GCI 50 index methodology

The starting universe consists of crypto-native digital assets — protocol and application tokens with a transparent supply schedule and publicly verifiable on-chain issuance. Wrapped or synthetic representations of an already-eligible asset are excluded to avoid double counting.

Assets must have a minimum trading history of 90 calendar days and a published, auditable circulating-supply figure sourced from at least two independent providers.

Candidates must exceed a minimum free-float adjusted market capitalisation threshold and a minimum median daily traded value measured over the trailing 90 days. Both thresholds are reviewed annually and published in advance of each reconstitution.

A liquidity buffer applies at each review: existing constituents are retained if they remain above 75% of the entry thresholds, reducing unnecessary turnover.

Stablecoins and other assets designed to track a fiat currency or a basket of fiat currencies are excluded from the GCI 50 price-return benchmark, since their inclusion would mute the market signal the index is designed to convey.

Meme and meme-community assets are excluded through a canonical taxonomy review. The classification, evidence, source and methodology version are retained on each asset record; ambiguous utility and protocol assets are not classified as memes without supporting evidence.

Also excluded: tokenised money-market and deposit instruments, assets with unresolved custody or transferability restrictions on major venues, and assets subject to a credible, ongoing solvency or security event under review by the index committee.

Prices are sourced only from venues that meet GCI's exchange quality standard, which considers order-book depth, uptime, fee structure integrity, withdrawal reliability, published proof-of-reserve or equivalent disclosure, and regulatory posture in the venue's home jurisdiction.

The target reference price for each constituent is a volume-weighted composite across eligible venues, with outlier prints removed. The index is calculated in USD.

Prelaunch calculations use a multi-provider reference-price layer with freshness, outlier, source-family and divergence checks. A blocking quality condition suspends publication rather than substituting a sample price.

Constituent size is measured using a free-float aware market capitalisation: circulating supply adjusted for tokens that are demonstrably locked, vesting, or held in protocol treasuries and foundation reserves subject to disclosed lock-ups.

Weights are then assigned on a modified market-capitalisation basis. A single-constituent cap of 20% is applied at each rebalance, with excess weight redistributed pro rata across uncapped constituents. Capping limits the dominance of the largest assets so the index measures the market, not one asset.

The index reconstitutes and rebalances quarterly, effective after the close on the third Friday of March, June, September and December, using data observed on a reference date five business days prior. Selection results are published in advance of the effective date.

Between scheduled reviews, extraordinary-event provisions allow removal of a constituent following delisting from a majority of eligible venues, a chain halt or irrecoverable security failure, a redenomination or migration that breaks continuity, or a supply event that makes market capitalisation unmeasurable. Removed constituents are replaced by the highest-ranked eligible non-constituent, or the weight is redistributed if no replacement qualifies.

GCI 50 publishes a USD-denominated normalized basket price. At each scheduled rebalance, methodology-determined target weights are converted into fixed constituent balances using reference prices. Those balances remain fixed until the next rebalance, so between rebalances the published price moves solely because constituent USD prices move.

Formally, GCI50 Price (USD)ₜ = ( Σ Balanceᵢ × ReferencePriceUSDᵢ,ₜ ) ÷ NormalizationFactor. Σ Balanceᵢ × ReferencePriceUSDᵢ,ₜ is the Basket Value in USD. ReferencePriceUSD is the canonical USD market price of each constituent taken from the multi-source consensus layer — the word ‘reference’ describes how the constituent price is sourced, not the unit of the published output, which is US dollars.

The normalization factor is a fixed scaling constant for the basket version, used only to keep the quoted price in a readable range. For calculation version gci50-basket-v3-usd it is 1,000,000,000 and it is never solved, tuned, or reset to make the published price equal a round number.

Balance derivation at a rebalance: Balanceᵢ = ( TargetWeightᵢ × V ) ÷ RebalanceReferencePriceᵢ. At version inception V is the observed aggregate free-float market capitalisation of the selected constituents — an observed market quantity, not a chosen anchor. At every subsequent rebalance V is the basket value immediately before the rebalance, so the published USD price is continuous across a composition change and is never rebased. Extraordinary methodology events (for example a delisting or a supply event requiring intervention) are the only circumstances in which balances change outside a scheduled rebalance, and each such change is recorded.

