive return. That baseline compensation for bearing risk is, in his view, the foundational test any asset class must pass.
That’s why he suggests prediction markets fail it. Without a systematic forecasting edge, one that consistently beats the oddsmakers, there’s no premium to collect. A participant could sell insurance on long shots, taking the other side of low-probability bets, but that strategy carries blow-up risk when the unlikely event hits, said Lewis.
"It doesn't have the characteristics of an asset class but it doesn't make them uninteresting, though. There's information to be extracted from, and this information can be useful for investors," he said.
One aspect of such information Lewis highlighted are the bid-ask spreads in prediction markets that function as a built-in cost - a tax on participation that eats into returns. He believes that reinforces the line between these platforms and genuine investing.
When market outcomes can be gamed
Still, he underscored how dismissing them entirely would be a mistake, particularly as institutional investors need to build baseline expectations before positioning portfolios, and prediction markets aggregate public sentiment in a way that could sharpen that process.
The open question is whether the signal is reliable, he noted, comparing it to political polling, where documented biases create gaps between survey results and actual outcomes. If prediction markets prove to be strong forecasters, that points to a wisdom-of-crowds effect worth paying attention to.
"As investors, we'd be stupid actually not to kind of use that to extract information," he said. "The odds are higher for me to want us to be an observer of that market to extract information from it than a participant," he said.
Data published earlier this year by the Crisil Coalition Greenwich also echoed Lewis’ view. According to the flash study, conducted among 53 US-based market structure specialists, 73 per cent think institutional investors will eventually find tangible value in data generated by prediction markets.
Additionally, 43 per cent expressed a positive view, citing the innovative nature of these platforms and the potential value they could add to the marketplace.
Still, Lewis sees several structural problems that stand between prediction markets and institutional adoption. Outcome ambiguity is near the top of the list. While some events resolve fair and square, like a baseball score that’s final, others resist easy settlement, particularly in geopolitics, where the definition of an event's conclusion can be contested and litigated, he noted.
Then, there’s also the problem of manipulation. Lewis points to a case in April 2026 where someone used a hair dryer to trigger a temperature-based payout in Paris, a small but telling example of how participants can interfere with outcomes, particularly in niche contracts. That vulnerability extends to information asymmetry, he said.
"While market regulations and law enforcement prevent individuals holding material non-public information to profit from it, prediction markets lack any mechanism to prevent someone in the know from placing bets," said Lewis.
Compliance infrastructure poses another barrier as financial professionals are required to disclose transactions and, in many cases, obtain pre-trade clearance. The fragmentation and breadth of prediction market platforms, Lewis suggests, make it difficult to apply equivalent oversight. According to Lewis, restricting participation based on insider knowledge is straightforward in narrow cases, for instance, barring hockey players from hockey-related predictions but anything else can become unworkable at scale without sophisticated monitoring tools.
“While it’s not impossible that a portion of the prediction markets could evolve into some form of listed assets and get the regulatory oversight and compliance required to evolve into an asset class, the question of how this would actually fit in a portfolio remains,” said Lewis.
Hedge, maybe? Asset class, never
“While one could see prediction markets as a potential tool to hedge or protect portfolios, we need to keep in mind that market participants are often quite bad at, for example, linking election results and stock market reaction.”
For prediction markets to find any institutional foothold, Lewis outlines a narrow set of conditions: hosting on a securities-grade exchange with full regulatory oversight, sufficient liquidity for large-scale participation, and a restricted perimeter of non-ambiguous contracts.
"I think we’re far from that but if you get all of that then it's not inconceivable that people would use that. But they would use it more as a hedge than as an asset class. I don't think these become an asset class ever," he said. “Can these become hedging instruments? Maybe.”
