CN112418575A - Futures initiative contract quantitative timing decision system based on cloud computing and artificial intelligence deep learning algorithm - Google Patents

Futures initiative contract quantitative timing decision system based on cloud computing and artificial intelligence deep learning algorithm Download PDF

Info

Publication number
CN112418575A
CN112418575A CN201910775950.4A CN201910775950A CN112418575A CN 112418575 A CN112418575 A CN 112418575A CN 201910775950 A CN201910775950 A CN 201910775950A CN 112418575 A CN112418575 A CN 112418575A
Authority
CN
China
Prior art keywords
algorithm
data
futures
deep learning
cloud computing
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
Application number
CN201910775950.4A
Other languages
Chinese (zh)
Inventor
刘畅
刘喆雄
潘龙佳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Liu Chang International Co Ltd
Original Assignee
Liu Chang International Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Liu Chang International Co Ltd filed Critical Liu Chang International Co Ltd
Priority to CN201910775950.4A priority Critical patent/CN112418575A/en
Publication of CN112418575A publication Critical patent/CN112418575A/en
Pending legal-status Critical Current

Links

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
    • 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
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • 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
    • 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
    • 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
    • G06Q40/06—Asset management; Financial planning or analysis

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Technology Law (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)

Abstract

The invention mainly belongs to the field of futures quantitative timing decision, and particularly relates to a futures initiative contract quantitative timing decision system based on cloud computing and an artificial intelligent deep learning algorithm. The method provided by the invention is used for constructing a quantitative transaction strategy by using a neural network algorithm and a deep learning algorithm by using a cloud computing and big data technology. The quantitative investment strategy of the invention can establish an effective investment model through computer programming, thereby realizing quantitative management of investment. The inventive content includes a quantitative investment strategy, a financial database, a backtesting framework and a trading system. The specific implementation mode comprises a data module, an analysis module, an algorithm generation module, a retest module and a transaction module. The invention solves the problem of excessive fitting of the AI algorithm, only millisecond-level calculation cost is needed for executing the algorithm, the application value of the algorithm is greatly improved, and enumeration of the whole parameter space is possible by optimizing a deep learning mode.

