HumbleWorth Domain Price Prediction

An AI-powered domain name valuation model that provides instant price estimates across three sales channels: auction, marketplace, and brokerage. This model powers the free HumbleWorth domain valuation service.

Model Description

This neural network model combines domain embeddings with word embeddings to predict domain name values. The architecture uses:

• Sentence Transformers (all-MiniLM-L12-v2) for text encoding
• Domain Tokenization with intelligent word segmentation
• Multi-layer Perceptron prediction head with dropout regularization
• Probability Curve Fitting using tanh functions to generate price distributions

The model was trained on multiple large-scale datasets including: - DNPric.es historical auction data (3M+ transactions over 20+ years) - Dropped and bargain domain datasets

Intended Use

Designed for: - Domain investors evaluating portfolio value - Quick price estimates for domain acquisitions - Bulk domain valuation (up to 2560 domains per request) - Market research and competitive analysis - Educational purposes understanding domain valuation

NOT designed for: - Financial advice or investment recommendations - Legal valuation for official purposes (court cases, tax assessments) - High-frequency trading or arbitrage strategies - Trademark or intellectual property valuations

Usage

Input

• domains: Comma-separated list of domain names (max 2560)
• Example: "example.com,test.net,domain.org"

Output

Returns structured JSON with valuations for each domain across three sales channels:

{
  "valuations": [
    {
      "domain": "example.com",
      "auction": 1500.0,
      "marketplace": 5000.0,
      "brokerage": 8000.0,
      "error": null
    }
  ]
}

Sales Channel Explanations

• Auction: Estimated value in competitive auction environments (50th percentile)
• Marketplace: Estimated value for direct marketplace sales (97.5th percentile)
• Brokerage: Estimated value through premium brokerage services (99.25th percentile)

API Compatibility

This model provides enhanced functionality compared to the original HumbleWorth API: - Same input/output format - Increased batch limit (2560 domains vs 20 domains) - Optimized batched processing for large requests - Drop-in replacement for https://valuation.humbleworth.com/api/valuation

Usage Tips

Best Practices: - Submit domains in lowercase for consistency - Batch multiple domains (up to 2560) for maximum efficiency - Include TLD extensions (.com, .net, etc.) - Avoid special characters or internationalized domains

Input Quality: - Valid, registered domain formats work best - Recently dropped domains may have less reliable estimates - Brandable domains typically get higher valuations than generic keywords - Shorter domains generally receive higher estimates

Processing Large Portfolios: - Single requests can handle up to 2560 domains efficiently - Internal batching (128 domains per batch) optimizes memory usage - Large requests may take longer but avoid rate limiting issues

Ethical Considerations

Model Limitations: - Estimates are computer-generated and should not replace human expertise - Market conditions change rapidly; historical data may not reflect current trends - Model may exhibit bias toward English-language domains - Valuations can vary significantly from actual sale prices

Responsible Use: - Always disclose that estimates are AI-generated when sharing with others - Consider multiple valuation sources for important decisions - Understand that domain values are highly subjective and market-dependent - Use estimates as starting points for further research, not final assessments

Transparency: - Model predictions are based on historical sales data through early 2024 - Training data may not represent all market segments equally - Performance may vary for newer TLDs or emerging market trends

Caveats and Recommendations

Technical Considerations: - Model optimized for CPU inference on small hardware (2GB RAM) - Response times typically under 2 seconds for single domains - Batch processing is more efficient than individual requests

Market Context: - Domain values are highly volatile and context-dependent - Economic conditions significantly impact domain market prices - Trademark issues and legal considerations are not evaluated - Seasonal trends and market cycles are not explicitly modeled

Data Freshness: - Training data extends through 2024 but may not capture very recent market shifts - Emerging trends in domain markets may not be fully represented - Consider supplementing with current market research for major decisions

Examples

Single Domain:

Input: "example.com"
Output: {"valuations": [{"domain": "example.com", "auction": 1200, "marketplace": 3500, "brokerage": 5200}]}

Multiple Domains:

Input: "tech.com,startup.net,innovation.org"
Output: {"valuations": [
  {"domain": "tech.com", "auction": 15000, "marketplace": 45000, "brokerage": 75000},
  {"domain": "startup.net", "auction": 800, "marketplace": 2400, "brokerage": 4000},
  {"domain": "innovation.org", "auction": 600, "marketplace": 1800, "brokerage": 3000}
]}

This model is part of the HumbleWorth ecosystem, providing free AI-powered domain valuations. With support for up to 2,560 domains per request, this Replicate model exceeds the bulk processing capability of humbleworth.com while offering programmatic access.