Polymarket Event-Linked Markets

The August 23 reference move collapsed lower bounds on ETH, BNB, DOGE, and SOL within 26 seconds — four markets in one family repricing on one underlying move, measured in the archive as four series moving in lockstep.

BTC was the odd one out on the same event: its lower bound took 37 minutes to repriced, a family-divergent lag that is invisible unless you compare the members.

Measuring the family together is what makes the structure usable: a 26-second synchronized collapse and a single 37-minute lagger are both signals that only exist relative to the co-series.

• Each market in a family has its own books, prices, and metrics at 250ms.
• The archive stores every member of the family, so a co-movement study is a many-market join on timestamp, not a manual spreadsheet.
• Event windows concentrate the action: the 65-second ETH repricing and the 26-second four-coin collapse are both bounded windows inside a single family rowset.
• Join with the underlying reference (futures price, CPI numbers, chain data) on the UTC second to attribute the family behavior.

• Lead-lag: which family member re-prices first (BTC lagging by 37 minutes is the canonical example).
• Dispersion spikes: family variance across members in the second of an event reference move.
• Liquidity flight: whether depth on the soon-to-settle members migrates to the unresolved ones.

• Pick an event family — four coins with lower bounds on the same date — and pull one hour around the reference time at 250ms for all four markets.
• Join the four series on the UTC second and compute the per-second cross-family spread of mid changes.
• Mark the second of the first member's reprice and the last member's; the lag is your family coherence number (the recorded 26-second collapse, the 37-minute lagger).
• Compute per member the depth swing across the event window (bid and ask sides separately) and rank the members by depth loss.

Event-linked markets are the strongest argument for whole-family data: the interesting structure — which member leads, which lags by minutes, which loses 96% of bid depth — only exists relative to the other family members.

The cross-market view is precisely what single-market feeds cannot give you; you not only buy the right market for the trade but the right family for the context.

The archive's contribution is that every member is stored at the same grid, so family studies are joins, not data-collection projects.

For event trading, the family series should be a standing dataset — refreshed on the event calendar, kept at 250ms for the window, coarse after.

Lead-lag tables compute from the family join and update live; they are the trading-ready version of the study that made the structure visible.

Event families are where Polymarket data pays for itself: the 26-second four-coin collapse and the single 37-minute lagger only exist as measurements when all the members sit in one dataset on one timestamp grid.

The family join is the whole method: pull every member at 250ms for the event window, mark the first and last repricing second, compute the depth swings per side, and the lead-lag structure of the event writes itself.

Whether you trade the family or study it, the standing dataset — refreshed on the event calendar, fine during the action — is the asset; ad-hoc pulls on the day will always arrive after the structure has moved.

Cross-market work is the honest edge the archive provides: single-market feeds cannot show you that BTC lagged 37 minutes while four peers repriced in 26. That asymmetry is a decision, and now it is data.

• cross market data
• repricing data
• settlement data