PHBench predicts Series A via Product Hunt

PHBench exposes the "hype vs. substance" gap in startup launches, proving that community upvotes are a weak proxy for venture success compared to founder track records. Traditional gradient-boosted models like XGBoost consistently outperform state-of-the-art LLMs like Gemini 3.1 Pro on this sparse tabular dataset. Maker influence, defined by team size and community followers, is the single most predictive signal for future funding rounds. B2B sectors such as fintech and payments convert to Series A at three times the baseline rate, validating that "boring" software remains the most investable. While reaching Rank #1 provides a 2.2x lift in funding probability, the model primarily captures genuine economic cycles rather than just platform noise.