CN111489259A - Stock market risk prediction intelligent implementation method based on deep learning - Google Patents
Stock market risk prediction intelligent implementation method based on deep learning Download PDFInfo
- Publication number
- CN111489259A CN111489259A CN202010298282.3A CN202010298282A CN111489259A CN 111489259 A CN111489259 A CN 111489259A CN 202010298282 A CN202010298282 A CN 202010298282A CN 111489259 A CN111489259 A CN 111489259A
- Authority
- CN
- China
- Prior art keywords
- data
- stock
- price
- deep learning
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013135 deep learning Methods 0.000 title claims abstract description 20
- 230000006870 function Effects 0.000 claims abstract description 9
- 238000013528 artificial neural network Methods 0.000 claims abstract description 6
- 230000004913 activation Effects 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 18
- 230000000630 rising effect Effects 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 3
- 230000007774 longterm Effects 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims 1
- 150000001875 compounds Chemical group 0.000 abstract 1
- 238000012706 support-vector machine Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000013178 mathematical model Methods 0.000 description 3
- 230000006403 short-term memory Effects 0.000 description 3
- 230000002354 daily effect Effects 0.000 description 2
- 238000007418 data mining Methods 0.000 description 2
- 230000003203 everyday effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000003340 combinatorial analysis Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Development Economics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Operations Research (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Technology Law (AREA)
- Educational Administration (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an intelligent realization method for stock market risk prediction based on deep learning, which forms more abstract high-feature representation by combining shallow features, thereby discovering a deep implicit relationship of data, and can better obtain a compound function by stacking a plurality of layers of neural networks and selecting a Sigmoid activation function.
Description
Technical Field
The invention relates to an intelligent stock market risk prediction realization method based on deep learning, and belongs to the technical field of communication.
Background
The stock market plays an important role in the financial market, and whether to accurately predict the fluctuation of the future whole stock market is also a key point for obtaining income in the stock market. Investors in recent years have tried to choose to predict stocks using expert analysis, combinatorial analysis, strategic research, and other related methods. Most of the methods adopt related contents of probability and statistics, and simultaneously use various mathematical models to finally obtain a prediction result. However, the trend of stock prices is influenced by various factors such as national policies, national economic conditions, international environments and enterprise management conditions, so that a single mathematical model cannot accurately describe the future trend of the stock market. Meanwhile, because the stock investment is accompanied by high risk and the stabilization of the income cannot be realized only by subjective experience of people and a mathematical model, the rising and falling characteristics of the stock market need to be mined according to historical data of a large number of stock trades, a certain reference is provided for stock prediction in a short period in the future, and the method has great significance for improving the income rate of stocks and is also an improvement part of the technology.
With the development and application of artificial intelligence methods in recent years, more artificial intelligence technologies are applied to stock market forecasting, and some Machine learning-based methods and data mining-based methods, such as Neural Network (NN) and Support Vector Machine (SVM) technologies, are more common, and are commonly applied to stock market forecasting. However, the SVM technique is difficult to implement for large-scale training samples, and the data in the stock market is large and various, and when the data size becomes large, the training time of the SVM becomes long, thereby bringing more time-consuming cost to the prediction. In addition, the classical support vector machine algorithm only gives out an algorithm of a two-class classification method, and in the practical application of stock data mining, multiple classes of problems are generally solved. In addition, for the SVM technology, the method is sensitive to missing data and to selection of parameters and kernel parameters, and the selection of the parameters is generally artificial and has certain randomness and unrepresentability, so that domain knowledge should be introduced in the stock market field, but no good method is available for solving the selection problem of kernel functions at present.
Disclosure of Invention
The invention aims to provide an intelligent realization method for stock market risk prediction based on deep learning, aiming at the defects of the prior art, the method forms more abstract high-feature representation by combining shallow features, thereby discovering the implicit relationship of data deep level, and through the stacking of multilayer neural networks and the selection of Sigmoid activation functions:
where f (x) represents the output layer, the output range is (0, 1), e represents a natural constant in mathematics, x represents the input value, and the function is to transform real values within one (— infinity, + ∞) to the interval [0,1 ].
