Event window
T−10CATALYSTT+10
01
Event studies
Measure repricing around elections, policy decisions, conflict developments, and other dated catalysts.
For quant & systematic research
Research the signal as it actually existed.
A half-hourly, point-in-time record of how prediction markets priced elections, conflict, and policy since January 2026.
Scoped evaluation extracts are available under a data agreement.
Point-in-time recordAs published
One observation
Preserved with the state that produced it.
as_of<timestamp>
series_id<theme>
composition_version<version>
fresh_markets<count>
integrity_statepublished
The reconstruction problem
A backtest cannot know what the dataset learned later.
01
Rebuilt later
Resolved contracts can disappear. The surviving universe already knows the outcome.
02
Captured in time
Each observation keeps the composition and quality state available at that moment.
03
What changes
The model tests the signal – not a cleaner history assembled with hindsight.
How the archive is made
Capture first. Research later.
Every half-hourly window becomes a durable record before resolution, reconstitution, or later information can change what was observable.
Data lineage30-minute cadence
01
Market state
Prices, liquidity, and contract status at the window.
02
Capture
A scheduled half-hourly observation – before resolution changes the record.
03
Published series
Aggregate level, composition version, and integrity state.
04
Research input
A time-aligned series ready to join to the assets you study.
Research workflows
Test the event – not the venue capacity.
Use event probability as an explanatory variable, timing input, or regime signal alongside the instruments you already trade.
Regime filterDaily
Elevated Baseline
02
Regime filters
Segment exposures or model behavior when market-implied event risk enters a different state.
Aligned series
Belief series Market data
03
Macro overlays
Align thematic probability series with rates, volatility, commodities, sectors, or proprietary signals.
Two data layers
Verify the benchmark publicly. Evaluate the deeper archive by agreement.
The public layer is enough to reproduce and cite the aggregate series. Scoped research extracts expose the window-level inputs needed for deeper evaluation.
Public snapshots
Aggregate record
- ✓Half-hourly series history
- ✓Composition membership and weights
- ✓Reconstitution and chain-link records
- ✓Series metadata and freshness counts
Research evaluation
Scoped historical extract
- +Per-window constituent prices
- +Bid, ask, and liquidity observations
- +Market-level freshness state
- +Data dictionary and field definitions
Point-in-time means operationally
The history stays honest after the outcome is known.
Archive integrity is a data-model property, not a marketing label.
01
As-published timestamps
The observation is bound to when it was calculated and published – not when it was later retrieved.
02
Composition by version
Your test can recover which markets and weights defined the thematic series at each point.
03
Quality states in the data
Freshness, staleness, and integrity conditions stay visible to the modeler.
04
No silent restatements
The series you tested does not quietly become a cleaner series after the fact.
53,000+
Half-hourly observations
30 min
Valuation cadence
Jan 2026
Publishing since
No
Silent restatements
Research evaluation
Bring one signal hypothesis.
We will scope the relevant history, deliver an evaluation extract and data dictionary, and let your team test the signal in its own environment.
01
Define the question
A theme, time horizon, catalyst set, or market relationship you want to test.
02
Evaluate the actual data
Receive a scoped extract with timestamps, quality states, and field definitions.
03
Decide with evidence
Commercial terms only if the dataset produces research value.