---
title: "How to analyze any RWA in 30 minutes"
author: Mariusz Szyma
date: 2026-09-15
lang: en
canonical: https://szyma.co/en/blog/rwa/thirty-minute-teardown/
series: "RWA: Past, Present and Future"
series_part: 11/11
series_url: https://szyma.co/en/blog/rwa/
data_as_of: 4–10 Sep 2026
---

# How to analyze any RWA in 30 minutes

A number about this market is not information until four things are attached to it: which size of the market it counts, which source level it came from, which date it was read, and which unit it uses. This chapter turns that habit into a 13-question checklist with a time budget, a five-level source hierarchy, and a fixed reading order for contract permissions and oracle freshness that can be run on any new product without help.

## What you'll learn

- Name the three sizes of the RWA market and the most common way of mixing them.
- Rank a source from offering document down to secondary portal, and say where most circulating figures sit.
- Read contract permissions in the order that actually matters, upgrade first.
- Settle an oracle's freshness with one function call.
- Run a 13-question teardown in 30 minutes, and treat "not disclosed" as a result.

## The market has three sizes

Three correct, disagreeing readings of "the market" in 2026: distributed value excluding stablecoins about 38.32 bn USD (August 2026); represented value 377.74 or 435.93 bn USD, two RWA.xyz subpages, same day, 3 September 2026; CoinGecko-methodology market cap 19.32 bn USD (data to 31 March 2026). None of them is wrong. Each counts a different set.

RWA.xyz introduced the distributed and represented split on 21 November 2025. Distributed assets can leave the issuing platform and move between wallets. Represented assets cannot. The multiplier between the two categories is about ten in 2026, and most of the represented side sits on one institutional network. A headline about a tokenization market above 400 bn USD is, in the main, a closed network rather than a market on which an order can be placed. [Chapter 1](/en/blog/rwa/what-is-an-rwa/) sets the split; this chapter only needs the working rule that follows from it.

Reclassification looks identical to a capital crash in a time series. About 36 bn USD on 31 December 2025, via Stobox, against 14.1 bn USD on 1 January 2026, RWA.xyz via MetaMask, reflects two taxonomy changes, 21 November 2025 and 17 March 2026, not a 60% collapse. Represented value swung over 100 bn USD between readings weeks apart in summer 2026 and fell 8.55% in 30 days to August 2026. The working rule: represented-value moves of more than a few percent a month default to a dataset change, not a flow.

Same-month tracker disagreement without any reclassification is ordinary. Three July 2026 distributed readings sat 20% apart, 31 bn, 33.1 bn and 33.5 bn USD. Tokenized Treasuries in May 2026 had two readings 38% apart, above 11 bn against 15.20 bn USD, from the same underlying data. The direction of the market does not depend on the tracker. The levels do. From about 5 to 6 bn USD in early 2025 to about 38 bn USD in August 2026 is a six- to sevenfold move on any tracker.

> **Watch out:** A number with no date is not a number with a wide confidence interval. It is a number with a range of possible table states. Represented value moved more than 100 bn USD between summer readings a few weeks apart.

## A five-level source hierarchy

**Level 1, the offering document.** Prospectus, private placement memorandum, issuance terms, redemption rules. Binds the issuer. Contains what is nowhere else: who may use the primary market, and under what process the issuer may change terms or end the product.

**Level 2, the regulator filing.** Securitize's Form 424B3 states that NAV supplied to smart contracts comes from the fund administrator or the transfer agent, and that the platform's assets are not freely traded on a secondary market and have no continuous market price. The SEC's 11 December 2025 no-action letter for the Depository Trust Company sits here too.

**Level 3, the issuer announcement.** Often true and always selective, because the sender picks the metric.

**Level 4, the dashboard.** A dated measurement whose filter and methodology are usually hidden.

**Level 5, the secondary portal.** Repeats a dashboard read late, with no methodology.

Most of the most granular figures from summer 2026, exact amounts, holder counts, thirty-day changes, were available only through tertiary sources quoting RWA.xyz with a read date. That citation proves the number sat on a dashboard that day. It is not an independent confirmation. The operational rule: a number from a secondary portal needs a fresh read from the source and the reader's own date, and if that read cannot be done the number is citable with a caveat, not citable.

Level 4 is not self-consistent. Total RWA holders on 3 September 2026: 3,230,169 on `/networks` against 3,298,816 on `/platforms`, a 68,647 gap on the same day. If the metric were a set of unique addresses, the two sums would match. They do not, so it is an aggregate of separate lists, probably without full deduplication across assets. RWA.xyz's formal definition of "Asset Holders" was never established.