The technical scheme adopted by the invention for solving the technical problems is as follows: the invention provides an intelligent realization method for stock market risk prediction based on deep learning, which comprises the following steps:
step 1: the historical trading data of the stock market is subjected to data preprocessing to obtain daily opening price, closing price, highest price, lowest price, bargaining price and bargaining volume of the stock, namely information data influencing the stock price, so that a long-term data sequence is obtained. And the data is normalized, and the influence of larger data difference is reduced.
And 2, selecting historical trading data of the stock in the past 10 years, and training the data by applying an L STM (L ong Short-Term Memory) model in deep learning to obtain a training set, wherein the training set mainly comprises the fluctuation condition of the past stock data.
And step 3: selecting the latest real data as a test set, predicting the stock closing price situation 15 days later by using a trained training model according to the transaction data 40 days before a certain day, comparing the closing price with the price on the day, making a data rule, regarding 2% of the price exceeding the price as rising and 2% of the price lower than the date as falling, and thus obtaining the situation of predicting the rising and falling of the stock in the future.
And 4, step 4: and adjusting different reasonable parameters, and respectively changing the number of the past test data, the future prediction days and the rise-fall evaluation value so as to obtain various different data prediction results.
And 5: the multiple data results are sorted and integrated by a time series method, so that the result can be accurately predicted every day in the future. And different fluctuation results caused by different parameters adopt a few principle of obeying most, and the obtained prediction precision is ensured to be higher.
Step 6: after the final prediction result is obtained, according to the rise and fall prediction conditions in the future period, when the rise and fall conditions in the future period are the same, buying or selling can be selected at the first rise point or fall point; when two successive changes and falls are opposite in a future period, buying (or selling) is carried out at the change and fall point in time.
The invention carries out risk prediction on the whole stock market based on a deep learning method, carries out prediction on the whole market of the financial stock market, and particularly adopts the combination of the deep learning and the financial stock, carries out self-learning self-adaptation through an L STM model in the deep learning, and obtains more accurate prediction on the whole stock market in the future according to the training of past historical data.
Has the advantages that:
1. the method carries out self-learning self-adaptation through a long-Short Term Memory network model (L ong Short-Term Memory, L STM) in deep learning, and well obtains accurate prediction of the future whole stock market according to training of past historical data.
2. According to the invention, a more abstract high-feature representation is formed by combining shallow features to discover a deep implicit relation of data, and a composite function can be better obtained by stacking a plurality of layers of neural networks and selecting a reasonable activation function.
3. The method has good technical effects on multiple indexes such as RMSE, error value, self-designed profit value and the like.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
The invention is described in further detail below with reference to the drawings.
As shown in FIG. 1, the invention provides an intelligent method for predicting risk of stock market based on deep learning, which comprises the following steps:
step 1: carrying out data preprocessing on historical trading data of stocks: the stock history data is obtained by first obtaining daily opening price, closing price, highest price, lowest price, bargaining price and bargaining volume (the bargaining volume is the volume ratio of the current day to the last day), namely information data influencing the stock price, thereby obtaining a string of long-term data sequences. And then defining interval data, and unitizing the data according to the price difference range of the highest price and the lowest price, thereby obtaining each parameter data capable of being used for training and reducing the influence of larger data difference.
Step 2, selecting historical trading data of the stock in the past 10 years, applying a L STM (L ong Short-Term Memory) model in deep learning, setting a time span of 40 days after removing holidays and rest days of stock trading, training the data by parameters of 2% of fluctuation amplitude and 3000 times of iteration after predicting the time span of 10 days, and obtaining a training model by a deep learning self-training method, wherein the training set takes a closing price as a reference and roughly comprises the fluctuation condition of the past stock data.
And step 3: selecting the latest real data as a test set, predicting the stock closing price situation 15 days later by using a trained training model according to the transaction data 40 days before a certain day, comparing the closing price with the price on the day, making a data rule, regarding 2% of the price exceeding the price as rising and 2% of the price lower than the date as falling, and thus obtaining the situation of predicting the rising and falling of the stock in the future.
