PHBenchSpot tomorrow's unicorns before investors do

PHBench

PHBench is an open benchmark that predicts Series A funding outcomes from Product Hunt launch signals. Built on 67,292 launches spanning seven years, it trains and ranks machine learning models to identify which products have the highest probability of securing venture capital within 18 months.

Product Highlights

• Open Benchmark: Transparent, reproducible framework with documented features and manually audited labels for research integrity
• Massive Dataset: Trained on 67,292 Product Hunt launches from 2019–2025, including 528 verified Series A winners
• Proven Predictive Power: Best model achieves 4.7× lift over random baseline, turning a 0.78% base rate into actionable signals
• Live Scoring: Every new Product Hunt launch receives a real-time prediction score from top-performing ensemble models
• Research-Grade Methodology: Developed by Vela Partners and University of Oxford with citable, peer-reviewed methodology

Use Cases

• Venture Capital Screening: Identify promising startups early by filtering thousands of launches to the top-ranked prospects
• Startup Benchmarking: Compare your Product Hunt performance against historical Series A winners using quantified signals
• Academic Research: Access a clean, documented dataset for studying startup success factors and fundraising predictors
• Model Development: Submit and test your own ML models against a held-out test set with standardized evaluation metrics

Target Audience

PHBench serves venture capitalists, angel investors, startup founders, and machine learning researchers who need data-driven methods to predict early-stage startup success from public launch signals.