Coinbase published its third-quarter results on October 29, and the figure the company wants the market to remember is simple: prediction markets are now running at more than $200 million in annualized revenue. Buried deeper in the same release is a number that matters far more to anyone trying to value the Nasdaq-listed exchange — a $641 million gap between GAAP net income and adjusted net income, almost entirely the product of unrealized gains on crypto assets held on the company's own balance sheet.
That gap is the real story of the quarter. Coinbase reported GAAP net income of $719 million, a swing of more than $1 billion from the $359.5 million net loss it posted in the prior quarter. Adjusted net income, which strips out the fair-value movement on those assets, was $78 million. The company's chief financial officer was unusually direct about the source, describing the swing as unrealized gains on crypto held for investment purposes. In plain terms, the two quarters are close to mirror images of each other, and the line separating them is the price of bitcoin, not the performance of the business.
None of that diminishes what is genuinely happening in the prediction market unit. Coinbase now says the business is generating more than $200 million annualized, growing at an 82 percent compound quarterly rate, with daily traders up 60 percent quarter over quarter. The company has also secured CFTC approval to list its own prediction market contracts, a change that moves it from distributing third-party products to issuing and market-making its own. That is a meaningful structural shift, and management's claim that it will "improve our economic model" is almost certainly an understatement: self-issued contracts let Coinbase capture the full spread and fee rather than a distribution cut, which could lift its take rate well above the level it earns as a pure intermediary.
The more interesting disclosure, and the one that received the least attention, is that more than half of Q3 contracts traded were non-sports. That single data point matters more than the 82 percent growth rate. Sports contracts are short-lived, highly seasonal, and sit in direct competition with DraftKings and FanDuel. Political, economic and crypto-price contracts are where prediction markets earn the right to be described as information-pricing tools rather than gambling substitutes. A business mix that is majority non-sports is a business that can plausibly claim a longer lifecycle.
But the scale needs to be kept in perspective. A $200 million annualized figure is a run-rate extrapolation, not realized revenue. On a quarterly basis the unit is producing something in the region of $50 million against roughly $1.2 billion in net revenue, or about four percent of the total. The 82 percent compound quarterly growth rate is mathematically unsustainable — sustained for four quarters it would imply an eleven-fold increase. Prediction markets are a narrative option on Coinbase's future, not a financial pillar of its present.
The pillar, or at least the most underappreciated part of the business, is subscriptions and services. That segment brought in $570 million, or 49 percent of net revenue, and it is recurring rather than transactional. Coinbase has spent years arguing that it is becoming a platform rather than a leveraged bet on trading volume, and this quarter offered the first real evidence. Total market trading volume fell 10 percent across the industry, yet Coinbase's revenue was essentially flat. Its market share rose to 10.7 percent, the fourth consecutive quarterly increase. That combination — shrinking market, stable revenue, rising share — describes consolidation, not growth, and it means smaller exchanges are absorbing a double hit.
The subscription base deserves its own scrutiny. Coinbase One membership grew 24 percent year over year to a record high, the fifth consecutive quarter of expansion, including through what management itself called a challenging market environment. Members generate 3.5 times the average revenue per user of non-members and hold 4.2 times the account balances. Those are striking numbers, and they are also a textbook case of selection bias: wealthy, high-volume users are more likely to buy a membership, which is not the same as a membership making users wealthy. The company still does not disclose the absolute number of members, only the growth rate, which usually signals a base that remains small relative to the whole.
Stablecoin economics sit underneath the subscription line, and they are the quietest risk in the entire report. Average USDC balances on the platform hit a record $22 billion. Interest earned on USDC reserves, split with Circle under a revenue-sharing agreement, is a core component of subscription revenue. That income is currently classified as recurring and stable, but it is a floating-rate exposure dressed as an annuity. A sustained rate-cutting cycle compresses reserve yields directly, and no hedging framework has been disclosed. The same $22 billion figure is a positive for Circle on scale and for Base on settlement volume, which makes falling rates a shared risk across all three.
Regulation cuts both ways here, and the reporting has been incomplete. The CFTC approval is a genuine milestone, placing Coinbase inside the designated contract market framework with its attendant obligations around market surveillance, position limits and large-trader reporting. But federal approval does not confer state-level immunity. The central legal question for event contracts in the United States is whether they constitute wagering under federal and state gambling statutes, and Kalshi is already fighting that battle in multiple states. No state-level litigation or injunction risk appears anywhere in the quarter's disclosure. That is a significant omission, because a single adverse state ruling could undermine the entire expansion thesis for the product line.
The non-sports mix may itself be a piece of regulatory positioning. Sports contracts are the easiest category to characterize as sports betting. Political and economic contracts sit closer to derivatives and information markets, where the legal footing is more favorable. Emphasizing that more than half of volume is non-sports is good business disclosure and good regulatory narrative management at the same time.
Competition will compress the advantage regardless. Kalshi holds a CFTC designated contract market license and has deeper category expertise. Polymarket has global reach and on-chain settlement through UMA's optimistic oracle, though it operates under US access restrictions and carries dispute-resolution risk. Robinhood is chasing the same retail wallet through licensed partners. Coinbase's moat is not technology — contract design, settlement rules and oracle mechanics are all replicable. Its moat is licensing, retail account infrastructure and fiat rails. The strategic purpose of prediction markets may not be to win a standalone category at all, but to raise wallet penetration among existing users.
Management's cost guidance suggests they see the trading environment clearly. The 2026 expense outlook was trimmed to $4.15–4.3 billion, a defensive posture that implies preparation for a prolonged period of subdued volume rather than a bet on near-term recovery. Combined with 15 consecutive quarters of positive adjusted EBITDA — $333 million this quarter, up 60 percent quarter over quarter — it paints a picture of a company managing for durability.
Taken together, the three exposures form a business that no longer fits the label of exchange. Coinbase now collects transaction fees, earns a share of stablecoin reserve interest, and carries a sizable crypto balance sheet whose marks flow straight through earnings. Valuing it as a pure volume-driven trading venue understates the recurring revenue but overstates the predictability of profit. Valuing it as a platform understates how much of its earnings power still depends on asset prices. There is also a subtler mechanical feedback loop: mark-to-market gains flatter reported profit, which supports the equity, which supports the capacity to hold and raise capital, which supports more holdings. That loop runs in reverse just as efficiently, and the prior quarter demonstrated it.
The narrative risk is that the market prices the wrong set of numbers. Prediction markets contribute roughly four percent of revenue while occupying the headline and the chief executive's characterization of the unit as one of the fastest-growing products by revenue. Note the qualifier: "one of." The gap between narrative weight and financial weight here is roughly ten to one. Meanwhile the GAAP profit that will anchor most screens is more than eight times the adjusted figure, and it will reverse the moment crypto prices turn.
What to watch is specific. Next quarter's absolute prediction market revenue, disclosed without the annualized wrapper. Whether the non-sports share holds above 50 percent. The size and composition of the self-held crypto position in the 10-Q notes, which remains the largest undisclosed variable in the earnings picture. Any state attorney general action or federal court injunction against event contracts. The timeline for the company's own CFTC-approved contracts going live. And USDC balances against the path of short-term rates.
Coinbase is not a prediction market company that happens to run an exchange. It is an exchange whose income statement has become a leveraged long position in crypto, wrapped around a genuinely improving recurring revenue base and a small, fast-growing, legally contested new product. Investors who price only the first of those three exposures are mispricing the other two.
Word count: 1523.
