Orbis Ranking

Overview

Every collection on Orbis receives a composite score from 0 to 100, computed from 10 independent on-chain and marketplace signals. The score reflects a collection's overall health, activity, and market position relative to its peers.

Rankings are recalculated periodically and update automatically. The algorithm uses percentile-based normalization — each signal is compared against all other collections, not measured in absolute terms. This keeps rankings fair regardless of collection size or age.

The 10 Signals

Each signal captures a different dimension of collection health. Weights are tuned using community survey ratings, so they sit close together — no single signal dominates. Weights may be adjusted over time and additional metrics added as the algorithm evolves.

How Scores Are Calculated

1. Signal Extraction

Raw values for all 10 signals are gathered from on-chain data, marketplace activity, and staking integrations. Each collection's data is evaluated independently.

2. Percentile Normalization

Raw values are converted to percentile scores (0–100) by ranking each collection against all others. A collection with the highest volume doesn't get "100" — instead, its score reflects its percentile position relative to the full marketplace.

This approach ensures that one dominant collection doesn't skew the entire ranking. Ties are handled by averaging their positions. Listing Health is ranked only against collections of a similar size, and Floor Stability and Wash Trade Health use their raw 0–100 value instead of a percentile.

3. Missing Data

If a signal isn't available for a collection (e.g., fewer than 3 days of price history), it receives a neutral score of 50. Missing data neither helps nor hurts a collection's ranking.

4. Weighted Composite

The composite is a weighted average of all 10 normalized signals. Each signal's contribution is proportional to its weight.

5. Wash Trading Penalty

Collections with detected wash trading lose up to 20 points from the composite, in proportion to how much of their trading is wash, and their volume and sales velocity are discounted before ranking — so faked activity can't buy a higher spot. The result is rounded to produce the final 0–100 score displayed on the marketplace.

6. Eligibility

Collections that have never traded, or launched in the last 24 hours, aren't scored yet. They show as unranked until they have a real market.

Always Improving

The Orbis Ranking algorithm is not static. It's designed to evolve alongside the marketplace.

• Signal weights may be adjusted as we learn which factors best reflect collection quality.
• New signals may be added as the ecosystem grows (e.g., social engagement, cross-marketplace data).
• The normalization approach ensures rankings remain fair even as the number of collections changes.
• Small collections and large collections are evaluated on equal footing through percentile-based scoring.

Our goal is to surface genuinely healthy, active collections — not just the largest or most hyped. The ranking rewards real utility, community engagement, and sustained market activity.