![Five levels of RWA research sources, from the offering document to the secondary portal; the one numeric point is the 68,647-holder gap between two RWA.xyz subpages on 3 Sep 2026.](https://szyma.co/blog/img/source-hierarchy.d333389f.svg)

## Five questions the offering document answers that no dashboard does

Read them in this order, because the first section decides the value of the rest.

Who has primary-market access, at what threshold, in what redemption window, and what exit exists for a holder who is not a primary client. BUIDL requires qualified-purchaser status and a 5 mn USD minimum under Rule 506(c) and Section 3(c)(7), with redemption in a T+0 or T+1 window. OUSG, USTB and USYC operate around 100,000 USD. BENJI starts at 20 USD. USDY is built as Regulation S outside the United States. For xStocks, redemption is open only to professionals through the issuer; a retail holder can only sell on the secondary market.

Under what process the issuer can change terms or end the product. That clause decided the largest wind-down in this class so far. On 12 May 2025 Anchorage Digital announced the acquisition of Mountain Protocol and the orderly wind-down of USDM, disabling mint the same day. About 10,820 holders, cap under 50 mn USD. Rewards ran to 11 June 2025, then the rate was set to 0%. On 22 August 2025 the primary market closed and remaining reserves went into a Uniswap pool against another stablecoin. The instrument that had been a claim for 1 USD against an issuer regulated by the Bermuda Monetary Authority became a token backed by USDC and honoured by an AMM. The underlying bills were fine. The product disappeared.

The legal structure and asset segregation: entity, jurisdiction, the document that constitutes the claim, and whether assets are legally separated from the issuer's estate. A BUIDL buyer does not buy Treasury bills. A PAXG buyer does not take title to a bar.

Who is custodian, administrator, transfer agent, auditor.

The full fee decomposition. Net yield depends on management fee, custody, platform, transfer agent and redemption fee, not on the underlying's rate.

An answer of "not disclosed" is recorded as a result. More than four such answers makes a product unmeasurable, whatever the quality of the marketing.

## Reading contract permissions in the order that actually matters

The order is the reverse of intuition. First: who can upgrade the implementation. That answer invalidates every other one, because if the code can be swapped, every answer about functions is temporary. Then the inventory of functions that move someone else's balance. Then today's holders of those roles. Then proof of use from event history. [Chapter 5](/en/blog/rwa/who-holds-the-keys/) is the worked reading; this is the checklist.

BUIDL: a single externally owned account owns the proxy and holds the MASTER role, able to swap the entire implementation via `setTarget` in one transaction with no timelock and no multisig, read 4 September 2026. `seize` requires the TRANSFER_AGENT role, held that day by one EOA and one contract. Event scan from 1 March 2024 to 4 September 2026: zero `Seize`, `OmnibusSeize`, `OmnibusBurn`, `Pause` and `Unpause` events, 208 `Burn` events, more than 1,000 `Issue` events. Zero events is confirmed for Ethereum only.

OUSG: no `seize`, `forcedTransfer` or `freeze` exists in the code. `BURNER_ROLE` was never granted to anyone, confirmed by scanning every `RoleGranted` event and calling `hasRole()`. Admin power sits on a 4-of-7 Safe for token admin, a 3-of-5 Safe for the KYC registry, and both a 4-of-7 and a 1-of-9 Safe for pause. Real power lives in the removable identity registry, not a freeze function.

The same product on different chains has a different attack surface. On Avalanche, BUIDL has `seize` and `omnibusSeize` but not `forcedTransfer`, `setTarget`, `freeze` or `mint`, with an empty EIP-1967 admin slot and an EOA owner with no code. On Solana it uses the Token-2022 `permanentDelegate` extension with a plain system-account upgrade authority. On Aptos, `ds_token::seize` is a Move entry function. Supply split, 4 to 5 September 2026: Solana 977.9 mn, Avalanche 564.2 mn, Ethereum 211.9 mn, Aptos 161.3 mn. Ethereum is about 11% of the four-chain total. An Ethereum-only permission audit describes a minority of the product.

Tool limits are part of the result. Avalanche's public node caps `eth_getLogs` at 2,048 blocks, so a full scan of 41,884,144 blocks needs 20,451 requests. Solana and Aptos usage was left unevaluated for lack of indexer access. "Not evaluated" is the correct answer, not "zero".

