WO2023185089A1 - 金融衍生品价格的预测方法、预测装置、存储介质和设备 - Google Patents

金融衍生品价格的预测方法、预测装置、存储介质和设备 Download PDF

Info

Publication number
WO2023185089A1
WO2023185089A1 PCT/CN2022/138142 CN2022138142W WO2023185089A1 WO 2023185089 A1 WO2023185089 A1 WO 2023185089A1 CN 2022138142 W CN2022138142 W CN 2022138142W WO 2023185089 A1 WO2023185089 A1 WO 2023185089A1
Authority
WO
WIPO (PCT)
Prior art keywords
disaster
prediction
data
financial
model
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.)
Ceased
Application number
PCT/CN2022/138142
Other languages
English (en)
French (fr)
Inventor
姜青山
吴胤旭
周亚雯
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.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
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 Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Publication of WO2023185089A1 publication Critical patent/WO2023185089A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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 OR CALCULATING; 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
    • Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping

Definitions

  • the invention belongs to the field of electronic information technology. Specifically, it relates to a prediction method, prediction device, computer-readable storage medium and computer equipment for the price of financial derivatives.
  • non-parametric estimation was used to study the impact of earthquakes on financial markets. Taking into account the volatility, dependence and "fat tail" characteristics of return data, researchers proposed a non-parametric estimation method without any requirements on data distribution. Instead, let the data speak for itself, which overcomes one of the major drawbacks of parameter estimation. It is also feasible to use this method to study the impact of the event for a period of time.
  • non-parametric estimation is not perfect, it is highly data-dependent and prone to overfitting.
  • neural networks can well avoid the problem of overfitting.
  • the values generated by traditional artificial neural networks may not be optimal. Therefore, the weights are adjusted to the optimal value for the determined number of epochs through backpropagation, the output layer is connected to the hidden layer, and the process is continuously repeated, so that the error is minimized and the prediction is improved.
  • the model will be trained to obtain a recurrent neural network based on traditional artificial intelligence. This process uses past data and makes predictions about the future. So past data should be remembered to predict the next value, however, the model cannot store long-term memory. In order to solve this problem, the LSTM model was born.
  • the technical problem solved by this invention is: how to retain all the characteristics of the original disaster data as much as possible while overcoming the problem of data sparsity to obtain a more accurate prediction method.
  • a method for predicting the price of financial derivatives includes:
  • the acquired financial derivatives transaction historical data set is divided into disaster-free transaction data sets and several types of disaster-related transaction data sets according to chronological order and disaster type, wherein the disaster-free transaction data set includes financial derivatives characteristic data and price data.
  • each type of disaster data set includes disaster characteristic data, financial derivatives characteristic data and price data;
  • the disaster-free transaction data set is used as a training sample to train the pre-built LSTM neural network model to obtain the disaster-free prediction sub-model, and the disaster-free transaction data set of each type is used as a training sample to train each pre-built LSTM neural network model.
  • the network model is trained to obtain a disaster prediction sub-model, and the disaster-free prediction sub-model and several disaster prediction models together form a total prediction model;
  • the data to be predicted obtained in real time is input into the total prediction model to obtain the price prediction sequence results of financial derivatives.
  • the financial derivatives transaction historical data set includes financial market historical data and disaster historical data
  • the acquired financial derivatives transaction historical data set is divided into disaster-free transaction data sets and several categories according to chronological order and disaster type.
  • Methods with disaster transaction data sets include:
  • the non-empty set of financial market historical data and disaster historical data are combined into disaster-free transaction data sets, earthquake disaster transaction data sets, flood disaster transaction data sets, and tropical storm disaster transaction data sets according to disaster types.
  • the several disaster prediction sub-models are earthquake disaster prediction sub-model, flood disaster prediction sub-model, and tropical storm disaster prediction sub-model.
  • the data to be predicted obtained in real time includes financial derivatives characteristic data and disaster characteristic data, and the data to be predicted in real time is input into the total prediction model to obtain the price prediction sequence results of financial derivatives.
  • Methods include:
  • the obtained prediction values are sorted in chronological order to obtain the price prediction sequence results.
  • the prediction device includes:
  • the data segmentation module is used to segment the acquired financial derivatives transaction historical data set into disaster-free transaction data sets and several types of disaster-related transaction data sets according to chronological order and disaster type, wherein the disaster-free transaction data set includes financial derivatives Product characteristic data and price data, each type of disaster data set includes disaster characteristic data, financial derivatives characteristic data and price data;
  • the model training module is used to use the disaster-free transaction data set as a training sample to train the pre-built LSTM neural network model to obtain a disaster-free prediction sub-model, and in turn uses the disaster-free transaction data set of each type as a training sample for each category.
