We show that economic theory predicts a proportionality tendency between GDP and the stock market, and that it implies that stocks have rational deviations away from this equilibrium, but that these are subsequently reversed. We corroborate these findings empirically, including an integral-correction mechanism in the econometric model: deviations from long-run proportionality accumulate and create a pullback mechanism that generates cycles of random lengths. The resulting model proves to have stable coefficient estimates (instead of sequence of breaks up & down trying to mimic a cycle), detects overreaction, predicts turning points, and has an impressive fit & predictive performance for stock returns: an in-sample R² of 61.4%, and an out-of-sample R² of 72.8% for 2013-2022.