Indications are presented for a significant connection between the relative motion of the planets and the appearance of energetic solar flares. Based on the records of the last four decades, the analysis highlights remarkable features and a lack of randomness in the data. The indications are supported further by the predictive power of a preliminary application to forecasting with machine learning methods.

A forecast for the evolution of solar cycle 25 in terms of solar flares activity is presented. The forecast is derived from an existing phenomenological model based on the coupling of an internal solar component and a planetary component. In addition to sufficient temporal resolution, the predictions are characterized by features which both differentiate the model from other methods for space climate prediction and make it falsifiable.

A phenomenological model is presented for the quantitative description of individual solar cycles' features, such as onset, intensity, evolution, in terms of the number of M and X-class solar flares. The main elements of the model are the relative ecliptic motion of the planets Jupiter and Saturn, and its synergy with a quasi-periodic component of solar activity. Using as input the temporal distribution of flares during cycle 21, the general evolution of cycles 22-24 is reproduced in notable agreement with the observations, including the resurgence of activity in the last months of 2017, and further predictions are provided for cycle 25. This deterministic description could contribute to elucidating the responsible physical mechanisms and forecasting space weather.