  • a pre-built LSTM neural network model is trained to obtain a disaster prediction sub-model, and the disaster-free prediction sub-model and several disaster prediction models together form a total prediction model;
  • the price prediction module is used to input the data to be predicted obtained in real time into the total prediction model to obtain the price prediction sequence results of financial derivatives.
  • the financial derivatives transaction historical data set includes financial market historical data and disaster historical data
  • the data segmentation module includes:
  • the sorting unit is used to add timestamps to each piece of financial market historical data and disaster historical data and sort them in chronological order;
  • the filtering unit is used to filter out the non-empty set of financial market historical data and disaster historical data in chronological order;
  • the combination unit is used to combine the non-empty set of financial market historical data and disaster historical data into disaster-free transaction data sets, earthquake disaster transaction data sets, flood disaster transaction data sets, and tropical storm disaster transaction data sets according to disaster types.
  • the data to be predicted obtained in real time includes financial derivatives characteristic data and disaster characteristic data
  • the price prediction module includes:
  • the judgment unit is used to judge whether the financial derivatives characteristic data is accompanied by disaster characteristic data at the same time in chronological order;
  • the first input unit is used to simultaneously input the financial derivatives characteristic data and the disaster characteristic data into the disaster prediction system with the same disaster type as the disaster characteristic data when the financial derivative characteristic data is accompanied by the disaster characteristic data at the same time.
  • the predicted value is obtained;
  • the second input unit is used to input the financial derivatives characteristic data into the disaster-free prediction sub-model to obtain the prediction value when the financial derivatives characteristic data is accompanied by disaster characteristic data at the same time;
  • the splicing unit is used to sort the obtained prediction values in chronological order to obtain the price prediction sequence results.
  • This application also provides a computer-readable storage medium that stores a prediction program for the price of financial derivatives.
  • the program for predicting the price of financial derivatives is executed by a processor, the above-mentioned financial derivatives are realized.
  • Price prediction methods are realized.
  • This application also provides a computer device, which includes a computer-readable storage medium, a processor, and a prediction program for financial derivative prices stored in the computer-readable storage medium.
  • a prediction program for financial derivative prices stored in the computer-readable storage medium.
  • the invention discloses a method, prediction device, storage medium and equipment for predicting the price of financial derivatives. Compared with the existing technology, it has the following technical effects:
  • This method first splits the original data set, trains individual prediction models separately, and then integrates them to obtain a total prediction model. It effectively solves the problem of disaster data scarcity, retains more information of the original data, and effectively integrates data from different disaster models. The impact of financial derivatives has led to an overall improvement in financial derivatives price predictions under sudden disasters.
  • Figure 1 is a flow chart of a method for predicting financial derivatives prices according to Embodiment 1 of the present invention
  • Figure 2 is a segmentation flow chart of a financial derivatives transaction history data set according to Embodiment 1 of the present invention
  • Figure 3 is a process diagram of prediction using a test data set according to Embodiment 1 of the present invention.
  • Figure 4 is a functional block diagram of a financial derivatives price prediction device according to Embodiment 2 of the present invention.
  • Figure 5 is a schematic diagram of computer equipment according to Embodiment 4 of the present invention.
  • each piece of data to be predicted is input into the matching prediction sub-model. Finally, each prediction result is arranged in chronological order to obtain a complete price prediction sequence. This method splits the Each data set formed after the classification is separately trained to predict the model, which can avoid the sparsity of the data. At the same time, because the original data is not simplified, more information can be retained.
  • the method for predicting financial derivative prices in Embodiment 1 includes the following steps:
  • Step S10 Divide the obtained financial derivatives transaction historical data set into disaster-free transaction data sets and several types of disaster-related transaction data sets according to chronological order and disaster type.
  • the disaster-free transaction data set includes financial derivatives characteristic data and prices.
  • Data, each type of disaster data set includes disaster characteristic data, financial derivatives characteristic data and price data;
  • Step S20 Use the disaster-free transaction data set as a training sample to train the pre-built LSTM neural network model to obtain a disaster-free prediction sub-model, and use each type of disaster-related transaction data set as a training sample to train each pre-built LSTM.
  • the neural network model is trained to obtain a disaster prediction sub-model, and the disaster-free prediction sub-model and several disaster prediction models together form a total prediction model;
  • Step S30 Input the data to be predicted obtained in real time into the total prediction model to obtain the price prediction sequence results of financial derivatives.
