Coinbase CEO's Earnings Call Remarks Stir Criticism Over Prediction Market Bets

On the evening of October 29, 2025, Coinbase CEO Brian Armstrong was wrapping up his company’s third-quarter earnings call when he went off script. Not in the way CEOs usually go off script — no slip about a pending deal, no accidental disclosure of nonpublic information. Instead, Armstrong pulled up a prediction market on his screen and started reading the words that people had bet money he would say.

“I was a little distracted because I was tracking the prediction market about what Coinbase will say on their next earnings call,” Armstrong told analysts and investors, “and I just want to add here the words Bitcoin, Ethereum, blockchain, staking, and Web3, to make sure we get those in before the end of the call.”

With that, approximately $84,000 in bets on Kalshi and Polymarket swung to settlement. Armstrong had just demonstrated, live on an earnings call for a publicly traded company, exactly how easily prediction markets can be manipulated by anyone with a microphone and an audience.

His response afterward, posted to X: “Lol, this was fun — happened spontaneously when someone on our team dropped a link in the chat.”

The crypto industry’s reaction was considerably less amused.

The mechanics of the stunt

Prediction markets like Kalshi and Polymarket allow users to wager on the outcomes of future events. In recent years, both platforms have expanded beyond traditional political and economic bets into increasingly granular territory: Will a specific company’s CEO say a particular word during an earnings call? Will a press conference mention a specific country? Will a particular phrase appear in a government report?

These so-called “mention markets” had gained traction in crypto circles as a form of entertainment trading. For the Coinbase Q3 call, users on Kalshi had placed $80,242 in total bets, while Polymarket users had wagered $3,912. The markets offered contracts on whether Armstrong would use specific terms — Bitcoin, Ethereum, blockchain, staking, Web3 — during the earnings presentation.

Around 8 p.m. Eastern on October 29, as Armstrong delivered his closing remarks, the odds on Kalshi shot to just under 100 percent for bets on “Bitcoin,” “Prediction Market,” and “Web3.” The timing aligned precisely with Armstrong’s deliberate recitation of the trigger words.

For anyone who had bet “yes” on those terms, the moment was profitable. For anyone concerned about the integrity of prediction markets as a financial instrument, it was something else entirely.

The criticism

Jeff Dorman, Chief Investment Officer at Arca, a digital assets investment firm based in Los Angeles, didn’t mince words.

“You need your head examined if you think it’s cute or clever or savvy that the CEO of the biggest company in this industry openly manipulated a market,” Dorman wrote in a public statement. He pointed to eight years of work by institutional crypto firms to get digital assets taken seriously by traditional finance, by regulators, and by the kind of pension funds and endowments whose capital the industry needs to grow.

Armstrong’s stunt, Dorman argued, undermined all of it.

The criticism from Dorman was notable because it came from inside the industry. This wasn’t an SEC commissioner or a CFTC enforcement attorney raising red flags. This was a crypto-native investment manager, someone who had spent nearly a decade building legitimacy for the asset class, saying that the CEO of its largest company had just handed regulators a reason to crack down.

Other voices echoed the concern. On social media, the reaction split along predictable lines. Crypto maximalists found the moment entertaining — “Keep breaking the fourth wall, Brian Armstrong. This is as good as it gets,” one user posted. But risk-averse participants in the ecosystem, the ones trying to build compliant infrastructure, saw it differently: “This prediction market stuff is going to end badly. Way too easy to manipulate.”

The regulatory question

Under the Commodity Futures Trading Commission’s (CFTC) rules, regulated exchanges are prohibited from listing contracts that are “readily susceptible to manipulation.” The phrase is deliberately broad, and it has been the basis for the CFTC’s scrutiny of prediction market contracts for years.

Armstrong’s stunt raised a straightforward question: if the CEO of a publicly traded company can move a prediction market by simply choosing to say a word, is that market “readily susceptible to manipulation”?

The answer, for anyone who watched the earnings call, was obviously yes. The harder question is what regulators do about it.

Kalshi, which operates under a CFTC license, has been in an ongoing legal and regulatory battle to expand the types of contracts it can offer. The company has positioned itself as a legitimate, regulated alternative to offshore prediction markets. Armstrong’s demonstration that a single individual could trivially manipulate outcome contracts on the platform was not helpful to that argument.

Polymarket, which operates as a decentralized prediction market and is not registered with the CFTC, faces a different set of issues. The platform has been the subject of a Department of Justice investigation and operates in a regulatory gray area. Armstrong’s stunt added another data point to the case that crypto-native prediction markets need clearer oversight.

