A geometrical shape in a time series can be a variation of the variable over a finite duration like a "S" or a "W" or other arbitrary variation. The data shape can

represent an event or can predict one in the evolution of process described by time series.