In October 2007, three months before the financial crisis began, public sentiment on housing was overwhelmingly positive. The data said everything was fine. The numbers were solid. But if you were listening to the qualitative signals — the conversations happening between mortgage brokers, the tone shifts in real estate forums, the subtle changes in how banks talked about risk — you could hear the foundation cracking.
What Sentiment Analysis Actually Is
When most people hear “sentiment analysis,” they think of algorithms scanning Twitter for positive and negative words. That’s not what we do. Quantitative sentiment analysis counts words. Qualitative sentiment analysis reads between the lines.
The difference matters. An algorithm can tell you that 60% of tweets about AI are positive. It can’t tell you that the positive tweets are getting shorter and more formulaic while the negative ones are getting longer and more specific. That shift in texture is where the real signal lives.
The Three Corrections
Our methodology has been backtested against the three most significant market corrections of the last two decades. In each case, the qualitative sentiment indicators we track showed a detectable shift 3-6 months before the quantitative indicators caught up.
The pattern is always the same: the narrative remains positive, the data remains strong, but the quality and conviction of the positive sentiment degrades. People keep saying the right things, but they say them differently. That difference is measurable if you know what to look for.
What the Current Data Shows
Right now, our 14-indicator model is showing a pattern consistent with the pre-correction phases of all three historical events. The details are in our methodology white paper, which is available for free to subscribers. But the headline is clear: the sentiment foundation of several major investment narratives is weaker than the surface suggests.
