Example of Physics-Based Sentiment Forecasts
To generate these forecasts, we used six of our 30+ physics-based measures of solar system orbital geometry. These include clusters of planets, along with the angles and distances among them. Each of the six drivers was selected because it showed strong statistical significance in explaining the actual 14-week Relative Strength Index (RSI) between 1940 and the late 2000s.
We then built a multi-driver model calibrated to explain the variability of the actual 14-week RSI over that historical period. To test it, we held the model constant and extended it forward from 2000 through 2024. The only inputs that changed were the locations of the planets each week (and therefore the orbital geometry), using NASA data. This gave us a physics-based forecast of the 14-week RSI for more than 20 years without readjusting the model’s parameters.
The chart above shows three things: the Predicted 14-week RSI (dotted green), the Actual 14-week RSI (solid green), and the U.S. stock market price on a log scale (brown). Notice how the peaks and troughs of the Predicted and Actual RSI generally line up. The figure highlights these inflection points so the alignment is clear.
Our statistical testing shows how powerful this relationship is. We found that 88% of the short-term turning points in the actual RSI occurred within plus or minus one week of the predicted turning points. By chance, you’d expect only about 0.25% alignment. The likelihood of achieving this 88% match rate by chance is extremely low (p < 0.001).
The difference between the actual and predicted RSI is also important. When the two diverge, it means economics, fundamentals, or major events are driving sentiment beyond the natural cycle. For example, during the Global Financial Crisis (2007–2009), the gap between the actual and predicted RSI widened sharply. Once we adjust for the naturally occurring shifts in sentiment, the impact of that economic stress becomes much clearer.