Stock trend prediction method based on 2-dimension flow prediction

Abstract

The invention relates to a stock trend prediction method based on 2-dimension flow prediction. The stock trend prediction method is characterized in that historical data is extracted, prediction is carried out in the horizontal and vertical time points, and the predicted results are combined to obtain the final prediction data. The stock trend prediction method is advantageous in that the one-step lagging problem in 1-dimension prediction can be effectively solved.

Description

technical field [0001] The invention relates to a method for predicting stock trends. Background technique [0002] The unstable and random nature of the stock market makes it a challenge to predict only tomorrow's stock prices. Trends in the stock market can be better estimated with an excellent, well-constructed feature set. Furthermore, when we build the right models to capture the less observable properties of changing trends, our predictive power improves. Recently, in many cities, especially in some big cities, the number of people investing and buying stocks has been increasing. Therefore, stock trend forecasting technology is particularly important. [0003] The current popular approach to stock forecasting is to use a binary event model. Based on this model, a feature set is built to better predict the future trend of the stock market. For example, Bayesian and support vector machines are used for prediction, which has high prediction accuracy and speed. Anothe...

AI Technical Summary

Problems solved by technology

[0004] The common problem with the above methods is that they all belong to the 1-dimensional forecasting algorithm, and the main disadvantage is that there will be a one-step lag problem, that is, the forecast after the time point when the stock data stream changes drastically will cause a large error

Most of the current forecasting algorithms have such problems, especially when the stock data flow is complex or changes drastically

Images