Stationarity

Very popular concept in time series.

The formal definition: A stationary time series is one whose properties do not depend on the time at which the series is observed.

Stationary data is one that has no obvious upward or downward trend and no seasonality. Also, the variability around a local mean does not change throughout time.

The opposite can be said about stock price, for example, during the top of the bull market, when higher prices have higher variations than during normal times.

Historically, people first developed methods to work with stationary data. Then they developed methods to transform non-stationary data into stationary.

But they also noticed you loose a lot of information in the process. So the next step in evolution was to develop methods that deal directly with non-stationary data (e.g. State-Space Models or integrated ARIMA)

The take home message: checking for stationarity is not always necessary. It depends on the model.