Answers are grounded in this paper and cite their sources.
In stock market, numerous ways have been used to predict the future asset return. Traditional time series models and neural network models are both popular in practice. Many of them have been shown that they do not make full use of the property of long-term dependency while with the help of graph structure, one can turn time series into complex network and keep the long-term dependency property. To make use of the property of the long memory of financial market in prediction, our framework first applies the visibility method to turn stock price data into graph structure, and then applies graph neural network to make graph classification task to let the network learn the overall topological structure of the graphs. The result shows that with the use of such property, our model can successfully forecast the future stock trend.
No takes yet. Share an insight, caveat, or question.
Chen et al. (2023) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: