Coinbase tokenized equity volume on Base, by oracle age at execution

Grayscale published a chart in August showing value locked in tokenized stocks across lending and trading protocols reaching roughly $100m, under the heading "tokenized equities have found liquidity, financial utility is next". Financial utility in that sense is lending, and lending reads a price from an oracle. This measures both halves for one set of assets: how much traded, and how old the published value was when it did.

Scope

This is not a reproduction of that chart. Grayscale's covers several chains and several protocols over fourteen months, sourced from Allium. This covers the Coinbase B20 tokenized equities on Base, their Chainlink aggregator feeds, and their Aerodrome Slipstream pools, from our own reads.

There is no value-locked series here because we do not read pool token balances. What we read is the tick book over a bounded range, which is executable depth and a different quantity. The series below is traded notional, taken from Swap logs.

The window is 12 August to 6 September 2026, twenty-six complete UTC days. That is the full extent of our record for these pools, not a chosen sample.

Traded notional, by day

Over the window the ten pools recorded 688,969 priced swaps totalling $262,587,871.

For scale, the cross-chain, cross-protocol figure Grayscale plots is a stock of about $100m in value locked, while this is a flow of $262.6m over 26 days at one venue for one issuer. The two are different quantities and are not directly comparable.

The venue count went from four to ten on 3 September

Four pools traded from mid-August: AAPLc, GOOGLc, METAc and NVDAc. On 3 September, at adjacent blocks minutes apart, six more were opened: AMZNc, MSFTc, MSTRc, SNDKc, SPCXc and TSLAc.

The six new pools reached 44.8% of daily volume on their second full day, and $13,145,511 over the four days they existed inside the window.

Traded notional by oracle age at execution

Each priced swap is matched to the feed for its ticker, and to how long that feed had been carrying the same published value at the swap's block time. Buckets are exclusive.

At or beyond six hours: $88,464,039, or 33.7%. At or beyond one hour: $162,086,277, or 61.7%.

Six hours is the threshold at which a latency window enters our headline statistics. A US regular session is six and a half hours, so it is the shortest gap not accounted for by a single quiet session. The definition is on the methodology page, and the individual windows are listed at /latency.

Feed state at the time of measurement

At 12:39 UTC on 7 September 2026, all thirteen feeds carried values published on Friday 4 September.

The remaining eight fall between those bounds. The pools traded $11,562,756 on 6 September.

Our own coverage over that period: the poller reading these contracts recorded 937 consecutive successful reads with zero failures, and last read the chain sixteen seconds before this table was taken. Coverage including its gaps is published at /status, so the elapsed column can be checked against whether anyone was watching.

Latency windows by when they fell

A deviation-triggered feed with no off-hours heartbeat could plausibly be quiet only when the underlying market is. The completed windows do not divide that way.

Of 221 completed windows of six hours or more across the ten feeds with a pool, 159 were not weekends, and 111 of those 159 overlapped a session that was open at the time. Across all thirteen feeds, including the three with no pool, it is 175 of 251.

Note the direction of that measurement. Nothing here decides whether a feed is stale by consulting a calendar. Elapsed time comes from the updatedAt the contract records, and the session label is applied to the result afterwards. That ordering is what allows the split above to be informative rather than circular.

Depth on the new pools

Executable depth walked from the tick book at 12:22 UTC on 7 September, within 2% of mid:

The two thinnest of the pools opened on 3 September carry roughly two orders of magnitude less depth than the August pools, while the six together accounted for 34.5% of volume on 6 September. The full walk across all ten pools, the modelled price impact at a range of sizes, and the inverse question of what depth a market of a stated size would require, are in what it costs to get out of a Coinbase tokenized equity.

What is not in these numbers

No composite score and no ordering of the assets. Every figure is per feed, per pool or per window, with its denominator shown. Tables print in catalog order rather than by their own result.

Volume is not a measure of depth, and neither is a measure of risk. A pool can carry high volume and thin depth at once, which four of these do. The oracle-age table describes when trades occurred relative to the last published value; it does not establish that any of those trades were mispriced, which would require a reference price we do not publish.

No lending market has listed any of these tokens, so nothing above describes anyone's live exposure.

Method

Published values and their timestamps come from latestRoundData() on the Chainlink aggregator proxy for each asset. Trades come from the Swap logs of the Aerodrome Slipstream pool, which carry the post-swap price, liquidity and tick; a pool's price history exists only as those logs. All timestamps are the chain's rather than our polling clock, so a measurement does not depend on when we looked.

Volume counts priced swaps only. An unseeded pool emits zero-amount, zero-liquidity Swap events; a pool that has emitted four hundred of those has not traded $0, it has not traded, and those are different claims. Days are whole UTC days, so the day of publication is excluded.

The figures on this page were measured once, at 12:39 UTC on 7 September 2026, and are not recomputed when you open it. The monitor carries the current numbers.