Stock index extreme value prediction method based on top view algorithm

Abstract

The invention relates to a stock index extreme value prediction method based on a top view algorithm. The method comprises an extreme value definition module, a visual graph network construction module, a completely unviewable graph network construction module, a weight-considering visual graph network construction module, a weight-considering unviewable graph network construction module, an extreme value prediction index construction module and an evaluation index construction module. According to the method, the defect that the time sequence is mapped into the complex network by using the visual graph algorithm is improved, the extreme value in the financial time sequence is predicted by using the improved method, and the prediction effect of the improved method is better. According to the method, more detail information is reserved in the process of mapping a time sequence to a complex network through a weighted visual graph algorithm based on an overlook angle; through a visual graph algorithm based on neighbor communities, the influence of adjacent nodes on the observation nodes is weakened.

Description

technical field [0001] The invention belongs to the field of computer technology, and relates to analysis and prediction of financial time series, and is especially applicable to the problem of extreme value prediction of stocks and stock indexes, and specifically relates to a method for predicting extreme values ​​of stock indexes based on a bird's-eye view visual graph algorithm. Background technique [0002] Nonlinear time series analysis is an active area of ​​research. Studying the structure of complex signals can yield information about the processes that generate these sequences, helping us to understand, model, and predict sequences. Over the past few years, scholars have proposed several methods for mapping time series to complex networks. Among these methods, the visual graph algorithm has received more attention due to its low complexity and good geometric properties. Applying the methods of complex network theory to characterize time series makes complex network...

AI Technical Summary

Problems solved by technology

The results show that: the horizontal visual graph network cannot effectively identify various time series signals (periodic, fractal, chaotic); for fractal signals, the visual graph and the finite traversal visual graph network have scale-free power-law forms, but the anti-noise ability is poor ; For periodic signals and chaotic signals, the finite traversal visual graph network has stronger noise immunity than the visual graph

Images