B3's ETF growth report combines a six-month custody change with an annual percentage

Brazil's B3 exchange has published an impressive account of the growth of exchange-traded funds, but two of its own releases describe different populations and a percentage comparison that demands careful reconciliation. The September 17, 2026 retail-investor release says individuals' ETF custody rose from R$26.9 billion to R$35.1 billion and describes the movement as 62% annual growth. Those printed amounts produce a rise of R$8.2 billion, or 30.5%, not 62%. A second B3 release dated July 30 says the broader market's ETF stock reached R$126 billion in June. The contradiction is not evidence of accounting misconduct. It appears to involve mixing annual and half-year reference periods and different ETF populations.[1][3][2]

A primary-record comparison suggests why. B3's March 2026 publication identifies R$26.9 billion as retail ETF assets in custody at the end of December 2025, following R$18.1 billion in December 2024. The September release itself repeats that December-to-December history and later has an official describing approximately 30% growth over the first six months of 2026. That is consistent with R$26.9 billion increasing to R$35.1 billion. Yet its earlier paragraph places the same amounts in a sentence about annual growth of 62%. The credible inference is that different period comparisons have been compressed into a single description, not that the published totals themselves are fictitious.[1][2]

For brokerage executives and financial publishers this matters more than a pedantic percentage correction. Retail-investor acquisition, total ETF market size, trading volume and assets held in custody are distinct business metrics. A platform claiming explosive growth from an exchange's press release might choose the largest percentage without identifying its underlying period and population. If a 62% annual asset-growth rate and a 30.5% six-month growth rate both exist, each requires its own correct starting date and denominator. The fact that B3's separate documents may support both does not make them interchangeable in one comparison.[1][3][2]

There is a separate product-coverage trap in the exchange's label 'ETF'. The July total-stock announcement refers to exchange-listed fund products across a wide range of exposures, including equities, fixed income, international markets and digital assets. The September retail-investor release counts people with ETF positions in custody, using a retail-investor population. Someone who buys an ETF providing international stock exposure through a Brazilian exchange remains a Brazilian ETF holder for this statistic; that person is not necessarily a new owner of a conventional domestic share or a customer of a foreign broker. This reinforces the need to preserve product definitions when estimating how much market growth creates addressable revenue for local intermediaries.[3][1]

Even if the retail custody balances were perfectly comparable across dates, per-investor averages would need explanation. Dividing R$35.1 billion by approximately 862,000 retail ETF holders yields about R$40,700 of custody value per counted person. That number is an arithmetic mean of an aggregate stock divided by a population count, not the median ETF account and not the typical investment of a newly acquired customer. Ownership is unevenly distributed, existing investors can add to positions, and market prices can rise or fall. B3 reports different series for custody, total stocks and trading; an accurate average therefore needs the exact population, product scope and measurement date disclosed beside it.[1][3]

Reconstructing the calendar bridge

The March B3 statement says retail ETF investors held R$18.1 billion in December 2024 and R$26.9 billion in December 2025. That is an R$8.8 billion increase, or 48.6%, consistent with the exchange's rounded 49% annual growth figure. The September statement says June 2026 retail ETF custody was R$35.1 billion. Against the December 2025 base of R$26.9 billion, the increase is R$8.2 billion, or 30.48%. The figures therefore form a coherent two-step series over eighteen months when the December and June dates are preserved.[2][1]

A full twelve-month comparison from June 2025 to June 2026 would instead require June 2025 custody. If the 62% annual growth figure were exact, the implied earlier value would be R$35.1 billion divided by 1.62, or around R$21.67 billion. That is a calculated implication, not a direct observation of the missing source table and not a newly verified B3 June 2025 value. It explains how a 62% annual change could be mathematically compatible with a 30.5% half-year change if the two percentages refer to different baselines. The September paragraph's explicit printed R$26.9 billion starting value does not itself support the 62% claim.[1][2]

The time dimension is visible in customer counts as well. B3 says retail ETF investors rose from 721,700 in December 2025 to approximately 862,000 in June 2026—an increase of 140,300, or 19.44% in six months. Using its stated June 2025 base of about 639,000 gives an increase of approximately 223,000, or 34.9% over twelve months. Both figures can legitimately describe the same June 2026 endpoint. An investor-acquisition chart that takes 223,000 as a six-month inflow would exaggerate the period-specific result by around 83,000 individuals.[1][2]

Neither investors nor their ETF shares are necessarily new to the financial system. The Brazilian CPF identifier measures unique natural persons under the selected custody definition, not all individual trades, nor the number of brokerage accounts across every institution. An investor can hold multiple ETFs and use different brokers. A new ETF holder may already own shares, government securities or funds, and a new custody position does not establish how much additional saving entered the country. Changes in customer counts cannot be treated as gross brokerage account openings for a specific commercial platform.[1]