And 4, step 4: adjusting different reasonable parameters: the following parameters (1) the number of past test data (selected to be 40.50.60 days in the past), (2) the number of future prediction days (selected to be 10.15.20 days in the future), and (3) the rise and fall evaluation values (with 0.5%, 1%, and 2% as standards) are respectively changed, so that various data prediction results are obtained.
And 5: and displaying various data results in a two-dimensional coordinate chart form, and superposing and integrating different graphs obtained by different parameters by a time sequence method to ensure that the results can be accurately predicted every day in the future. And different fluctuation results caused by different parameters adopt a few principle of obeying most, and the obtained prediction precision is ensured to be higher.
Step 6: according to the different prediction result graphs under different parameters, overlapping to obtain a final prediction result graph, and according to the rise and fall prediction conditions in the future period in the graph, when the rise and fall conditions in a period of time in the future are the same, buying (or selling) can be selected at the first rise (or fall) point; when two successive changes and falls are opposite in a future period, buying (or selling) is carried out at the change and fall point in time. After screening and listing each transaction time point obtained under different fluctuation conditions, the investment selection time for obtaining more stable income in the future stock market can be obtained.
For the stock market, a series of contents such as market price fluctuation, market rule change, enterprise decision and the like can change future trends in real time, so that the problems cannot be solved by only using the traditional methods of mathematical modeling and operation research model statistical analysis, and for a simple artificial intelligence method, L STM model can process and train a large amount of data, and meanwhile, the use of the STM model can be greatly improved in accuracy compared with the traditional SVM and BP model, and the regression prediction method has a plurality of profit values in the Root mean square Error (RMsquared), the network model (Back prediction, BP) compared with the traditional linear regression algorithm (L initial regression), and the like.
Claims (3)
1. A stock market risk prediction intelligent implementation method based on deep learning is characterized by comprising the following steps:
step 1: carrying out data preprocessing on historical trading data of the stock to obtain daily opening price, closing price, highest price, lowest price, bargaining price and bargaining volume of the stock, namely information data influencing the stock price, thereby obtaining a string of long-term data sequences;
selecting historical trading data of the stock in the past 10 years, and training the data by applying an L STM model in deep learning to obtain a training set, wherein the training set comprises the fluctuation condition of the stock data in the past;
and step 3: selecting the latest real data as a test set, predicting the stock closing price situation 15 days later by using a trained training model according to the transaction data 40 days before a certain day, comparing the closing price with the price on the day, making a data rule, regarding 2% of the price exceeding the price as rising and 2% of the price lower than the date as falling, and thus obtaining a situation of predicting the rising and falling of the stock in the future;
and 4, step 4: adjusting different reasonable parameters, and respectively changing the number of past test data, the future prediction days and the rise-fall evaluation value so as to obtain various different data prediction results;
and 5: the data results are sorted and integrated by a time series method, so that the result can be accurately predicted for each day in the future, wherein different fluctuation results caused by different parameters are ensured;
step 6: after the final prediction result is obtained, according to the rise and fall prediction conditions in the future period, when the rise and fall conditions in the future period are the same, buying or selling is selected at the first rise point or fall point; when two successive changes and falls are opposite in a future period, buying or selling is carried out at the change and fall point.