![Four-step order for reading a contract's permissions, applied to BUIDL and OUSG as of 4 Sep 2026; use on Avalanche, Solana and Aptos was not evaluated.](https://szyma.co/blog/img/permission-reading-order.7841a92d.svg)

## One function call settles an oracle's freshness

Eight production NAV oracles, read 4 September 2026 at 11:08 UTC, gave USCC 225 days stale with no error flagged, and ACRED never written (`roundId = 0`, `answer = 0`). The rest ranged 0.1 to 21.7 hours old. Seven of eight had `maxPriceDeviation` confirmed at 0, disabled. USCC's call reverts, so its state is unknown, not zero. [Chapter 6](/en/blog/rwa/oracles-and-nav/) is the measurement. The teardown only needs the method: call `latestRoundData()`, read `updatedAt`, and stop if the consumer has not checked freshness itself. Push feeds do not stream. They write when a threshold or a heartbeat is met, and they return the last value with a true old timestamp.

NAV is computed off-chain by a fund administrator and only carried by the oracle. Daily publication is not daily valuation if the underlying is priced quarterly and interpolated in between.

**An attestation** tests a narrow claim at one moment, usually that circulating tokens are covered by the fair value of reserves, and does not confirm reserve composition, encumbrances or internal controls. **An audit** examines a full period. USDC: monthly attestation (Deloitte) plus weekly disclosures. USDT: quarterly (BDO Italia). BUIDL: annual fund audit (PwC).

## Four numeric traps that share one root cause

A missing unit.

A number with no date, after represented value moved more than 100 bn USD between summer readings.

Represented value counted as if tradable. JMWH, 861 mn USD, listed on no DEX and no CEX. JSOY_OIL, 300 mn USD, mint and burn only. In the commodity aggregate they look like 1.16 bn USD of market.

A cumulative figure reported as a period figure. BENJI's 211 mn USD-plus of peer-to-peer transfers is a five-year total, not annual, 8.6% of the fund over five years.

Counting positions instead of value. A newly listed stock token at 763 USD counts identically to a billion-dollar fund. Tokenized stocks are about 77 to 79% of total RWA holders. Holders measure state. Active addresses measure use. [Chapter 10](/en/blog/rwa/what-works-and-whats-next/) scores those tests at segment level; the teardown computes them for one product.

Turnover-to-AUM, with the ceiling caveat, because transfer volume includes mint and redemption: Securitize about 1.4× a year; tokenized stocks about 51×; tokenized gold about 71×; PAXG about 31×. Active-address-to-holder: Securitize 61 active against 1,872 holders, 3.3%. DeFi-use: 7.4 bn USD of RWA deposited in DeFi in the second quarter of 2026, still under 10% of tokenized value; Aave Horizon 0.43% of the distributed market. Many of the largest assets were never designed to trade. Absence of turnover is a specification for a product that does not promise liquidity, and a failure for a product sold as liquid exposure to an illiquid asset. The marketing of the two is often identical.

## The 13-question framework, with a time budget

What it tokenizes. What rights the token actually gives. Where the underlying is and who controls it. The legal structure. How mint and redemption work. Who the client is. Where the yield comes from. The business model. Why this chain. The risks. Whether the product makes sense. Whether it has product-market fit. Whether it is distributed or represented.

Time budgets below are the author's estimate from how long each check took in the source material, not a separately measured figure. They add to about 30 minutes if the offering document is to hand and the chain is one the reader can query. They add to more than 30 if the document is missing, which is itself a result.

![Thirteen questions, what to check for each, where to check it, and how many minutes it should take: a working 30-minute teardown, with time budgets the author's estimate rather than a measured figure.](https://szyma.co/blog/img/teardown-card-rwa.2d6f616c.svg)

> **Example:** Run questions 5, 10 and 13 on USDM as it stood in May 2025 and the product is already scored before the wind-down is announced. Primary-market access, the issuer's right to end the product, and the fact that the token was distributed, were all in the documents. The dashboard still showed a Treasury-backed dollar until the reserves moved.

> **My take:** The skill this course is for is not collecting figures. It is refusing to interpret a figure until the unit, the date, the source level and the size of the market it counts are written next to it. A 30-minute teardown that ends with four answers of "not disclosed" is a completed teardown.

