A0754
Title: Phase spline-analysis for time series dynamics
Authors: Lyudmila Gadasina - Saint-Petersburg State University (Russia) [presenting]
Lyudmila Vyunenko - Saint-Petersburg State University (Russia)
Ivan Labutkin - Saint-Petersburg State University (Russia)
Abstract: The analysis of the medium- and long-term time series dynamics was carried out using the phase shadow concept. The approach includes the following steps: determining a time interval, considered as a minimum unit within which the dynamics of the series can be interpreted as short-term volatility; time series smoothing by one-dimensional adaptive regression splines with the number of splines corresponding to the selected time intervals; constructing the phase shadow treated as the projection of the first derivative of the resulting smooth function $y$ onto the plane $(y,y')$. Phase spline analysis allows one to identify and visualize cycles, bubbles, and structural shifts for a time series of economic data. The approach was tested on the NASDAQ index, and Bitcoin prices were taken over the period from November 2019 to January 2023, with an interval of 24 hours.