Description

Futures initiative contract quantitative timing decision system based on cloud computing and artificial intelligence deep learning algorithm
Technical Field
The invention mainly belongs to the field of futures quantitative timing decision, and particularly relates to a futures initiative contract quantitative timing decision system based on cloud computing and an artificial intelligent deep learning algorithm.
Background
Quantitative trading refers to the investment of securities traded by computer technology with modern statistical and mathematical methods. Quantitative trading selects various high-probability events which can bring excess income from huge historical data to formulate strategies, verifies and solidifies the rules and strategies by using a quantitative model, and then strictly executes the solidified strategies to guide investment so as to obtain the sustainable, stable and excess return higher than average income. Quantitative trading originated in the stock market in the seventies of the last century, and then rapidly developed and popularized, especially in the futures trading market, programmed to become mainstream. There are data showing that futures programmed trading in mature markets abroad already accounts for 70% -80% of the total volume of trading, while it is just starting at home. The disadvantage of emotional fluctuation of traders in manual trading becomes more and more a barrier to profit, and the natural accuracy and 100% execution rate of programmed trading bring advantages to the profit.
The research scholars respectively put forward a large number of different theoretical analysis research methods aiming at various observable obvious characteristics of the price fluctuation rule of the futures, but because the stock exchange market has a set of complicated and unknown operation mechanism, the factors influencing the price fluctuation rule of the stock exchange and the investment earning rate of the stock investors in the operation mechanism are various, and a large number of influencing factors interact together, so that the whole exchange system becomes a complex nonlinear learning system. For the learning of a nonlinear learning system, the artificial neural network algorithm can obtain more accurate prediction precision than other prediction methods. By carefully analyzing historical information of futures and selecting a proper artificial intelligence algorithm (such as an artificial neural network algorithm), a futures trading price model with high prediction precision and strong real-time performance is constructed, and the method has very important theoretical and application values on the stability and continuous development of financial markets and the improvement of the investment profitability of vast investors.
The invention utilizes cloud computing and big data technology, uses neural network algorithm and deep learning algorithm to carry out quantitative analysis on futures initiative contracts, constructs a futures initiative contract quantitative timing decision-making system based on cloud computing and artificial intelligent deep learning algorithm, and realizes excess income of futures investment.
Disclosure of Invention
The invention relates to a quantitative investment strategy, which can establish an effective investment model through computer programming, thereby realizing quantitative management of investment. The specific inventive content comprises a quantitative investment strategy, a financial database, a backtesting framework and a trading system.
The invention is realized by the following technical scheme:
the quantitative investment dependence statistical and metering method of the invention establishes a proper strategy and obtains investment profit by a computer automatic (semi-automatic) transaction means, and is a product combining statistics, computers and financial disciplines. The quantitative timing decision system of the invention utilizes a statistical model to select proper time to buy and sell specific futures to obtain benefits.
Furthermore, the invention firstly constructs a financial database at the cloud, the financial database is formed by combining financial theory knowledge and computer application software technology and carrying out operation, processing and processing on financial and other related data, the financial database is a 'data platform' capable of providing data and related services, and the database is mainly used for storing original data, storing the data after the original data is cleaned or transformed, and storing transaction and result data generated by backtesting.
The retest frame body of the invention consists of three modules: the strategy module generates a trading signal according to the parameters and rules of the strategy, the trading module determines the position of each stage according to the signal and records the expense and fund condition generated in the trading process, and the retest expression module evaluates the strategy and outputs related indexes, graphs and test reports.
The quantitative transaction system of the invention comprises three models: an alpha model, a risk control model, and a transaction cost model. The investment portfolio building model is a trading strategy formed by comprehensively using an alpha model, a risk control model and a trading cost model. And finally, the investment portfolio construction model completes tasks by executing the model. Wherein the alpha model is used to predict the future trends of the financial product being traded. The risk control model includes loss prevention and fund management to reduce loss and manage funds. The transaction cost model decomposes a large transaction into a plurality of small transactions through mathematical calculation, thereby achieving the purpose of reducing the commission charge.
Furthermore, the invention mainly realizes the landing of the quantitative timing decision of the futures initiative contract through a data module, an analysis module, an algorithm generation module, a retest module and a transaction module.
