Reference

Data, coverage, and definitions

The short version of what can be compared, what cannot, and which data capabilities unlock each analytical module.

Comparison rules

Unavailable is not zeroA missing fee, outcome, or open-interest series is omitted and named—not plotted on the axis.

Aligned windows by defaultComparison views use a common recent window; each venue's full history is an explicit alternate view.

Shares and levels are differentAbsolute volume answers scale. Within-venue share answers shape. They are separate controls.

Participant P&L requires identityAnonymous buyer-price bias is not labeled taker P&L when aggressor or buyer identity is unknown.

Capability registry

Module	Required data	What becomes possible
Activity	Date + normalized quantity	Volume, growth, market share
Trade size	Trade-level prints	Size composition, block share, largest trades
Taker behavior	Price + quantity + aggressor side	Taker dollars, yes/no preference
Open interest	Comparable daily snapshot	Standing positions and carefully defined turnover
Fee economics	Price/quantity + computable schedule	Effective fee rate and exchange revenue
Calibration	Price + outcome + event clustering key	Actual-vs-implied and honest intervals
Participant P&L	Price + outcome + buyer/aggressor identity	Taker/maker results and market-level P&L
Parlay structure	Parlay identity + leg count	Adoption, length, correlation
Parlay economics	Price + parlay outcome + buyer identity	House edge, realized P&L, cash-outs

Core definitions

Volume is the venue's reported trading quantity. ForecastEx reports matched pairs; most other venues report contracts. Mixed-unit tables never total them.
Taker-side volume is the dollars paid by the aggressor, not contract face value.
Fee cost is what one trader pays to execute. Exchange revenue is what the venue retains across charged sides after published rebates.
Open interest is an end-of-day stock. Turnover is a secondary diagnostic defined literally as reported volume divided by prior reported open interest over the stated window.
Calibration error is actual win rate minus implied probability. Error intervals cluster related prints at the event level where the data supports it.
Taker P&L is used only where aggressor identity is observed. Otherwise the dashboard says buyer result or price bias.

Taxonomy and chart-level notes

The detailed source-label mappings stay next to the canonical comparison that uses them:

Each chart keeps only one useful sentence in the reading path. Sampling, joins, mappings, and uncertainty details belong in a collapsed chart note or on this reference page—not in a wall of defensive prose before the result.