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

Trend prediction method based on attention mechanism and reinforcement learning

A technology of reinforcement learning and trend prediction, applied in the field of information science, can solve problems such as price evolution, and achieve the effect of reducing decision-making errors

Pending Publication Date: 2022-02-15
SUN YAT SEN UNIV
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

2. Prices evolve in a trending manner
However, the patent does not involve entering the market as much as possible when the futures price is low and has an upward trend, and reducing the probability of making a wrong decision

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
  • Trend prediction method based on attention mechanism and reinforcement learning
  • Trend prediction method based on attention mechanism and reinforcement learning
  • Trend prediction method based on attention mechanism and reinforcement learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0110] Such as figure 1 As shown, a trend prediction method based on attention mechanism and reinforcement learning includes the following steps:

[0111] S1: Select a futures product and obtain the historical futures market data of the selected product;

[0112] S2: Perform data cleaning on the data;

[0113] S3: Preprocess the data;

[0114] S4: Use pre-processed data to pre-train the feature extraction model;

[0115] S5: Splicing the output of the feature extraction model with the original data for reinforcement learning model training;

[0116] S6: Use the trained reinforcement learning model for decision-making, and derive the decision-making sequence;

[0117] S7: Test on the backtest platform.

[0118] In this example, first select the futures variety, try to choose a variety with a longer establishment time and a larger trading volume, and according to the fractal theory, judge whether the futures is suitable for data enhancement through testing, to a certain ext...

Embodiment 2

[0120] Such as figure 1 As shown, a trend prediction method based on attention mechanism and reinforcement learning includes the following steps:

[0121] S1: Select a futures product and obtain the historical futures market data of the selected product;

[0122] S2: Perform data cleaning on the data;

[0123] S3: Preprocess the data;

[0124] S4: Use pre-processed data to pre-train the feature extraction model;

[0125] S5: Splicing the output of the feature extraction model with the original data for reinforcement learning model training;

[0126] S6: Use the trained reinforcement learning model for decision-making, and derive the decision-making sequence;

[0127] S7: Test on the backtest platform.

[0128] The specific process of step S1 is:

[0129] S11: Due to the high noise of financial data, compared with other machine learning tasks, training trend prediction models often requires more data to achieve better results. The selected data is minute-level data. Accor...

Embodiment 3

[0140] Such as figure 1 As shown, the present invention provides a trend prediction method based on attention mechanism and reinforcement learning, comprising the following steps:

[0141] S1: Select the futures product according to the fractal theory, and obtain the historical futures market data of the selected product, including the following steps:

[0142] S11: Calculate skewness and kurtosis based on futures historical market data, and screen futures varieties. When 1

[0143] Skewness SKE, also known as the third standard central moment of the probability model, is usually used to describe the symmetry of the data, and the calculation formula is as follows:

[0144]

[0145] Kurtosis KUR, also known as the fourth standard central moment of the probability model, is usually used to describe the tail thickness of the data, and the calculat...

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 provides a trend prediction method based on an attention mechanism and reinforcement learning, and the method makes full use of the market information, better captures the market dynamics, enters as much as possible when the futures price is at a low level and has a rising trend, and reduces the probability of decision errors. Firstly, futures varieties are selected, varieties which are long in establishment time and large in trading volume are selected as far as possible, whether futures are suitable for data enhancement or not is judged through testing according to the fractal theory, and the problem that the data volume is small is solved to a certain extent. Then, through data cleaning and a feature extraction model based on an attention mechanism, the ability of the model to extract financial data features is optimized to serve as the input of a final strategy network, finally, a reinforcement learning algorithm is used, continuous action control is introduced, and an intelligent agent with strategy randomness and robustness is obtained; and the intelligent agent can make decisions according to market dynamics.

Description

technical field [0001] The present invention relates to the field of information science, and more specifically, relates to a trend prediction method based on attention mechanism and reinforcement learning. Background technique [0002] Traditional financial investment methods rely to a large extent on requiring investors to have relatively rich financial knowledge, relatively strong ability to resist pressure in the face of market downturns, and decisive decision-making ability. The market is in trouble. Because there are three basic principles in market analysis: 1. Price includes everything. 2. Prices evolve in a trending manner. 3. History repeats itself over and over again. So in recent years, financial innovation and artificial intelligence technology have been successfully combined in various financial applications, including quantitative trading (QT). Although a large amount of research has been dedicated to constructing better QT strategies using deep learning (...

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): G06Q40/06G06Q40/04G06N3/04G06N3/08
CPCG06Q40/06G06Q40/04G06N3/08G06N3/048G06N3/045
Inventor 潘炎杜劭旭印鉴
Owner SUN YAT SEN UNIV
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