The reference basket is a calculation construct only. GCI holds no assets and GCI 50 is not a fund, a net asset value or an investable product; the published USD price is a measurement of the basket, not a quote for anything purchasable. This calculation is published as calculation version gci50-basket-v3-usd. Values published under earlier semantics (gci50-basket-v2, and the retired solved-divisor v1) retain their original calculation version, are not restated, and are never compared against v3 prices.

An index committee owns the methodology and meets at least quarterly. Discretion is constrained: it may be exercised only where the rules are silent or where mechanical application would produce a materially unrepresentative outcome, and each exercise is recorded.

Every methodology change is versioned and published with an effective date, a summary of the change, the rationale and the expected impact on composition. This page reflects methodology version 0.2-draft-usd and calculation version gci50-basket-v3-usd. These versions describe the rules document and calculation semantics respectively.

Every published observation carries a data mode. Live indicates a price calculated from inputs captured contemporaneously by GCI at the stated observation time. Point-in-time backtest indicates a price calculated from historical inputs captured as-of the observation date and free of subsequent restatement. Reconstructed backtest indicates a price calculated from historical inputs rebuilt after the fact.

Historical benchmark results may incorporate reconstructed point-in-time market data where complete historical inputs are unavailable. GCI distinguishes reconstructed backtests from calculations produced using contemporaneously captured data. Reconstructed results are not equivalent to live results and are labelled as such in the interface and in the public data API.

All calculations are bounded to their observation time: a price dated at a given moment is computed only from inputs observable at or before that moment, so later restatements cannot influence an earlier published value. Inputs are normalised into a single canonical GCI data model before calculation, and each observation retains its source, observation time, retrieval time and quality state so any published price can be reconstructed and audited.

Inputs are screened before calculation and assigned a quality state of valid, warning, degraded or rejected. Rejected inputs are excluded from the calculation, and a blocking data-quality condition suspends publication rather than allowing a degraded price to be published.

Official prices are append-only. If an official value must be corrected, the replacement is published as a new record that supersedes the original; the original remains available with the correction reason and timestamp, so public history can never be silently rewritten.

While the methodology is in draft, GCI publishes a launch profile that states explicitly which rules above are enforced now and which are deferred. A deferred rule is disclosed here and in the public data API — it is never silently treated as satisfied. These notes are updated as coverage improves and will be versioned when the methodology is finalised.

Market observations: prelaunch GCI 50 calculations draw from configured exchange and oracle paths. Canonical reference prices apply source-family quorum, freshness, venue-anchor, outlier and divergence controls. Missing or non-compliant held-asset prices block publication rather than triggering a silent substitute.

Universe and supply metadata: market capitalisation, circulating supply and listing-age evidence are recorded with provider provenance. Slower-changing metadata may be carried forward only within its stated freshness policy, with its age and quality retained. Metadata aggregators are not the pricing source of record.

Listing age: verified first-listing dates are not yet available for every asset. The draft screen accepts provider-reported or provider-derived proxy dates, and the evidence class behind each constituent's listing date is recorded and exposed in the public constituent data. Assets with no listing-date evidence at all are ineligible.

Liquidity: the median traded value is measured over a trailing 30-day window built exclusively from venue-measured daily volumes captured contemporaneously by GCI — aggregator-reported volumes never enter the median. A median requires at least 30 real daily observations; assets whose history is shorter — or whose liquidity figures lack sample-count evidence — are ineligible rather than passed on incomplete data.

Venue coverage: venue counts are derived only from real, timestamped observations — never assumed or fabricated. Any minimum-venue requirement that is not yet enforced remains an explicit prelaunch limitation rather than being treated as satisfied synthetically.

Reconstitution governance: reconstitutions are proposed mechanically by the calculation system and take effect only after explicit approval by an authorised operator. Each decision is recorded with the approver's identity, timestamp and a written note.