If $R^2$ is high, the model is good at both predicting future trends and explaining past patterns

Explaining past patterns is about fit ($R^2$), while predicting future trends requires that the underlying relationship remains stable and hasn't changed

A high correlation coefficient guarantees accurate predictions of future values

Prediction and explanation are the same thing in regression analysis