Index of /v3/
Our own capture. Live data from PendulumFlow.
Format, provenance and known issues for this data: V3 · our capture. What we measured about it, and where it falls short: the audit.
Checksums for this prefix:
SHA256SUMS.txt, verify a download with
sha256sum -c. It currently names 1184 hourly Parquet files and 2279 sidecars (the manifests and capture-timing files), through
2026-10-06T13. Hours after that are served and are NOT in
the ledger yet; until its next extension, verify those against the sha256 in
each hour's own manifest.json.
Most hours here hold two files: the Parquet and its
manifest.json. 1104 of 1193 also carry a witness_stats.json holding capture timing, not market data (2026-08-21T19 to 2026-10-06T22, with 4 hours inside that range still waiting for one). It is optional:
an hour without it is not missing any data.
Download this prefix in bulk
You do not need to open the hour pages, and you should not scrape this one.
SHA256SUMS.txt lists every file under this prefix, so it is both the
download list and the verification file.
# every file URL under /v3/
curl -s https://archive.pendulumflow.com/v3/SHA256SUMS.txt | awk '{print $2}' \
| sed 's|^|https://dl.pendulumflow.com/v3/|'
# fetch them, four at a time, keeping the directory layout
curl -s https://archive.pendulumflow.com/v3/SHA256SUMS.txt | awk '{print $2}' \
| xargs -P4 -I{} curl -sf --create-dirs -o {} https://dl.pendulumflow.com/v3/{}
# then verify the lot, in the directory you downloaded into
curl -sO https://archive.pendulumflow.com/v3/SHA256SUMS.txt && sha256sum -c SHA256SUMS.txt
# fetched only PART of it? filter the ledger to MATCH WHAT YOU TOOK
grep <the prefix you fetched> SHA256SUMS.txt | sha256sum -c -Fetched only part of it? Filter the checksum list to what you took first, for example
grep 2026-10-06 SHA256SUMS.txt | sha256sum -c - for one day, or every file you
skipped prints as FAILED.
Add | grep 'parquet$' to skip the manifests. Ranges within a file work
too, see the format notes.
1193 hours, 3490 objects.
Thin hours inside these windows line up with the venue pausing, though that is not the whole story for every one: the 26 August entry, for one, records a stretch where the venue was emitting and we were not receiving. Polymarket's status page recorded each incident; the link on each line is their record, and details is what we measured and what we are not claiming.
2026-09-23T18:06Zto2026-09-23T18:24Z: Trading degraded for eighteen minutes. Trading degraded on the predictions CLOB; investigating at 18:06, resolved at 18:24, 18:06 to 18:24 UTC (their record). 1 hour · details2026-09-19T19:38Zto2026-09-19T20:20Z: Trading paused by degraded block production on Polygon. Trading paused; the venue attributed it to degraded block production on Polygon PoS delaying transaction confirmations, and cited Polygon's own incident, 19:38 to 20:20 UTC (their record). 2 hours · details2026-09-19T16:54Zto2026-09-19T18:58Z: Trading degraded, then cancel-only mode. Trading degraded across the CLOB, both websocket transports, market data and on-chain settlement; the venue moved to cancel-only mode at 17:59 while it found the root cause, and re-enabled trading at 18:58, 16:54 to 18:58, cancel-only from 17:59 UTC (their record). 3 hours · details2026-09-11T16:22Zto2026-09-11T17:02Z: Trading degraded for forty-one minutes. Trading degraded on the predictions CLOB; the venue reported a fix in progress at 16:22 and resolution at 17:02, 16:22 to 17:02 UTC (their record). 2 hours · details2026-09-03T17:06Zto2026-09-03T21:35Z: Venue paused trading, then twice more in cancel-only mode. Delayed open order read responses; trading paused, resumed through a cancel-only window, then returned to cancel-only when the degradation came back, 17:06 to 21:35, paused from 17:06, cancel-only until 19:35, cancel-only again from 19:58 to about 20:34 UTC (their record). 5 hours · details2026-09-03T13:07Zto2026-09-03T13:26Z: Venue paused trading for about twenty minutes. Delayed open order read responses; trading paused while a fix was applied, 13:07 to 13:26 UTC (their record). 1 hour · details2026-09-01T22:30Zto2026-09-02T04:58Z: Venue restart, then a paused database migration. Delayed open order read responses traced to database replica lag; an emergency CLOB restart, then a longer paused window to apply a fix with their infrastructure provider, opened 20:21 on 1 September, restart 22:30 to 23:54, trading paused again 01:30 to 04:58 UTC (their record). 6 hours · details2026-08-31T23:09Zto2026-09-01T01:07Z: Venue paused trading, cancel-only mode. Delayed open order read responses; trading paused and put in cancel-only mode while a fix was applied, 20:28 to 01:07, with trading paused from 23:09 and resuming about 00:59 UTC (their record). 2 hours · details2026-08-31T06:30Zto2026-08-31T10:47Z: Venue stopped trading for four hours. Trading paused while a fix was applied, with the resumption target moved twice, then cancel-only before trading resumed, opened 06:23, trading paused 06:30, cancel-only from 10:24, trading resumed 10:40, resolved 10:47 UTC (their record). 5 hours · details2026-08-30T17:55Zto2026-08-30T18:29Z: Venue paused trading for about half an hour. Trading paused while the venue investigated, then cancel-only while a fix was monitored, opened 17:36, trading paused 17:55, cancel-only from 18:18, resolved 18:29 UTC (their record). 2 hours · details2026-08-28T04:00Zto2026-08-28T05:00Z: Scheduled CLOB maintenance. Scheduled CLOB maintenance, 04:00 to 05:00 UTC (their record). 1 hour · details2026-08-26T04:03Zto2026-08-26T07:57Z: Venue maintenance, feed interruption. Scheduled CLOB maintenance, planned for about ten minutes, extended twice, 04:00 to 07:30 UTC (their record). 4 hours · details2026-08-22T13:13Zto2026-08-22T13:51Z: Venue paused trading for about forty minutes. Trading paused, then monitored before it resumed, 13:13 to 13:51 UTC (their record). 1 hour · details2026-08-19T09:19Zto2026-08-19T10:58Z: Trading outage, order submissions failing. Trading outage with order submission failures, 09:19 to 10:58 UTC (their record). 3 hours · details2026-08-19T04:00Zto2026-08-19T05:30Z: Venue infrastructure maintenance. Announced infrastructure updates, run as maintenance, announced 17:30 on 18 August, ran 04:00 to 05:30 UTC (their record). 2 hours · details
23 hours are marked small, meaning under a quarter the size of the median hour here (1.0 GiB). That is a fact about volume, not a verdict about capture. Each marker is a link: where the exchange recorded an incident covering the hour it leads to what we know about it, and where nothing is on record it leads to the definition. Hover for the reason without leaving the page. Some of these hours pass the completeness check for all sixty minutes and simply had little happening; a quiet market and a capture problem look the same from the outside. If the difference matters to your work, read the hour's manifest and the audit rather than inferring from the size.
Looking for data we don't publish yet? Tell us.
Pendulum Flow Discord
Something wrong with an hour, or a file you need that is not here? Say so in the Discord, where the people who run the site and prediction market data geeks compare notes.
Join the Discord