Coinbase’s response

A Coinbase spokesperson told reporters that Armstrong’s comments were made “in a lighthearted, offhand way referencing online discussion around the earnings call.” The company emphasized that it has internal controls prohibiting employee participation in company-related prediction markets.

The statement was carefully worded. Coinbase didn’t claim the stunt was appropriate or inappropriate — it framed it as spontaneous and informal, something that happened in the spirit of the moment rather than as a premeditated market-moving action.

Whether regulators accept that framing remains an open question. At minimum, Armstrong demonstrated that the CEO of the world’s largest publicly traded cryptocurrency exchange is aware of, tracks, and is willing to interact with prediction markets that reference his company’s public communications. For a company that spent years navigating SEC enforcement actions and that recently won a legal victory establishing that not all crypto tokens are securities, the voluntary creation of new regulatory exposure was a choice.

The financial context

The irony of the prediction market stunt is that Coinbase’s actual Q3 2025 numbers were strong enough to speak for themselves. The company reported $1.9 billion in net revenue, a 55 percent year-over-year increase. Trading volumes were up. Subscription and services revenue — the recurring, less volatile side of Coinbase’s business — continued to grow.

In a normal earnings cycle, those numbers would have been the story. Instead, every major financial and technology outlet led with the prediction market theatrics. Bloomberg ran a piece headlined “Coinbase CEO Stunt Exposes Prediction Market Vulnerability.” TechCrunch went with “Coinbase CEO Brian Armstrong trolls the prediction markets.” Fast Company called it “word salad.”

Armstrong, whose net worth Forbes estimates at over $10 billion, moved $84,000 in markets by saying five words. The amount is trivial relative to his wealth, to Coinbase’s market cap, and to the broader crypto economy. But the principle it demonstrated — that prediction markets on subjective, controllable outcomes are inherently gameable — is not trivial at all.

What it means for the Bay Area

Coinbase is headquartered in San Francisco — or, more precisely, it operates as a “remote-first” company that maintains its legal headquarters in the city and its CEO in the Bay Area. Armstrong’s public profile is intertwined with the region’s tech identity: a Y Combinator alum who built Coinbase from a San Francisco apartment into the first major crypto company to list on the Nasdaq.

The prediction market stunt sits at the intersection of two Bay Area obsessions: cryptocurrency and the emerging prediction market ecosystem. Kalshi is backed by Sequoia Capital and other top-tier Silicon Valley venture firms. Polymarket has drawn investment from Peter Thiel’s Founders Fund. The platforms represent a bet by some of the Bay Area’s most prominent investors that prediction markets will become a mainstream financial instrument.

Armstrong’s demonstration that those instruments can be trivially manipulated by a single person with influence over the outcome is the kind of stress test that Silicon Valley’s “move fast and break things” ethos doesn’t always account for. The move was technically legal — Armstrong didn’t trade on prediction markets himself, and Coinbase’s internal policies prohibit employee participation. But it exposed a structural vulnerability that neither Kalshi’s regulatory framework nor Polymarket’s decentralized architecture was designed to handle.

For the Bay Area’s fintech ecosystem, which has invested billions in the premise that prediction markets represent a more efficient form of information discovery, the question Armstrong raised — albeit accidentally, and with a grin — is the one that matters most: what happens when the person who controls the outcome is also watching the market?

The bigger picture

Prediction markets have been heralded as one of the most promising applications of blockchain technology. The premise is appealing: by allowing people to bet on outcomes with real money, these platforms aggregate information more efficiently than polls, pundits, or surveys. The success of Polymarket’s political betting during the 2024 presidential election — where the platform’s odds proved more accurate than most traditional polling — gave the concept enormous credibility.

But the Armstrong incident highlighted a distinction that prediction market enthusiasts have sometimes been reluctant to confront. Markets that bet on verifiable, external events — who wins an election, what GDP growth will be, whether it will rain — are fundamentally different from markets that bet on the behavior of specific individuals who know they’re being watched.

Word-mention markets, in particular, are a category where the “outcome” is entirely under the control of the person being bet on. An earnings call is not a coin flip. It’s a scripted presentation by a CEO who can choose to say whatever they want. The moment Armstrong pulled up the prediction market and started reading from it, the market ceased to function as an information aggregation tool and became something closer to a game show.

Whether that’s a fatal flaw in the prediction market model or simply a feature that needs better contract design is a question the industry will have to answer. Armstrong, for his part, seems unbothered. “This was fun,” he wrote.

The $84,000 in bettors who got paid probably agree. The regulators watching from Washington may not.