Methodology

1. Channels. We start from a curated list of finance YouTubers seeded by a public analysis video. An admin can add more channels and adjust their reliability weight (1–100).

2. Transcripts. We pull each new video and fetch its captions. When captions aren't available, we fall back to AI transcription of the audio.

3. Mentions. Google Gemini reads the transcript and extracts every ticker mention, with a bullish / neutral / bearish stance and a confidence score.

4. Scoring. For each ticker over a 14-day rolling window we sum weight × stance × confidence, normalize to 0–100, and apply a small bonus when 3+ distinct channels mention it.

5. Review. An admin reviews drafts and writes the "why to buy" reasoning before publishing.

Scores below 40 are never published.