Everywhere else on this site a trade is a 24-hour aggregate someone else
computed. Here it is a transaction. Every swap against a tokenized-stock pool on Solana is a
signature on that pool's address, so the trade can be read from the chain itself: which mint
moved, in which direction, at what price, paid for by which wallet. That is only true of the
DEX pools. The exchanges that list these tokens publish no per-trade feed without a key, so
this page covers the sampled pools and nothing else. Nothing below is interpolated or
smoothed: the tape shows decoded transactions, and the replay animates only the
hours that were collected.
ObservedCollector run CoverageDecoded trades from sampled on-chain DEX poolsLimitCentralized-exchange trades are not included
Showing the bundled sample fixture
(stocks/fixtures/stocks-trades.sample.json), not the published
capture. Do not cite any number on this page while this banner is visible.
Loading…
The tape
The accumulating database history, newest first and loaded 20 trades at a time.
The page re-reads it every few minutes, so a new collection appears on page one without a reload.
The 24-hour replay below uses the collector's complete rolling capture.
The twenty newest decoded swaps, newest first. A buy is the
pool losing the tokenized stock to a buyer, a sell is the pool gaining it. The
side is read from the pool's own balance change; transaction labels are ignored. Price is
USD where the pool's quote token has a USD rate, otherwise the price in the quote token itself.
routed marks a transaction that moved more than two mints: an aggregator or
arbitrage path in which this pool was one leg. Routed trades are mostly bots, so the tape hides
them unless you tick the box below; the replay and every total still count them.
…
…
Routed trades moved more than two tokens in one transaction: one leg of an arbitrage bot or aggregator path.
No decoded trades in this capture yet.
The last 24 hours, replayed
One bar per hour, stacked by venue, over the 24 hours ending at the last
collector run. The cursor sweeps the window in about half a minute and the counters above it
accumulate as it passes each hour. Every bar is built from decoded trades; nothing is drawn by
interpolation.
…
How this page is made
What the tape covers. The sampled Solana DEX pools only: the top pools by
24-hour volume across the tokenized-stock mints, refreshed on every collector run. There are
no CEX trades here at all: the exchanges listing these tokens publish per-trade data only
behind a key, so a token that trades mostly on an exchange will look quiet on this page even when
it is active.
How a trade is decoded. Each swap is read from the transaction's own token
balances, before and after, restricted to the accounts the pool owns. The pool's
change in the tracked mint is the size; its change in the quote mint is what was paid; the
price is one divided by the other, converted to USD with the pool's quote rate. The side
follows from the direction: a pool that gained the token was sold into. Nothing is taken from
an instruction name or a log line, so a swap routed through an aggregator is decoded exactly
like a direct one, and marked routed when the transaction moved more than
two mints.
Why routed trades are hidden by default. In a 24-hour sample of 2,762 decoded
trades (30 September 2026), 69% were routed, and about 4 in 5 of those ran through arbitrage or
bot trading programs rather than a public aggregator such as Jupiter; their median size was
about $24, against about $350 for a direct swap. The tape hides them so the trades people place
are readable; the checkbox above the tape shows them again. The page does not see which
program ran a trade, so a routed swap a person placed through an aggregator is hidden too.
Why failed transactions are shown. Most signatures on a busy Solana pool
never trade: they are bots racing each other and losing, and they fail. They have to be counted
to read the tape correctly: a pool with a 90% failure rate has heavy competition to trade,
and much less real trading than its signature count suggests. The failed share is per pool on the
chips, and over the whole window in the header line. Failed transactions are counted, never
fetched and never decoded, so they cost nothing and can never enter the tape.
What the replay shows. It animates the collected trades and only those. An
hour that began before collection started is hatched, because it was never sampled: a short
bar there means we were not collecting yet. Trades may still have happened. Volume is priced with each
pool's own quote rate at collection time, so a stale rate makes a stale USD number. The
trade counts never depend on a price.
Distinct traders are counted as distinct fee-payer wallets over the hours
the cursor has passed, from the trades themselves rather than by summing the per-hour counts,
which would count one wallet once per hour it traded in. An aggregator or a relayer can pay
the fee for someone else's swap, so read this as distinct paying wallets.
How fresh it is. RWA Sonar's server collects once an hour: it reads the
newest signatures of each sampled pool from a Solana RPC, fetches a capped number of the
successful ones, decodes them and stores them. This page only reads what the server stored
(your browser never contacts a Solana node) and re-reads it every five minutes. So the
tape is up to about an hour behind the chain, and it samples each pool once an hour. It does
not capture every trade: on the busiest pools more swaps happen between collections
than one collection reads.