Time SeventeenDocs
Concepts

Index Lifecycle

Methodology publication, rebalance, suspension, degradation, correction, and point-in-time reproduction.

Draft v0.1 — design contract, not yet a live service

This documentation set defines a proposed Time Seventeen protocol for review and implementation planning. https://api.t17.capital is the approved and reserved REST origin, but neither that origin nor any documented endpoint is claimed to be live. The WebSocket origin remains unspecified. Credentials, chain IDs, contract addresses, rate limits, and SLAs remain unapproved; IDs, prices, and timestamps shown throughout are illustrative, not real production data.

An Index moves through a lifecycle of its own, separate from the lifecycle of any single Market it contains.

Index lifecycle stages

StageDescription
CreationAn Index is defined with a methodology reference, initial constituent proposal, and status draft.
ReviewConstituent eligibility, weights, and methodology are reviewed before publication.
PublicationThe first Index Version is published with effective_from; the Index status becomes active.
RebalanceA new Index Version is published with its own effective_from; the prior version receives an effective_to and remains permanently readable.
Constituent addition/removalExpressed as a change between two Index Versions, never as an in-place edit of a published version.
SuspensionThe Index (or a specific Constituent) is marked suspended when inputs cannot be trusted; publication of new Index Price Observations pauses or is flagged suspended.
DegradationThe Index continues publishing but with a degraded state and a documented reason (for example a stale or missing constituent observation) rather than silently omitting the issue.
ArchivalAn Index that is retired keeps its full version and observation history readable; archival changes status, it does not delete records.
Correction / backfillA previously published Index Price Observation is superseded by a new revision carrying supersedes_id and correction_reason; the original observation remains in the history.

Point-in-time reproducibility

Because an Index Version is immutable after publication, a client can always answer: “which methodology, constituent set, weights, source observations, ratings, and calculation status produced this index value?” This is why rebalances always publish a new version rather than mutating an old one, and why corrections supersede rather than overwrite.

Handling degraded inputs

The Index Lifecycle documents how constituent-level problems surface at the index level instead of being hidden:

  • Stale observation — a constituent’s most recent Market Price Observation is older than its expected freshness window; the Index Price Observation reflects this in its coverage_ratio and quality-state field (quality_state in the API reference).
  • Missing observation — a constituent has no eligible observation for the calculation window; it is excluded and listed under a missing-constituents field, not defaulted to zero.
  • Outlier observation — an observation that fails validation rules is excluded and listed under excluded_inputs with a reason, not silently averaged in.
  • Venue outage — Markets on an unavailable Venue are treated as missing inputs for the duration of the outage.
  • Market halt / resolution — a Constituent’s underlying Market transitioning to halted or resolved status changes how (or whether) it contributes weight, governed by the Index Version’s rebalance and normalization rules.

Rebalance schedule and normalization

Rebalance cadence, weight normalization formula, and weight-capping rules are methodology-version concerns, documented per methodology rather than hard-coded into the Index resource. Draft v0.1 does not fix a numeric rebalance interval, weight cap, or normalization constant — those are production configuration pending an approved methodology.

Where to go next