Testing for nonlinearity in time series: the method of surrogate data
Conference proceedings on Interpretation of time series from nonlinear mechanical systems
| Hi-index | 0.98 |
We revise the method of shuffled surrogate data for financial time series. We take into account calendar effects such as the day-of-the-week and the holiday effect. More precisely, we shuffle the data that belong to a particular calendar event separately from the other data. Our modification keeps these calendar effects unchanged for the shuffled surrogate time series. As a consequence, the shuffled surrogate data preserve not only the distribution of the original data. but also the distribution of the calendar effects. Empirical results are presented based on the returns of the Nasdaq Composite index, and the General Index of the Athens Stock Exchange. Significant differences are observed between the ordinary and the proposed method.