The data module of the invention adopts big data technology, the needed basic data is cleaned, extracted and counted by features, and is collected into the database, the data collection program needs to set three parameters, and the initial date and the ending date of the market and the transaction data of the related varieties are obtained. The data cleaning firstly needs to detect the distortion value of the original futures contract, and fills the corresponding distortion value with the price data of the previous 1 minute, if the price data of the previous 1 minute is also distorted, the data is delayed forward, and so on, and the data of other dimensions (such as the volume of finished goods) is not modified.
The analysis module of the invention constructs a 'commodity futures price index' to be used as a trading signal of a certain commodity futures, and calculates the commodity futures index in an equity mode.
The algorithm generation module provided by the invention mainly applies an artificial intelligence deep learning algorithm and a cloud computing technology, adopts an AI learning algorithm, and carries out parameter cross validation and algorithm calling on algorithm parameters through cloud computing. The method comprises the following steps:
step 1, initialization, setting the state of the node to be represented, representing the node
Figure 565677DEST_PATH_IMAGE001
Connected nodes
Figure 342134DEST_PATH_IMAGE002
Randomly initializing the state of each nodeAs a weight matrix
Figure 687796DEST_PATH_IMAGE003
。
Step 2, randomly selecting a training sample to be input into the network, and updating the state of each node in the first hidden layer
Figure 214592DEST_PATH_IMAGE004
。
Step 3, the hidden node state is calculated according to the step 2
Figure 442660DEST_PATH_IMAGE004
Updating the status of a visual node
Figure 763920DEST_PATH_IMAGE005
：
Figure 800140DEST_PATH_IMAGE006
。
Step 4, calculating to obtain the state of the visible node according to the step 3
Figure 130627DEST_PATH_IMAGE005
The hidden layer state is updated again, which is recorded as
Figure 201483DEST_PATH_IMAGE007
：
Figure 896906DEST_PATH_IMAGE008
。
Step 5, randomly selecting the next training sample, turning to step 2, if the samples of the training set in the current round are all input, respectively calculating the change amount of the weight according to the formula
Figure 482740DEST_PATH_IMAGE009
Amount of change of noise control parameter
Figure 554601DEST_PATH_IMAGE010
Updating the weight matrix and the noise control number:
Figure 260389DEST_PATH_IMAGE011
Figure 408604DEST_PATH_IMAGE012
。
step 6 goes to step 2, and the next round of training is entered until a predetermined number of times is reached or the change in the weight matrix is sufficiently small (i.e., the weight matrix is changed
Figure 199843DEST_PATH_IMAGE013
) When this is the case, the first training is finished.
And 7, taking the output obtained by the first CRBM as the input of a second CRBM, repeating the steps 1 to 6, and training the two CRBMs until all CRBMs forming the DBN are trained, and finishing the training of the DBN.
Finally, the transaction module of the present invention is calculated based on the algorithm and the model
Figure 826127DEST_PATH_IMAGE014
The value is compared with the true price, if the true price is
Figure 855263DEST_PATH_IMAGE014
If the value is within the range, sending an instruction at the cloud end according to the transaction module to perform transaction; if the true price is
Figure 705539DEST_PATH_IMAGE014
If the value is out of the range, sending an instruction at the cloud end according to the transaction module to monitor the real price in real time until the real price reaches the target value
Figure 452915DEST_PATH_IMAGE014
Within the value range, a transaction is conducted.
According to the invention, through effective integration of the contents of the quantitative investment strategy, the financial database, the retest frame and the transaction system, a futures initiative contract quantitative timing decision-making system based on cloud computing and an artificial intelligent deep learning algorithm is developed, and excess income of the conventional futures investment is realized. The invention has the beneficial effects that:
1. the algorithm is innovative. The problem of overfitting of the AI algorithm is solved, the effect outside the sample on most varieties is good, and the cross-variety effect of the algorithm is relatively stable;
2. innovation in development and application. In the learning stage, parameters can be learned on a common server generally for two days, and on a real disk, only millisecond calculation cost is needed for executing an algorithm, so that the application value of the algorithm is greatly improved;
3. innovation in algorithm development. The general algorithm can enumerate parameters rarely, because the calculation cost is too high, the algorithm optimizes a deep learning mode to enable enumeration in the whole parameter space to be possible, and meanwhile, the learning mode of the algorithm also ensures that the extrapolation of the sample space is controlled.
Drawings
FIG. 1 is a decision framework diagram of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting.
As shown in the figure, the invention adopts 5 modules to realize the landing of the quantitative investment strategy. The system comprises a data module, an analysis module, an algorithm generation module, a retest module and a transaction module, and is completed through MATLAB programming.
The data module of the invention adopts big data technology to clean, extract and count the characteristics of the required off-line financial data and real-time financial data, and collects the data into the database, and the data collection program needs to set three parameters to obtain the initial date and the ending date of the market and the transaction data of related varieties. The method comprises the steps of firstly carrying out wavelet denoising processing on directly acquired data, and then dividing the data into a training data set and a prediction data set. The data cleaning firstly needs to detect the distortion value of the original futures contract, and fills the corresponding distortion value with the price data of the previous 1 minute, if the price data of the previous 1 minute is also distorted, the data is delayed forward, and so on, and the data of other dimensions (such as the volume of finished goods) is not modified.
