Is My Collection Playable?

Not every NFT collection makes a good deduction game. Here's how to evaluate yours.

Quick Checklist

• At least 6 meaningful trait categories (Background, Body, Eyes, etc.)
• No single trait value dominates > 80% of the collection
• Most trait combinations are unique across the collection
• Traits are relatively independent (Body type shouldn't predict Background color)

The CQS Score

Run the analysis yourself:

# Clone the repo
git clone https://github.com/NormieLabs/NormaLabs-HQ

# Install dependencies
pip install numpy matplotlib

# Run analysis on your collection metadata JSON
python3 obsidian_vault/guessmynft/research/guessability/analyze_collection.py \
  your-collection.json \
  --output my-collection

Your metadata JSON should have this format:

{
  "name": "My Collection",
  "characters": [
    {
      "id": 1,
      "name": "NFT #1",
      "attributes": [
        {"trait_type": "Background", "value": "Red"},
        {"trait_type": "Body", "value": "Dragon"}
      ]
    }
  ]
}

Interpreting Results

CQS	Verdict
≥ 0.85	✅ Excellent — ideal for all modes including wagers
0.70 – 0.84	✅ Good — suitable, minor imbalances
0.55 – 0.69	⚠️ Fair — playable but some NFTs have significant advantage
< 0.55	❌ Poor — not recommended

Common Anti-Patterns

The "None" problem: If 95% of your collection has "No Accessory" for the Accessories trait, that category provides almost zero information. Questions about it are wasted turns.

Correlated traits: If every Dragon body always has a Fire background, knowing the body type reveals the background for free. This reduces effective trait categories.

Duplicate combinations: If many NFTs share identical trait vectors, they become indistinguishable by questions alone — requiring random guessing at the end.