  • the financial derivatives transaction history data set includes financial market historical data and disaster historical data
  • the financial market historical data is financial derivatives characteristic data and corresponding price data.
  • disaster historical data web crawler technology is used to crawl the corresponding property loss amount, death toll and other information from disaster news to form a disaster historical database.
  • the source of disaster news can be news websites of countries around the world, and the disaster history database is stored in a local database.
  • step S10 includes the following steps:
  • Step S101 Add timestamps to each piece of financial market historical data and disaster historical data and sort them in chronological order;
  • Step S102 Filter out the non-empty set of financial market historical data and disaster historical data in chronological order;
  • Step S103 Combine the non-empty set of financial market historical data and disaster historical data into disaster-free transaction data sets, earthquake disaster transaction data sets, flood disaster transaction data sets, and tropical storm disaster transaction data sets according to disaster types.
  • each piece of data after obtaining the historical data of financial market conditions and disasters, each piece of data will be timestamped, merged by timestamp, sorted by time, and split into a training set and a test set. .
  • the same process is performed on the training set and the test set respectively, that is, split by disaster type. First, determine whether the data is an empty set. If it is an empty set, skip the piece of data. Otherwise, obtain the first piece of data. Then, determine whether the first piece of data is accompanied by earthquake disaster characteristic data. If so, add the earthquake disaster.
  • the financial derivatives transaction historical data set is divided into a disaster-free transaction data set, an earthquake disaster transaction data set, a flood disaster transaction data set, and a tropical disaster transaction data set.
  • Storm disaster transaction data set includes financial derivatives characteristic data and price data
  • the earthquake disaster transaction data set includes financial derivatives characteristic data, price data and earthquake disaster characteristic data
  • the flood disaster transaction data set includes financial derivatives product characteristic data, price data and flood disaster characteristic data
  • the tropical storm disaster transaction data set includes financial derivatives characteristic data, price data and tropical storm disaster characteristic data.
  • each LSTM neural network model includes a three-layer structure, an input layer, a hidden layer and an output layer.
  • An LSTM layer with dimension N function where the hidden layer is an LSTM structure, the dimension of the hidden layer is J ⁇ K, and there are J network nodes; the Linear function is used as the activation function of the output layer, and the output node is 1.
  • the disaster-free transaction data set is used to train to obtain the disaster-free prediction sub-model
  • the earthquake disaster transaction data set is used to train the earthquake disaster prediction sub-model
  • the flood disaster transaction data set is used to train the flood disaster prediction sub-model.
  • the tropical storm disaster prediction sub-model was trained using the tropical storm disaster transaction data set.
  • the data to be predicted obtained in real time includes financial derivatives characteristic data and disaster characteristic data.
  • the model is actually applied, it is judged in chronological order whether the financial derivatives characteristic data is accompanied by disaster characteristic data at the same time; if If yes, then input the financial derivatives characteristic data and disaster characteristic data into the disaster prediction sub-model of the same disaster type as the disaster characteristic data at the same time to obtain the predicted value; if not, input the financial derivatives characteristic data into the disaster-free prediction sub-model.
  • the predicted values are obtained; the obtained predicted values are sorted in chronological order to obtain the price prediction sequence.
  • the second embodiment also discloses a prediction device for financial derivative prices.
  • the prediction device includes a data segmentation module 100 , a model training module 200 and a price prediction module 300 .
  • the data segmentation module 100 is used to segment the acquired financial derivatives transaction history data set into disaster-free transaction data sets and several types of disaster-related transaction data sets according to chronological order and disaster type, where the disaster-free transaction data set includes financial derivatives characteristics.
  • Data and price data, each type of disaster data set includes disaster characteristic data, financial derivatives characteristic data and price data.
  • the model training module 200 is used to use the disaster-free transaction data set as a training sample to train the pre-built LSTM neural network model to obtain a disaster-free prediction sub-model, and in turn uses each type of disaster-related transaction data set as a training sample to train each pre-built LSTM neural network model.
  • the constructed LSTM neural network model is trained to obtain a disaster prediction sub-model, and the disaster-free prediction sub-model and several disaster prediction models together form a total prediction model.
  • the price prediction module 300 is used to input the data to be predicted obtained in real time into the overall prediction model to obtain the price prediction sequence results of financial derivatives.
  • the data segmentation module 100 includes a sorting unit, a filtering unit and a combining unit.
  • the sorting unit is used to add timestamps to each financial market historical data and disaster historical data and sort them in chronological order;
  • the filtering unit is used to filter out non-empty financial market historical data and disasters in chronological order.