The analysis module of the invention constructs a 'commodity futures price index' as a trading signal of a certain commodity futures, and calculates the commodity futures index in an equity mode:
Figure 297691DEST_PATH_IMAGE016
the exponential rate of return equals: the average value of the rate of return of all tradeable varieties in a certain trading period. The exponential price is equal to the exponential price at the previous time multiplied by the exponential rate of return at that time. When the quantitative investment strategy is evaluated, the method is mainly evaluated through comprehensive indexes of investment income and risk. The method mainly comprises three indexes: annual income, maximum withdrawal, sharp rate.
The algorithm generation module provided by the invention mainly applies an artificial intelligence deep learning algorithm and a cloud computing technology, adopts an AI learning algorithm, and carries out parameter cross validation and algorithm calling on algorithm parameters through cloud computing. The method comprises the following steps:
step 1 initialization, setting node status
Figure 36977DEST_PATH_IMAGE017
It is shown that,
Figure 553540DEST_PATH_IMAGE004
representation and node
Figure 521627DEST_PATH_IMAGE001
Connected nodes
Figure 807246DEST_PATH_IMAGE002
State of (2), randomInitializing the state of each node as a weight matrix
Figure 248592DEST_PATH_IMAGE003
。
Step 2, randomly selecting a training sample to be input into the network, and updating the state of each node in the first hidden layer
Figure 721293DEST_PATH_IMAGE004
。
Step 3, the hidden node state is calculated according to the step 2
Figure 555387DEST_PATH_IMAGE004
Updating the status of a visual node
Figure 554568DEST_PATH_IMAGE005
：
Figure 651968DEST_PATH_IMAGE006
。
Step 4, calculating to obtain the state of the visible node according to the step 3
Figure 459300DEST_PATH_IMAGE005
The hidden layer state is updated again, which is recorded as
Figure 362665DEST_PATH_IMAGE007
：
Figure 622877DEST_PATH_IMAGE008
。
Step 5, randomly selecting the next training sample, turning to step 2, if the samples of the training set in the current round are all input, respectively calculating the change amount of the weight according to the formula
Figure 953495DEST_PATH_IMAGE009
Amount of change of noise control parameter
Figure 446793DEST_PATH_IMAGE010
Updating the weight matrix and the noise control number:
Figure 91532DEST_PATH_IMAGE011
Figure 471829DEST_PATH_IMAGE012
。
step 6 goes to step 2, and the next round of training is entered until a predetermined number of times is reached or the change in the weight matrix is sufficiently small (i.e., the weight matrix is changed
Figure 691458DEST_PATH_IMAGE013
) When this is the case, the first training is finished.
And 7, taking the output obtained by the first CRBM as the input of a second CRBM, repeating the steps 1 to 6, and training the two CRBMs until all CRBMs forming the DBN are trained, and finishing the training of the DBN.
An optimal network structure of the DBN is determined in an experimental mode, and the optimal network structure is divided into contents in 3 aspects: (1) determining the number of nodes of an input layer; (2) determining the number of nodes of each hidden layer; (3) the number of hidden layer nodes is determined, no specific rule can be followed, an approximate range is generally obtained according to experience, and then the number of hidden layer nodes is selected through multiple trial calculations. The following are several empirical formulas for calculating the number of hidden layer nodes:
Figure 891626DEST_PATH_IMAGE018
wherein the content of the first and second substances,
Figure 792586DEST_PATH_IMAGE019
in order to hide the number of layer nodes,
Figure 27389DEST_PATH_IMAGE020
in order to input the number of nodes,
Figure 168652DEST_PATH_IMAGE021
in order to output the number of nodes,
Figure 308646DEST_PATH_IMAGE022
is composed of
Figure 295188DEST_PATH_IMAGE023
Constant in between.
The retest module comprises a retest platform and an online display, the retest is carried out at the cloud end by using an artificial intelligence algorithm, the effect of the strategy is checked by using historical data, and opening and leveling signals generated by the strategy in the operation of a real disk are simulated, so that a income curve representing historical performance is obtained as a result. By analyzing and improving the result, the modification direction of the strategy is determined, and even whether the real disk is used or not is determined. All the yields are calculated in the form of logarithmic yields, and the yields in a plurality of stages are calculated as follows:
Figure 102607DEST_PATH_IMAGE024
。
finally, the transaction module of the present invention is calculated based on the algorithm and the model
Figure 149191DEST_PATH_IMAGE014
The value is compared with the true price, if the true price is
Figure 776482DEST_PATH_IMAGE014
In the value range, sending an instruction at the cloud end according to the trading module, and trading through a market trading API; if the true price is
Figure 284824DEST_PATH_IMAGE014
If the value is out of the range, sending an instruction at the cloud end according to the transaction module to monitor the real price in real time until the real price reaches the target value
Figure 228640DEST_PATH_IMAGE014
Within the value range, a transaction is conducted. The output module outputs the transaction after the transaction is executedAnd the input module acquires information from a remote transaction platform again, and the transaction is made to be in the decision system at all times in a circulating manner.