2. The method for intelligently predicting risk in stock market based on deep learning as claimed in claim 1, wherein the method combines features of shallow layer to form moreAdding abstract high-feature representation to discover data deep-level implicit relations, and through stacking of multilayer neural networks and selecting activation functions:
where f (x) denotes the output layer, the output range is (0, 1), e denotes the natural constants in mathematics, x denotes the input values, and the function is the transformation of real values within one (- ∞, + ∞) into the interval [0,1 ∞ ]]。
3. The intelligent realization method for risk prediction of stock market based on deep learning as claimed in claim 1, wherein the method is used for predicting the whole market of financial stock market, and the self-learning self-adaptation is carried out by L STM model in deep learning by combining deep learning and financial stock.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010298282.3A CN111489259A (en) | 2020-04-16 | 2020-04-16 | Stock market risk prediction intelligent implementation method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010298282.3A CN111489259A (en) | 2020-04-16 | 2020-04-16 | Stock market risk prediction intelligent implementation method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111489259A true CN111489259A (en) | 2020-08-04 |
Family
ID=71812867
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010298282.3A Pending CN111489259A (en) | 2020-04-16 | 2020-04-16 | Stock market risk prediction intelligent implementation method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111489259A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111932379A (en) * | 2020-09-23 | 2020-11-13 | 北京口袋财富信息科技有限公司 | Data processing method and device, electronic equipment and readable storage medium |
| CN112884211A (en) * | 2021-02-02 | 2021-06-01 | 上海卡方信息科技有限公司 | Stock price prediction system and method based on deep learning |
| CN113159941A (en) * | 2021-02-02 | 2021-07-23 | 上海卡方信息科技有限公司 | Intelligent streaming transaction execution method and device |
| WO2023185125A1 (en) * | 2022-04-02 | 2023-10-05 | 富途网络科技(深圳)有限公司 | Product resource data processing method and apparatus, electronic device and storage medium |
-
2020
- 2020-04-16 CN CN202010298282.3A patent/CN111489259A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111932379A (en) * | 2020-09-23 | 2020-11-13 | 北京口袋财富信息科技有限公司 | Data processing method and device, electronic equipment and readable storage medium |
| CN112884211A (en) * | 2021-02-02 | 2021-06-01 | 上海卡方信息科技有限公司 | Stock price prediction system and method based on deep learning |
| CN113159941A (en) * | 2021-02-02 | 2021-07-23 | 上海卡方信息科技有限公司 | Intelligent streaming transaction execution method and device |
| WO2023185125A1 (en) * | 2022-04-02 | 2023-10-05 | 富途网络科技(深圳)有限公司 | Product resource data processing method and apparatus, electronic device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111489259A (en) | Stock market risk prediction intelligent implementation method based on deep learning | |
| CN110147389A (en) | Account number treating method and apparatus, storage medium and electronic device | |
| Ren | Research on financial investment decision based on artificial intelligence algorithm | |
| CN115660725A (en) | Method for depicting multi-dimensional energy user portrait | |
| CN106097094A (en) | A kind of man-computer cooperation credit evaluation new model towards medium-sized and small enterprises | |
| Sajid et al. | An ensemble LGBM (light gradient boosting machine) approach for crude oil price prediction | |
| Liu et al. | Stock price trend prediction model based on deep residual network and stock price graph | |
| CN114004530B (en) | Enterprise electric power credit modeling method and system based on ordering support vector machine | |
| Gao et al. | Establishment of Economic Forecasting Model of High‐tech Industry Based on Genetic Optimization Neural Network | |
| CN111815458A (en) | Dynamic investment portfolio configuration method based on fine-grained quantitative marking and integration method | |
| Borysenko et al. | Intelligent forecasting in multi-criteria decision-making. | |
| Akter et al. | Early Prediction of Cryptocurrency Price Decline: A Deep Learning Approach | |
| Sun et al. | Short-term stock price forecasting based on an SVD-LSTM model | |
| Purwinarko et al. | Crude oil price prediction using Artificial Neural Network-Backpropagation (ANN-BP) and Particle Swarm Optimization (PSO) methods | |
| Zha et al. | A hierarchical reinforcement learning framework for stock selection and portfolio | |
| Malik et al. | Towards a Stock Price Prediction on Time Series Data using Long-Short Term Memory Method | |
| CN112132352A (en) | Stock price trend prediction method based on attention and dense connection | |
| CN112949883A (en) | Thermal load prediction method and device | |
| Huang et al. | Crude oil price forecasting base on an EEMD and multi-scale time series analysis combination methods | |
| Lin et al. | Feature transformation and simulation of short term price variability in reinforcement learning for portfolio management | |
| Mitra et al. | Predicting Indian basket crude prices through machine learning models-a comparative approach | |
| Gao et al. | Enterprise financial forecasting based on machine deep learning | |
| Li et al. | Prediction and Sensitivity Analysis of Companies’ Return on Equity Based on Deep Neural Network | |
| Varadharajan et al. | Stock Closing Price and Trend Prediction with LSTM-RNN | |
| Supriyanto | Comparison of Grid Search and Evolutionary Parameter Optimization with Neural Networks on JCI Stock Price Movements during the Covid 19 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200804 |