Polymarket Repricing Data

On 2026-08-23, ETH, BNB, DOGE, and SOL all printed lower bounds inside a 26-second window while BTC's market took another 37 minutes to do the same. The record shows depth leaving and ask size multiplying before each print, with the price following the depth.

For reproduction: pull the affected series' books at 250ms across the recorded window; the divergence between BTC and the rest is measurable as bound-hit timestamps per series.

During an ETH selloff, Binance printed its low at 14:42:50 UTC after falling from $2,452.84 to $2,393.34 (about $59 in 65 seconds); Polymarket's 12:00–16:00 window contract hit its 0.065 low at the same clock second, preceded by a collapse from 0.815.

The claim depends on timestamp alignment across venues. The recorded Polymarket book stamps events in UTC at 250ms, which is the resolution required to make the same-second assertion.

A BTC strike ladder repriced from 0.044 to 0.996 in a single second, while the same measurement recorded a 99.6-cent hole in the middle of the ladder — a mid of 0.4995 with nothing executable behind it.

This pairs two phenomena in one record: a real repricing (whole ladder, one second) and a book artifact (empty rung, unexecutable mid). Distinguishing them is exactly what full-book capture enables.

• Take the recorded ETH case (Binance futures 2,452.842 → 2,393.34 between 14:41:45 and 14:42:50 UTC) and pull the Polymarket mid and ladder for that same window at 250ms.
• Mark the exact second the mid crossed from 0.815 to 0.065 and the floor settled at the new reference; the same-second delivery at 14:42:50 is the anchor of the study.
• Compute the mid-to-floor gap on each side of the reprice; the divergence between headline and executable is the real prototype of repricing risk.
• Extend to the August 23 family event: four coins' lower bounds collapsed in 26 seconds while BTC took 37 minutes, with ETH bid depth −96% and ask depth +491% in the window.

Repricings are where the venue earns its bad and good reputation at once: the ladder can move the full width (0.044 to 0.996) in one second, yet the same-second reference alignment (a futures print and a Polymarket floor settling together at 14:42:50) shows the market does close gaps fast when it decides to.

Depth is the honest co-conspirator: the August 23 redistribution — bid depth collapsing 96% while ask depth multiplied by 491% — is the mechanism by which repricing picks sides, and the archive's full-ladder frames are the only record that shows the mechanism.

A repricing study is a four-field story: reference level, UTC second, mid, and executable floor. The archive hands all four on the same 250ms grid.

The measured contrasts — 65 seconds for ETH, 26 seconds for four coins, 37 minutes for BTC — are the texture that separates a reprice story from a reprice number.

Before you trade a repricing, replay it; the same-second reset is the phenomenon, and the ladder frame at that second is the evidence.

Repricing data is the link between market microstructure and the reference events that drive it; pair it with settlement-data when the study is about endings and with cross-market-data when it is about the whole family.

Each measured case in this page is stored at 250ms with the same field names as every other dataset, so the switch from one page's material to another is a query-parameter change, not a migration.

That shared format is the reason a family can be studied across settlement, repricing, replay, final-seconds, and cross-market frames with one pull and one schema.

The cleanest way to see a reprice is by replaying the ladder frame by frame across the event; at 250ms the sequence is explicit, and the same-second alignments become visible instead of asserted.

• Research blog
• Order book data
• Backtesting data
• Crypto data