Claims (3)

1. A futures initiative contract quantitative timing decision system based on cloud computing and an artificial intelligence deep learning algorithm is characterized in that a data module adopts a big data technology, required basic data are cleaned, extracted and subjected to feature statistics and collected into a database, a data collection program needs to set three parameters, and the initial date and the final date of a market and transaction data of related varieties are obtained. The data cleaning firstly needs to detect the distortion value of the original futures contract, and fills the corresponding distortion value with the price data of the previous 1 minute, if the price data of the previous 1 minute is also distorted, the data is delayed forward, and so on, and the data of other dimensions (such as the volume of finished goods) is not modified.
2. A futures initiative contract quantitative timing decision system based on cloud computing and an artificial intelligence deep learning algorithm is characterized in that an algorithm generation module mainly applies the artificial intelligence deep learning algorithm and a cloud computing technology, adopts an AI learning algorithm, and carries out parameter cross validation and algorithm calling on algorithm parameters through the cloud computing. The method comprises the following steps:
step 1, initialization, setting the state of the node to be represented, representing the node
Figure RE-151088DEST_PATH_IMAGE001
Connected nodes
Figure RE-811746DEST_PATH_IMAGE002
Randomly initializing the state of each node as a weight matrix
Figure RE-718522DEST_PATH_IMAGE003
；
Step 2, randomly selecting a training sample to be input into the network, and updating the state of each node in the first hidden layer
Figure RE-607980DEST_PATH_IMAGE004
；
Step 3, the hidden node state is calculated according to the step 2
Figure RE-549392DEST_PATH_IMAGE004
Updating the status of a visual node
Figure RE-928420DEST_PATH_IMAGE005
：
Figure RE-276487DEST_PATH_IMAGE006
；
Step 4, calculating to obtain the state of the visible node according to the step 3
Figure RE-766375DEST_PATH_IMAGE005
The hidden layer state is updated again, which is recorded as
Figure RE-499975DEST_PATH_IMAGE007
：
Figure RE-49905DEST_PATH_IMAGE008
；
Step 5, randomly selecting the next training sample, turning to step 2, if the samples of the training set in the current round are all input, respectively calculating the change amount of the weight according to the formula
Figure RE-383804DEST_PATH_IMAGE009
Amount of change of noise control parameter
Figure RE-677382DEST_PATH_IMAGE010
Updating the weight matrix and the noise control number:
Figure RE-327806DEST_PATH_IMAGE011
Figure RE-986320DEST_PATH_IMAGE012
；
step 6 is transferred to step 2, and the next round of training is entered until the preset number of times is reached or the change of the weight matrix is small enough (i.e. the weight matrix is changed
Figure RE-114506DEST_PATH_IMAGE013
) When so, the first training is finished;
and 7, taking the output obtained by the first CRBM as the input of a second CRBM, repeating the steps 1 to 6, and training the two CRBMs until all CRBMs forming the DBN are trained, and finishing the training of the DBN.
3. A futures initiative contract quantitative timing decision system based on cloud computing and artificial intelligence deep learning algorithm is characterized in that a transaction module calculates according to algorithm and model
Figure RE-149458DEST_PATH_IMAGE014
The value is compared with the true price, if the true price is
Figure RE-919968DEST_PATH_IMAGE014
If the value is within the range, sending an instruction at the cloud end according to the transaction module to perform transaction; if the true price is
Figure RE-483805DEST_PATH_IMAGE014
If the value is out of the range, sending an instruction at the cloud end according to the transaction module to monitor the real price in real time until the real price reaches the target value
Figure RE-854612DEST_PATH_IMAGE014
Within the value range, a transaction is conducted.
CN201910775950.4A 2019-08-22 2019-08-22 Futures initiative contract quantitative timing decision system based on cloud computing and artificial intelligence deep learning algorithm Pending CN112418575A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910775950.4A CN112418575A (en) 2019-08-22 2019-08-22 Futures initiative contract quantitative timing decision system based on cloud computing and artificial intelligence deep learning algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910775950.4A CN112418575A (en) 2019-08-22 2019-08-22 Futures initiative contract quantitative timing decision system based on cloud computing and artificial intelligence deep learning algorithm

