<p>Spectrum of solar activity oscillations contains apart from the well-known 11-year activity cycle a continuous component, which includes, in particular, quasy-biennual oscillations as well as long-term oscillations including so-called Gleisberg cycle.&#160; We suggest to consider the mid-term solar variability in terms of statistical dynamic of fully turbulent systems, where solid arguments are required to accept an isolated dominant frequency in a continuous (smooth) spectrum. What about the timescales longer than the Schwabe cycle, we consider them as a presence of long-term memory in solar dynamo and discuss statistical test for veryication of this interpretation. Sequences for statistical long-term forecast of solar activity are discussed.</p>