  • the combination unit is used to combine non-empty historical financial market data and disaster historical data into disaster-free transaction data sets, earthquake disaster transaction data sets, flood disaster transaction data sets, and tropical storm disaster transaction data sets according to disaster types.
  • For the specific working processes of the sorting unit, filtering unit and combining unit please refer to the relevant description of Embodiment 1, and will not be described in detail here.
  • the price prediction module 300 includes a judgment unit, a first input unit, a second input unit and a splicing unit.
  • the judgment unit is used to judge whether the financial derivatives characteristic data is accompanied by disaster characteristic data at the same time in order of time;
  • the first input unit is used to determine whether the financial derivatives characteristic data is accompanied by disaster characteristic data at the same time.
  • the financial derivatives characteristic data and the disaster characteristic data are simultaneously input into the disaster prediction sub-model with the same disaster type as the disaster characteristic data to obtain the predicted value;
  • the second input unit is used when the financial derivatives characteristic data is accompanied by a disaster at the same time.
  • the splicing unit is used to sort the obtained prediction values in chronological order to obtain the price prediction sequence results.
  • the first input unit, the second input unit and the splicing unit please refer to the description of Embodiment 1, and will not be described again here.
  • Embodiment 3 also discloses a computer-readable storage medium.
  • the computer-readable storage medium stores a prediction program for the price of financial derivatives.
  • the prediction program for the price of financial derivatives is executed by the processor, the above-mentioned prediction of the price of financial derivatives is realized. method.
  • Embodiment 4 also discloses a computer device.
  • the computer device includes a processor 12 , an internal bus 13 , a network interface 14 , and a computer-readable storage medium 11 .
  • the processor 12 reads the corresponding computer program from the computer-readable storage medium and then runs it, forming a request processing device at the logical level.
  • one or more embodiments of this specification do not exclude other implementations, such as logic devices or a combination of software and hardware, etc. That is to say, the execution subject of the following processing flow is not limited to each A logic unit can also be a hardware or logic device.
  • the computer-readable storage medium 11 stores a financial derivatives price prediction program. When the financial derivatives price prediction program is executed by the processor, the above-mentioned financial derivatives price prediction method is implemented.
  • Computer-readable storage media includes permanent and non-transitory, removable and non-removable media and may be implemented by any method or technology to store information. Information may be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer-readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage , magnetic tape cartridges, magnetic disk storage, quantum memory, graphene-based storage media or other magnetic storage devices or any other non-transmission medium, can be used to store information that can be accessed by a computing device.
