Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Investment portfolio generation system and method based on K-line graph and convolution auto-encoder

A technology of convolutional auto-encoder and generation system, which is applied in the field of portfolio generation system based on K-line graph and convolutional auto-encoder. and other issues to achieve the effect of increasing revenue and increasing revenue

Pending Publication Date: 2022-06-03
SHANGHAI UNIVERSITY OF FINANCE AND ECONOMICS
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Linear correlation measures cannot capture the nonlinear dynamic characteristics of stocks and most correlation measures cannot take into account the time-invariant factors that are more important for stock correlations

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Investment portfolio generation system and method based on K-line graph and convolution auto-encoder
  • Investment portfolio generation system and method based on K-line graph and convolution auto-encoder
  • Investment portfolio generation system and method based on K-line graph and convolution auto-encoder

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0029] This embodiment provides a portfolio generation system based on K-line graph and convolutional autoencoder, including:

[0030] K-line chart generation module: used to obtain the stock market data of each stock in the window period closest to the current according to the set window period (20 days in this embodiment) and generate the K-line chart of each stock. Specifically, in this embodiment, the K-line chart is generated by using the daily lowest price, highest price, opening price, and closing price of each stock in the stock market in the last 20 days;

[0031] Feature extraction module: used to input the K-line chart of each stock into the convolutional autoencoder to obtain the deep feature representation of each K-line chart. In this embodiment, the convolutional autoencoder is constructed based on the structure of VGG16. VGG16 uses a multi-layer stacked 3*3 convolution kernel to replace the larger convolution kernel in AlexNet. Its advantage lies in the multi-l...

Embodiment 2

[0046] This embodiment provides a method for using the system described in Embodiment 1, such as figure 1 As shown, the specific process is:

[0047] S1. The K-line graph generation module obtains the stock market data of each stock in the nearest window period according to the set window period (20 days in this embodiment) and generates the K-line graph of each stock.

[0048] S2. The feature extraction module inputs the K-line chart of each stock into the convolutional autoencoder to obtain the deep feature representation of each K-line chart.

[0049] S3. The similarity calculation module regards each stock as a node, and calculates the cosine similarity between stocks according to the depth feature representation of the K-line graph of each stock obtained by the feature extraction module. The formula is as follows:

[0050]

[0051] where X i with X j are the eigenvector representations of two stocks i and j, respectively;

[0052] According to the cosine similarity...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses an investment portfolio generation system and an investment portfolio generation method based on a K-line graph and a convolution auto-encoder, deep features of stocks are extracted by using the K-line graph reflecting historical performance of the stocks and the convolution auto-encoder, and the stocks are clustered by using a modularity optimization thought. The method comprises the following steps: firstly, generating a K-line graph reflecting stock performance based on an opening price, a closing price, a highest price and a lowest price of stock history, then inputting the generated K-line graph into a convolutional auto-encoder to extract depth feature representation of stocks, and calculating cosine similarity among the stocks based on the depth feature representation of each stock to construct a stock network; and then the stocks are clustered by using a modularity optimization method, and finally the stocks are selected from each cluster according to the Sharp ratio to construct an investment portfolio. According to the invention, a computer vision technology is combined, and a combinatorial investment strategy capable of effectively increasing income is provided from the perspective of network science.

Description

technical field [0001] The invention relates to the application field of deep learning in finance, in particular to an investment portfolio generation system and method based on a K-line graph and a convolutional autoencoder. Background technique [0002] Investment decision-making is an important research field in quantitative finance and behavioral finance, among which portfolio construction and optimization is one of the most important decision-making problems in this field. Financial institutions try to structure and optimize investment portfolios to maximize investor returns while minimizing risk to investors. [0003] At present, many techniques consider the similarity between stocks when constructing investment portfolio strategies, but most similarity measures use linear indicators such as Pearson correlation coefficient for measurement, which cannot capture the nonlinear dynamic characteristics of stocks. In addition, most of the deep learning-based portfolio strat...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06Q40/06G06Q40/04G06F17/18G06K9/62G06N3/04G06N3/08
CPCG06Q40/06G06Q40/04G06F17/18G06N3/08G06N3/045G06F18/23213
Inventor 刘建国胡国圣郭强欧阳谢斐杨凯
Owner SHANGHAI UNIVERSITY OF FINANCE AND ECONOMICS
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com