I've been working on a side project that uses an LLM pipeline to track how financial narratives evolve across Financial news and social media discussion. Not just how much people are talking about something, but how they feel about it, broken down into ten distinct emotions, tracked over time.

The attached heatmap shows the emotion z-scores for the geopolitical_risk narrative over the past month (April 2026). Each row is an emotion, each column is a day, and the color intensity shows how far above or below normal that emotion is running.

How to read it

Green for negative emotions (panic, fear, frustration, skepticism, uncertainty) = those emotions are below their baseline, ie people are calmer than usual

Red for negative emotions = those emotions are elevated, people are more afraid/frustrated than normal

The scale flips for positive emotions (optimism, confidence, hope, greed, euphoria) --> red means elevated positive sentiment, green means it's suppressed

Z-scores range from roughly -3 to +2.5, so…