## Key takeaways

1. Three correct 2026 readings of "the market" disagree: distributed excluding stablecoins about 38.32 bn USD, represented 377.74 or 435.93 bn USD on the same day, CoinGecko-methodology cap 19.32 bn USD; mixing them is the most common error in the sector.
2. Represented-value moves of more than a few percent a month default to a dataset change; a one-day drop from about 36 bn to 14.1 bn USD at the 2025–2026 turn was two taxonomy changes, not a 60% collapse.
3. Five source levels, offering document then filing then announcement then dashboard then secondary portal; most circulating figures sit on the last two, and RWA.xyz's two holder totals on 3 September 2026 differ by 68,647.
4. More than four answers of "not disclosed" in the offering-document questions makes a product unmeasurable, independent of its marketing.
5. Permission reading order is upgrade, then function inventory, then today's role holders, then event history; BUIDL's strongest power is `setTarget` on one EOA, and zero `Seize` events are confirmed for Ethereum only.
6. Oracle freshness is one call to `latestRoundData()`; on 4 September 2026 USCC was 225 days stale with no error, and ACRED had never been written.
7. Four numeric traps share a missing unit: no date, represented counted as tradable, cumulative reported as periodic, positions counted as value.
8. Turnover-to-AUM is an upper bound, because volume includes mint and redemption; Securitize at about 1.4× a year is a register, gold at about 71× is a market.

## Glossary

- **Attestation**: a test of a narrow claim at one moment. It does not confirm reserve composition, encumbrances or internal controls.
- **Audit**: an examination of a full period's statements. BUIDL has an annual fund audit by PwC; USDC has a monthly attestation by Deloitte.
- **Source hierarchy**: five levels from offering document to secondary portal, ranked by how close the source sits to a statement that binds the issuer.
- **Upper bound** (turnover): on-chain transfer volume includes mint and redemption, so the ratio cannot be smaller than secondary trading and can be much larger.
- **latestRoundData()**: the standard call that returns `roundId`, `answer` and `updatedAt` from an aggregator-style price contract.

## Go deeper

- [Who holds the keys](/en/blog/rwa/who-holds-the-keys/): the permission-reading order here restates that chapter's method as a repeatable checklist.
- [Oracles, NAV and proof of reserves](/en/blog/rwa/oracles-and-nav/): the one-function-call freshness test is drawn from that chapter's own measurement of eight production feeds.
- [What works, what does not, and the next five years](/en/blog/rwa/what-works-and-whats-next/): turnover, active-address ratio and DeFi use are scored at segment level there; this chapter computes them for one product.

## Sources

- 🟢 RWA.xyz, `/platforms` and `/networks`, 3 Sep 2026: represented 377.74 and 435.93 bn USD; holders 3,230,169 and 3,298,816; stablecoins 303.04 to 303.21 bn USD.
- 🟢 RWA.xyz, `/stocks`, 2 Sep 2026: 3,918 assets, 2.54 million holders, 152,006 active addresses.
- 🟢 RWA.xyz, `/platforms/securitize`, about Aug 2026: 4.99 bn USD, 1,872 holders, 61 active addresses, 565.71 mn USD monthly volume.
- 🟢 Author's on-chain readings, Ethereum, 4 Sep 2026, 11:08 UTC: eight NAV oracles, BUIDL and OUSG permissions, event scans from 1 Mar 2024. Method in [chapter 5](/en/blog/rwa/who-holds-the-keys/) and [chapter 6](/en/blog/rwa/oracles-and-nav/).
- 🟢 CoinGecko, *RWA Report 2026*, data to 31 Mar 2026: 19.32 bn USD market cap; gold and perpetuals volumes.
- 🟢 SEC, DTC no-action letter, 11 Dec 2025; Securitize Form 424B3, 2026, on NAV source and absence of a continuous secondary price.
- 🟢 Mountain Protocol, *USDM Wind-Down Overview*, 15 Aug 2025.
- 🟡 Stobox citing Canton, year-end 2025 distributed reading of about 36 bn USD.
- 🟡 Tiger Research, 27 Feb 2026: JMWH 861 mn USD unlisted; JSOY_OIL 300 mn USD mint-and-burn.
- 🔴 CoinPaprika citing RWA.xyz, Aug 2026: distributed 38.32 bn USD; Canton represented-value figure, unverified in a later pass.
- 🔴 MetaMask citing Franklin Templeton, 3 Aug 2026: BENJI cumulative peer-to-peer volume.

## Caveats

The formal definition of RWA.xyz's "Asset Holders" metric was never established. Present that as a standing gap in the sector's data, not as a puzzle this chapter solves.

Time-per-question budgets in the teardown card are the author's estimate, not a measured figure. The caption states that.

Zero `Seize` events for BUIDL is confirmed only on Ethereum. Avalanche, Solana and Aptos were not evaluated. Reporting "zero" for those chains would be a fabrication.

Canton Network's about 371.3 bn USD of represented value remains unverified after a dedicated September 2026 search and is not used as a confirmed input here, even though it is the usual explanation for the size of the represented category.

The 56% and 70.6% idle figures from [chapter 10](/en/blog/rwa/what-works-and-whats-next/) are not restated as a single number in this checklist. The teardown exists so the reader computes turnover and activity for the product in front of them.