Bullish reported July 2026 metrics this week. Total volume: $30.7 billion. Down 42.9% year-over-year. Average spread: 2.62 basis points. Up 72.4% year-over-year.
One number is catastrophic. The other is framed as resilience. Both appear in the same disclosure, side by side, as if they carry equal analytical weight. They don't.
The "average trading spread" is not a bid-ask spread. It is not a commission schedule. It is a blended ratio — commissions relative to trading volume, adjusted for perpetual contract fair value changes and rebates. Three variables. One opaque figure. No external decomposition possible.
This is not my first encounter with a platform that defines its own profitability metrics in ways that resist verification. In 2020, I audited a DeFi lending protocol celebrating 500% APY while its oracle mechanism consumed stale data feeds. The community called it innovation. The code called it a re-entrancy waiting to happen. Hype is just noise in the signal. But when a company hand-builds the signal itself, noise becomes structural.
The July disclosure is the clearest recent example of a metric doing narrative work. Volume falls 43% year-over-year, and the accompanying spread metric suggests the exchange is monetizing more efficiently. Both facts are true within their own definitions. The question is what those definitions actually measure.
The Context: Institutional Polish, Non-Standard Plumbing
Bullish is a Gibraltar-regulated, institutionally focused central exchange backed by Block.one capital. Brendan Blumer's vehicle. Peter Thiel among the early backers. Its most distinctive asset is CoinDesk — the only major exchange in the world with a media outlet under the same corporate umbrella. Exchange traffic and editorial influence, vertically integrated into a single entity.
The company runs a disclosure cadence that mimics public-company reporting: monthly operational data, quarterly preliminary figures, annual reports. That cadence is the brand. "Institutional-grade" is the pitch. Data published on schedule, every schedule. The underlying definitions, however, are anything but standard.
This is what a governance review looks like when the subject is a private exchange: you read the numbers they publish, you map the definitions they choose, and you ask what those definitions are designed to reveal — and what they are designed to conceal.
The July disclosure reveals plenty. Spot trading volume: $29.1 billion, down from $45.5 billion in June and $48.8 billion in July 2025. Derivatives volume — inferred by subtracting spot from total — collapses to roughly $1.6 billion in July. June's inferred figure: approximately $5.3 billion. July 2025: approximately $5.1 billion. A 68.6% year-over-year decline in perpetual contract trading.
The broader market narrative in July 2026 is bullish. Prices elevated. Attention abundant. Yet this exchange's own data shows a transaction exodus. Industry-wide spot volumes declined between 20% and 35% over the same period — painful numbers, but nowhere close to the 42.9% contraction Bullish recorded. The excess decline points to something beyond market beta: market share loss, client migration, or a structural problem in the venue's product suite.
The comparison to the prior year sharpens the picture. Total volume fell $23.2 billion year-over-year. Spot fell $19.7 billion. Yet the derived revenue proxy barely moved — down roughly 1.6%. That combination, massive volume loss with flat revenue, is the single most important fact in this disclosure. It tells you the exchange captured more value from a dramatically smaller activity base. Whether that is strategy or survival is the open question.
That divergence between market narrative and exchange-level activity is the signal worth dissecting.
The Core: Building the Revenue Model the Company Won't Disclose
Since Bullish won't disclose clean revenue, we construct a proxy. Volume times spread.
July 2026: approximately $8.04 million. June 2026: approximately $13 million. July 2025: approximately $8.2 million.
Month-over-month: down 38.3%. Year-over-year: down 1.6%.
The year-over-year flatness is the framing the company prefers. It's also misleading. The proxy revenue fell by roughly $5 million in thirty days. That is not market noise. That is a structural event — a major client departure, a market dislocation, or an intentional reduction in market-making activity.
Now decompose the monthly move. Spread expanded 2.3% month-over-month. Volume fell 39.7%. The revenue collapse is almost entirely volume-driven. The spread improvement narrative cannot offset a transaction exodus. The resulting mix is "smaller venue, better unit economics" — if the unit economics are real.
Take the ETH data. Spot ETH volume dropped from $11.1 billion to $3 billion year-over-year. A 73% decline against an overall spot decline of 40.4%. ETH is deteriorating far faster than the rest of the book. This is not market beta. It is a product-level problem: deteriorating ETH liquidity, or institutional execution channels migrating to other venues where depth is deeper and custody is more established.
The derivatives picture is worse. Perpetuals make up roughly 5% of total volume. Industry benchmarks: Coinbase runs near 50%, Binance above 70%. A derivatives line at 5% isn't a business line — it's a checkbox. For an institutionally positioned exchange in a bull market, this is the clearest sign of competitive marginalization. Perpetual volume fell about 69% month-over-month and about 69% year-over-year. Both directions pointing down with similar magnitude suggests the June spike was itself an anomaly — a single event pulse that temporarily masked the underlying collapse.
That is a critical point. The June proxy revenue of approximately $13 million is the outlier, not July. June's average spread was 2.56 basis points — lower than July's 2.62. Lower spread, far higher revenue: the June revenue spike was purely volume-driven. One institutional client, one large event trade, or one algorithmic campaign. When that pulse ended, revenue reverted to a level nearly identical to a year earlier. The "stability" of the year-over-year proxy is not strength — it is a flatline with a temporary interruption.