Total ETF stock is a different number again

B3's July 30 release describes a market-wide ETF stock of R$126 billion in June 2026, up from R$91 billion in December 2025 and R$64 billion in June 2025. That is about 38.5% over six months and 96.9% over twelve months, on that broader measure. The September retail release describes R$35.1 billion held by individual investors at the June 2026 endpoint. Those two values are not competing estimates for exactly the same population. The larger figure concerns a broader ETF stock, while the smaller one is explicitly a retail custody measure.[3][1]

As an illustrative scale comparison, R$35.1 billion is about 27.9% of R$126 billion. That calculation is useful only if the two reported stock measures use compatible product and valuation perimeters and refer to the same date. The B3 releases do not supply a full reconciliation of institutional, foreign and individual holdings on those exact definitions. Therefore 27.9% must not be promoted as an independently certified retail ownership share of every ETF listed on B3. The correct conclusion is that the releases describe materially different asset populations and that the exact ownership bridge remains to be verified.[3][1]

The July release also reports 204 ETFs listed in June, up from 175 in December 2025 and 131 in June 2025. Product variety therefore expanded substantially. But a new product listing is not the same as a new investor, and a larger fund stock can result from market appreciation as well as subscriptions. The value of assets in custody reflects market prices, currency movements and flows; it cannot be divided by the increase in ETF investors to measure the average investment made by newcomers. That calculation would mistakenly assign all valuation and incumbent-flow changes to new customers.[3]

A further measurement difference concerns trading volume. B3 reported ETF average daily trading value of R$1.85 billion in the first half of 2026, up 16% from the previous year's corresponding period. At the same time, its June 2026 ETF trading distribution included institutions, non-residents and individual investors with different shares. This is a transaction-flow measure, not the stock of funds in custody. Comparing a 16% activity increase directly with a 96.9% one-year stock increase could be informative, but the periods and definitions must first be aligned. It cannot prove liquidity improved for every ETF or that spreads narrowed.[3]

The exchange identified individual investors as responsible for about 14.9% of ETF trading volume in June, compared with 55.3% from institutions and 26.7% from non-residents. Those rounded figures do not sum to 100%, reflecting other participant categories and rounding. They show that more individual ETF holders do not automatically mean individuals dominate transactions. A brokerage platform may see a rising number of custody accounts while institutional activity remains the larger source of exchange trading notional. That distinction changes revenue, execution and product-development assumptions.[3]

The conflicting numbers deserve a fair, exact interpretation

The most defensible reading is not that the exchange is hiding assets or intentionally misstating market growth. The September release includes figures consistent with two different growth intervals, and its later commentary explicitly refers to about 30% custody growth over six months. The phrase identifying the R$26.9 billion-to-R$35.1 billion movement as 62% annual growth is mathematically inconsistent with those printed endpoints. That might be an editorial date substitution, a mislabelled baseline or the conflation of two valid growth metrics from a longer table. Without a final underlying June 2025 custody row, a definitive cause cannot be established.[1][2]

There is also a smaller count variation across B3 releases. The July announcement places investors with ETF custody at approximately 870,000 in June; the September retail analysis gives around 862,000. It also uses slightly different comparisons with the preceding year. This could arise from source populations, data extraction dates, rounding or definitions of relevant accounts. The difference is worth obtaining an explanation for, but does not justify an inference that one of the datasets is fabricated. Analysts must preserve each release's population and methodology rather than silently replacing one figure with the other.[3][1]

B3’s September statement itself supplies an important qualification: an official describes roughly 30% custody growth during the first six months. That is consistent with the printed December-to-June amounts. The separate 62% annual figure may use a different June 2025 base, which was not independently established in the material reviewed here. No comment request was sent. This analysis identifies the unresolved baseline, not intentional misstatement, an accounting violation or an error in the annual figure itself.[1][2]

Implications for market statistics and investor communications

The key methodological distinction for financial intermediaries is between new investors, funded brokerage accounts, ETF product count, total fund stock, retail custody stock and transaction notional. Each is capable of increasing at a different rate and none alone establishes the profitability of a broker, the returns earned by clients or the cost of execution. A data vendor building a Brazilian trading-sector dashboard should keep the source date, numerator, denominator, reference period and coverage group as first-class fields rather than store every percentage under a generic growth metric.[1][3]

The new analysis consists of a multi-release temporal reconciliation and identification of a clearly bounded numerical inconsistency. The exact R$8.2 billion rise from R$26.9 billion to R$35.1 billion equals 30.5%; if 62% is also valid, it must refer to another base. The separate broad ETF stock of R$126 billion cannot be treated as the retail custody figure, and customer counts differ across two B3 publications. The records warrant scrutiny precisely because Brazil's capital-market expansion is significant enough that a wrongly defined growth rate could propagate into broker presentations and investor research.[2][1][3]

A definitive annual reconciliation would require the underlying June 2025 retail-custody balance and consistent definitions for the two June investor counts. Until then, the established result is narrower: the printed R$26.9 billion and R$35.1 billion endpoints imply approximately 30.5%, while a 62% annual change requires another baseline. Brazil’s ETF market is expanding on the published series, but the period and asset population must travel with each growth figure.[1][2][3]

Report an error or submit an update