Performance
Price appreciation, depreciation, and relative performance across benchmark indices.
Our Approach
Rare Index provides institutional-grade market intelligence for the alternative asset market, treating Cards, Watches, Wine, Video Games, Toys, Coins, Jewelry, and Other as benchmarkable markets.
Our indices track price movements, relative performance, and market signals using a rigorous, transparent methodology built on the same principles that underpin traditional financial benchmarks.
We employ a matched-basket, chain-linked return methodology — each index movement is computed only from assets actually priced on both adjacent observation dates, then chain-linked into a continuous series. This approach prioritizes consistency, comparability, and resistance to manipulation: an index can only move when real, matched observations move it.
This page outlines the key components of our methodology. We publish it because transparency is not optional when institutional capital is at stake.
What We Measure
Rare Index quantifies performance, risk, and market signals across curated asset universes.
Market Breadth
Breadth of participation across categories to assess market health.
Momentum
Signal momentum computed from stored price history — statistically unusual behavior, never advice.
Coverage & Depth
Matched basket size, constituent count, and coverage published with every index observation.
Data Reliability
Confidence scoring on every published measurement, from constituent depth to basis integrity.
✓ We Measure
- ✓ Verified pricing from qualified, licensed sources
- ✓ Public auction results and marketplace data
- ✓ Third-party graded assets and condition tiers
- ✓ Observable market activity with strictly positive prices
× We Do Not Measure
- × Private or unreported transactions
- × Subjective condition adjustments without grading
- × Hypothetical, modeled, or estimated pricing
- × Interpolated or forward-filled values — gaps stay honest
Index Construction
Rare Index uses a matched-basket, chain-linked methodology to ensure stable, comparable, and manipulation-resistant benchmarks.
How It Works
01
Matched Basket
Each index link is computed only from assets with a valid representative price on both adjacent observation dates. Assets phase in and out under governed identity rules; historical baskets are never rewritten.
02
Equal-Weight Returns
Each link averages the price relatives of the matched basket with equal weighting — no fixed base-period weights, no survivorship distortion.
03
Chain-Linking
Links compound into a continuous index level. Rare Index derived indices are normalized to a base value of 100 at their first sufficiently-covered observation; a reading of 112 means the basket has appreciated 12% from that base.
04
Quality Gates
A return publishes only when the matched basket meets minimum depth thresholds. When overlap is insufficient, the index holds its prior level, the return is withheld, and the observation is flagged — never estimated.
ⓘ
Partner mirror indices (Liv-ex wine, WatchCharts watches) are published at the partner's native levels and cadence — never recomputed, rebased, or converted. Returns exist only between actual partner-published points.
Cadence & Honest Gaps
Rare Index facts are observation-dated. Values publish when real observations support them — never interpolated, never forward-filled.
Ingestion — Daily snapshots from licensed and qualified sources.
Computation — Index, confidence, and signal computation runs under governed publication discipline; every published value traces to an immutable governed publication.
Partner cadence — WatchCharts data follows a daily cadence; Liv-ex wine indices publish monthly — a 28–31 day gap between wine observations is on schedule, not missing data.
Where a value cannot be computed from real observations, the platform shows an honest em-dash — honest gaps are a feature of the methodology, not a defect.
Confidence Scoring
Every published measurement carries a Confidence profile scoring the trustworthiness of the measurement itself — a 0–100 composite with LOW, MEDIUM, and HIGH bands, built from equally weighted dimensions including constituent depth, coverage completeness, basis integrity, and constituent-level confidence. Confidence scores data quality; it never predicts markets.
Signal Methodology
Rare Index signals detect statistically unusual behavior relative to an asset's own governed history — momentum, volatility anomalies, and observation activity — measured in observations, not calendar days. Signals describe what the data is doing; they never estimate economic value and never constitute advice. Confidence may flag a signal's reliability but never suppresses it.
Data Governance
Every price is attributed to its source. Every index is built on published methodology under versioned governance — methodology changes are governed events, never silent edits. Publications are immutable: once published, a served dataset is never rewritten. The eligibility gate is universal: active constituents, unquarantined observations, index-eligible sources, and strictly positive prices.
Raw source data is preserved untouched; all normalization, currency conversion, and computation happens in governed compute layers with full audit trails.
1M+ITEMS TRACKED
330K+AUCTION RESULTS (99.8% IMAGED)
100M+PRICE OBSERVATIONS
17+VERIFIED DATA SOURCES