# Extracting regimes from a time series

Trying to find a good way to solve the following problem: I have daily time series $$\{x_n\},$$ say for the last 10 years. During that time span there are several shorter periods (say, Dec '12 - April '14, May '16 - Oct '16, Nov '18 - Jan '19), that I am interested in and that I believe can be identified using my time series. In the ideal case, it would be something simple, like $$\{x_n\}$$ is decreasing exactly during those periods, but it doesn't look that easy. What would be an appropriate statistical technique to identify the properties of my time series, that cut out given periods in the best way?