Publications (1)

Publication Number Publication Date
CN112418575A true CN112418575A (en) 2021-02-26

Family

ID=74779549

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910775950.4A Pending CN112418575A (en) 2019-08-22 2019-08-22 Futures initiative contract quantitative timing decision system based on cloud computing and artificial intelligence deep learning algorithm

Country Status (1)

Country Link
CN (1) CN112418575A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113312334A (en) * 2021-05-28 2021-08-27 海南超船电子商务有限公司 Modeling analysis method and system for big data of shipping user
CN114092254A (en) * 2021-11-26 2022-02-25 桂林电子科技大学 Consumption financial transaction method based on artificial intelligence and transaction system thereof

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113312334A (en) * 2021-05-28 2021-08-27 海南超船电子商务有限公司 Modeling analysis method and system for big data of shipping user
CN113312334B (en) * 2021-05-28 2023-05-26 海南超船电子商务有限公司 Modeling analysis method and system for big data of shipping user
CN114092254A (en) * 2021-11-26 2022-02-25 桂林电子科技大学 Consumption financial transaction method based on artificial intelligence and transaction system thereof

Similar Documents

Publication Publication Date Title
Li et al. Risk prediction in financial management of listed companies based on optimized BP neural network under digital economy
WO2021082809A1 (en) Training optimization method for foreign exchange time series prediction
Treleaven et al. Algorithmic trading review
Conegundes et al. Beating the stock market with a deep reinforcement learning day trading system
WO2021082810A1 (en) Construction method for foreign exchange time series prediction
CN108305167A (en) A kind of foreign currency trade method and system enhancing learning algorithm based on depth
CN112418575A (en) Futures initiative contract quantitative timing decision system based on cloud computing and artificial intelligence deep learning algorithm
Huang et al. Deep reinforcement learning for portfolio management
Chalvatzis et al. High-performance stock index trading: making effective use of a deep LSTM neural network
CN113919944A (en) Stock trading method and system based on reinforcement learning algorithm and time series model
Zhang et al. TradeBot: Bandit learning for hyper-parameters optimization of high frequency trading strategy
Labadie et al. Optimal algorithmic trading and market microstructure
Ge et al. Single stock trading with deep reinforcement learning: A comparative study
Crawford et al. Automatic high-frequency trading: An application to emerging chilean stock market
CN114998010A (en) Stock trading decision method based on deep reinforcement learning and market emotion
CN115423499A (en) Model training method, price prediction method, terminal device, and storage medium
Huang et al. Algorithmic trading using combinational rule vector and deep reinforcement learning
Guo et al. Market Making with Deep Reinforcement Learning from Limit Order Books
Weng Quantitative Trading Method based on Neural Network Machine Learning
Hao et al. Application of Deep Reinforcement Learning in Financial Quantitative Trading
Zhang Forecasting financial performance of companies for stock valuation
Zong et al. Deep Reinforcement Learning for Pairs Trading: Evidence from Soybean Commodities
Huang et al. Comprehensive scoring trading model based on LSTM prediction
Supriyanto Comparison of Grid Search and Evolutionary Parameter Optimization with Neural Networks on JCI Stock Price Movements during the Covid 19
Wan Trading strategy model based on LSTM neural network and Extreme Value-Dynamic programming

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20210226

WD01 Invention patent application deemed withdrawn after publication