Resource Notes
Use, understand, and apply the model
Key takeaway
Three rarity methodologies: Trait Count Score (simplest — counts traits), Inverse Frequency Score (most common — 1/trait_frequency), and Statistical Rarity (multiply trait frequencies). For a 10,000 PFP collection: Common 50% (20-50 score), Uncommon 30% (50-100), Rare 15% (100-200), Epic 4% (200-500), Legendary 1% (500+). Rarity matters most in first 30 days post-reveal.
The Three Rarity Score Methodologies
Method 1: Trait Count Score (Simplest)
Score = number of traits the NFT has. NFTs with fewer traits are rarer. Problem: Does not account for trait rarity within each category.
Method 2: Inverse Frequency Score (Most Common)
For each trait the NFT has: add `1 / trait_frequency` to the rarity score. Example: Background Blue (30%) → 3.33; Eyes Laser (2%) → 50.00; Hat Crown (5%) → 20.00; Mouth Smile (25%) → 4.00; Total score = 77.33. Higher score = rarer. This is the standard used by Rarity.tools.
Method 3: Statistical Rarity (Most Sophisticated)
Multiply trait frequencies: P(NFT) = 0.30 × 0.02 × 0.05 × 0.25 = 0.0000075 (0.00075%). Lower probability = rarer. Does not handle 'none' traits well but is statistically rigorous.
Rarity Distribution for a 10,000 PFP Collection
Reference Table
| Rarity Tier | Count | Score Range | % of Collection |
|---|---|---|---|
| Common | 5,000 | 20–50 | 50% |
| Uncommon | 3,000 | 50–100 | 30% |
| Rare | 1,500 | 100–200 | 15% |
| Epic | 400 | 200–500 | 4% |
| Legendary | 100 | 500+ | 1% |
Collector Guidance: When Rarity Matters
For trading: Rarity score matters most in the first 30 days after reveal, when the market is discovering which tokens are rare. Rare items often trade at 5–20× floor price immediately post-reveal. For long-term holding: Fun > rarity for most successful PFP communities. For creators: Release rarity methodology before mint (not just after reveal).