Hi HN, I dropped out of college 3 months ago to solve my own problem: I kept building SaaS products that nobody wanted. I realized I was guessing instead of measuring. So I built a system to automate "Market Signal Detection" using LLMs. The Architecture: 1. Ingests data primarily from Reddit (niche communities), Hacker News, and rising Search Queries. 2. Filters for "Negative Sentiment" + "Financial Pain" (e.g., people complaining about losing money). 3. Calculates a "Velocity Score" to spot trends before they hit mainstream Google Trends. I wrote a detailed technical breakdown of the methodology and the "3 Signals" I look for (Bleeding Neck, Fragmentation, Wallet Index). You can read the full guide (no paywall) here: https://blog.vect.pro/trend-hacking-guide The tool itself is part of the platform I'm building, but the guide stands alone as a framework for finding ideas manually if you prefer. Would love feedback on the signal detection logic. - Afraz