The Hidden Variable: Rebates and the Definitional Trap
Now the forensic detail that deserves attention. Rebates are folded into the spread definition. If maker rebates increased — whether to attract liquidity or to compensate existing market makers — the spread mechanically expands even when the platform charges users nothing extra. The 72.4% year-over-year spread increase may reflect cost-structure deterioration, not pricing power.
The data cannot distinguish between "we charge more per unit" and "we pay more to keep the book alive." Those scenarios have opposite implications for profitability. The company's chosen metric conflates them. If the math doesn't separate, the conclusion is indeterminate.
Also worth flagging: the disclosed data is labeled "unaudited preliminary estimates." Adjustments are possible when the quarterly settlement closes. Anyone comparing this month's numbers to next quarter's audited figures should expect revisions — in either direction. Precise to two decimal places, yet explicitly subject to change.
Competitive Calibration
Calibrate against the industry. Bullish's July volume of $30.7 billion works out to roughly $1 billion per day. Coinbase has historically traded between $1 billion and $5 billion daily during normal conditions. Binance operates at multiples of that. Bullish is running at one-twentieth to one-fiftieth the scale of the top venues.
The product mix compounds the problem. Spot is 95% of volume; derivatives are 5%. The institutional clients Bullish courts typically demand sophisticated derivatives products for hedging. A venue without meaningful perp liquidity cannot serve that demand. The volume structure is essentially a warning label: this is a spot venue with a derivatives afterthought.
The Regulatory Overhang
The regulatory dimension deserves attention. Bullish operates from Gibraltar with ties to the Block.one ecosystem. If it serves US users — and the annual-report cadence suggests ambitions beyond the EU — its perpetual products fall under the Commodity Exchange Act's jurisdiction. Unregistered offshore derivatives services for US clients have become a serious enforcement priority. That exposure does not appear in monthly trading metrics, but it is a tail risk that institutional counterparties price into their venue selection.
Then there is the CoinDesk factor. A media outlet owned by an exchange creates a structural conflict of interest that no amount of editorial firewalls can fully resolve. Market participants will always discount the independence of coverage produced under the same corporate roof as a trading venue. And if editors face pressure — real or perceived — the reputational damage flows back to the exchange brand. The media asset cuts both ways.
The Contrarian Angle: What the Bulls Got Right
I have spent two decades in this industry, and I know a Ponzi when I see one. Bullish is not a Ponzi. There is no token, no emissions schedule, no incentive flywheel. It runs a traditional exchange revenue model, where fee income cycles with market activity. That is a real business. The absence of token-incentive distortions is a genuine positive.
The disclosure cadence deserves credit. Monthly numbers, quarterly preliminaries, annual reports. Most unlisted crypto companies publish a tweet. Bullish publishes data. That discipline is a governance signal, and it matters.
The CoinDesk asset is also undervalued in bearish analysis. Owning a media outlet creates a distribution moat no other exchange has. In an attention-driven market, that is a structural advantage. Exchange plus media plus compliant disclosure: that is a credible pre-IPO narrative, assuming editorial credibility survives the integration.
And the unit economics did improve. Proxy revenue is nearly flat year-over-year despite a 43% volume decline. The platform extracts more per unit of activity — if the spread expansion is genuine pricing power rather than rebate distortion. In that scenario, Bullish is becoming a higher-margin business at a smaller scale. A defensible thesis for a privately held entity preparing for a public listing. Shrink to profitability is a legitimate strategy when the owners control the timeline.
There is also a counterfactual worth considering: had Bullish not raised its effective take rate, the revenue proxy would have collapsed alongside volume. The team chose to defend revenue, not market share. In a bull market, that is an unusual choice. In a pre-IPO process, it is a rational one.
The problem is that "if." The definitional ambiguity prevents external verification. Bullish is asking the market to accept a profitability signal it cannot independently validate.
The Takeaway: The Audit Comes Before the IPO
This matters because Bullish is clearly positioning for a public listing. The annual report reference. The standardized disclosure rhythm. The careful metric framing. These are pre-IPO behaviors. When that prospectus lands, every definition will face scrutiny from analysts who reject blended metrics on principle.
The question is not whether July was a bad month. It was. The question is whether the metrics tell you why. A spread definition that blends commissions, fair value changes, and rebates is a narrative instrument, not an analytical one. "Fully audited" in crypto too often means someone signed off on the code, not someone verified the economics.
For analysts, the lesson extends beyond Bullish. When a platform controls the definitions of its own health metrics, period-over-period comparisons become an exercise in trust. The industry's stated commitment to transparency is sound — but only when the metrics being published are the ones that matter.
Check the source code, not the roadmap. For protocols, that means reading the contract. For a CEX, the equivalent is reading the footnotes. Bullish's footnotes are doing heavy lifting.
The market sits in a bull phase. Prices high. Attention abundant. Bullish's own data shows transaction volume contracting 43% year-over-year while its self-defined spread metric expands 72.4%. One of those signals is honest. The other is a construction.
I would ask the company one question: if rebates were stripped from the spread calculation, what would the 2.62 basis points become? The answer distinguishes pricing power from cost. Until that decomposition is public, the spread improvement is just another promise waiting to be audited.
And from where I sit, promises don't settle. Findings do.