Analysis of Stock Price Data: Determinition of The Optimal Sliding-Window Length

• Main Authors: Xuebin Liu, Xuesong Yuan, Chang Liu, Hao Ma, Chongyang Lian
• Format: Article
• Language: English
• Published: Frontiers Media S.A. 2021-09-01
• Series: Frontiers in Physics
• Subjects: time series visualization; complex network; sliding window-based visibility graph; multiplex visibility graph; stock price
• Online Access: https://www.frontiersin.org/articles/10.3389/fphy.2021.741106/full

Over the recent years, the study of time series visualization has attracted great interests. Numerous scholars spare their great efforts to analyze the time series using complex network technology with the intention to carry out information mining. While Visibility Graph and corresponding spin-off technologies are widely adopted. In this paper, we try to apply a couple of models derived from basic Visibility Graph to construct complex networks on one-dimension or multi-dimension stock price time series. As indicated by the results of intensive simulation, we can predict the optimum window length for certain time series for the network construction. This optimum window length is long enough to the majority of stock price SVG whose data length is 1-year. The optimum length is 70% of the length of stock price data series.