This free PoC uses keyword rules, not an AI analyst. It can miss sarcasm, context and the difference between network adoption and token value.
How scores are calculated
Each item is Bullish if only positive terms match, Bearish if only negative terms match, Mixed if both match, or Neutral if neither matches. Simple negation is supported.
Overall score = 100 × (Bullish count − Bearish count) / all items. Above +15 is Bullish; below −15 is Bearish; otherwise Mixed, or Neutral if no directional terms were found. With 1–4 items, “Limited sample” shows the available distribution without an overall verdict. With zero items, “No data collected” is shown. These are sample-size indicators, not calibrated confidence levels.
News analysis uses headlines only. Reddit posts use titles and feed text; comments use comment text only. Repeated text and very similar news headlines within 72 hours are deduplicated. Promotional keywords are filtered; some spam and paraphrases can remain. Each item has equal weight, so active threads can dominate. No engagement weighting.