  • PRAM phase change memory
  • SRAM static random access memory
  • DRAM dynamic random access memory
  • RAM random access memory
  • ROM read-only memory memory
  • EEPROM electrically era

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Finance (AREA)
  • Strategic Management (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Data Mining & Analysis (AREA)
  • General Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Game Theory and Decision Science (AREA)
  • Technology Law (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Tourism & Hospitality (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Quality & Reliability (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本发明公开了一种金融衍生品价格的预测方法、预测装置、存储介质和设备。预测方法包括：将获取到的金融衍生品交易历史数据集按照时间顺序和灾难类型分割成无灾难交易数据集和若干类有灾难交易数据集；将无灾难交易数据集作为训练样本训练得到无灾难预测子模型，依次将每一类有灾难交易数据集分别作为训练样本训练得到有灾难预测子模型，将无灾难预测子模型和若干个有灾难预测模型共同构成总预测模型；将实时获取到的待预测数据输入到总预测模型中，得到金融衍生品的价格预测序列结果。通过将原始数据集进行拆分后分别训练对应的预测模型，有效克服了灾难数据稀疏性问题，同时原始数据集未进行简化处理，保留了更多信息，有利于提高预测准确度。

Description

金融衍生品价格的预测方法、预测装置、存储介质和设备 技术领域
本发明属于电子信息技术领域，具体地讲，涉及一种金融衍生品价格的预测方法、预测装置、计算机可读存储介质和计算机设备。
 背景技术
如今全球范围内，自然灾难的发生频次越来越高，洪水、地震、疫情等都在潜移默化的影响着各行各业，最终集中反映在金融市场上。根据灾难降临时其带来影响的数据，对金融市场进行预测，来指导投资者及时的改变自己的投资策略躲避风险与损失是当前急需解决的问题。目前的预测方法包括事件分析法、非参数估计法、基于机器学习的方法简单线性回归
目前的事件分析法对金融市场的研究多停留在描述性比较与分析的层面，而进行定量研究时亦多采用事件分析法，在思路上可谓非常直观，也是最为普遍的方法，通过测量事件发生当日的异常收益及事件窗口内的累积收益，并检验其统计显著性，进而验证某一事件的发生是否对市场行为产生影响。但是，要用如此简单的方法对现实数据进行拟合，需要对数据进行非常严格的分布假设，而这在现实世界中往往是难以成立的。
进一步地，使用非参数估计研究地震对金融市场的影响，考虑了收益数据的波动性、依赖性及“肥尾”等特性，研究者们提出了非参数估计方法，对数据分布没有任何要求，而是让数据自己说话，这就克服了参数估计对一大弊端。同时运用这一方法对事件发生后一段时期的影响进行研究也是可行的。但是，非参数估计也非十全十美的，它具有很高的数据依赖性，且容易过度拟合。
通过减少或控制训练周期，在数据出现拐点前停止训练，神经网络可以很好地避免过度拟合的问题。传统的人工神经网络的生成值可能不是最优值。因此，通过反向传播将权重调整为所确定历元数的最佳值，将输出层连接到隐藏层，并不断重复该过程，使得误差降到最低，预测得到改善。此过程完成后，将对模型进行训练，得到基于传统人工智能的递归神经网络。这个过程使用过去的数据，并作出未来的预测。因此应记住过去的数据以预测下一个值，然而，该模型不能存储长期记忆。为了解决该问题，LSTM模型自此诞生。
由于灾难数据的稀疏性，若将数据不经处理后地喂入LSTM模型，将导致预测结果极差或无法得到结果。因此，大多数研究选择将灾难特征直接处理成零一变量，或者将预测结果处理成分类变量，使得结果更加具备可解释性，但这两种方式都将原数据进行了简化可能会丢失很多信息。
 技术问题
本发明解决的技术问题是：如何尽可能地保留灾难原始数据的所有特征，同时克服数据稀疏性的问题，以获得准确性更高的预测方法。
 技术解决方案
一种金融衍生品价格的预测方法，所述预测方法包括：
将获取到的金融衍生品交易历史数据集按照时间顺序和灾难类型分割成无灾难交易数据集和若干类有灾难交易数据集，其中所述无灾难交易数据集包括金融衍生品特征数据和价格数据，每类所述有灾难数据集包括灾难特征数据、金融衍生品特征数据和价格数据；
将无灾难交易数据集作为训练样本对预先构建的LSTM神经网络模型进行训练得到无灾难预测子模型，依次将每一类所述有灾难交易数据集分别作为训练样本对每一个预先构建的LSTM神经网络模型进行训练得到有灾难预测子模型，将所述无灾难预测子模型和若干个有灾难预测模型共同构成总预测模型；
将实时获取到的待预测数据输入到所述总预测模型中，得到金融衍生品的价格预测序列结果。
优选地，所述金融衍生品交易历史数据集包括金融市场行情历史数据和灾难历史数据，将获取到的金融衍生品交易历史数据集按照时间顺序和灾难类型分割成无灾难交易数据集和若干类有灾难交易数据集的方法包括：
将各条金融市场行情历史数据和灾难历史数据都加上时间戳并按时间顺序进行排序；
按照时间先后顺序，筛选出非空集的金融市场行情历史数据和灾难历史数据；
将非空集的金融市场行情历史数据和灾难历史数据按照灾难类型组合成无灾难交易数据集、地震灾难交易数据集、洪水灾难交易数据集、热带风暴灾难交易数据集。
优选地，若干个有灾难预测子模型分别是地震灾难预测子模型、洪水灾难预测子模型、热带风暴灾难预测子模型。
优选地，所述实时获取到的待预测数据包括金融衍生品特征数据和灾难特征数据，所述将实时获取到待预测数据输入到所述总预测模型中，得到金融衍生品的价格预测序列结果的方法包括：
按照时间先后顺序，依次判断在同一时间上金融衍生品特征数据是否附带有灾难特征数据；
若有，则将所述金融衍生品特征数据和灾难特征数据同时输入到与灾难特征数据的灾难类型相同的有灾难预测子模型中，得到预测值；
若无，则将所述金融衍生品特征数据输入到无灾难预测子模型中，得到预测值；
将得到的各个预测值按照时间先后顺序进行排序得到价格预测序列结果。
本申请还公开了一种金融衍生品价格的预测装置，所述预测装置包括：
数据分割模块，用于将获取到的金融衍生品交易历史数据集按照时间顺序和灾难类型分割成无灾难交易数据集和若干类有灾难交易数据集，其中所述无灾难交易数据集包括金融衍生品特征数据和价格数据，每类所述有灾难数据集包括灾难特征数据、金融衍生品特征数据和价格数据；
模型训练模块，用于将无灾难交易数据集作为训练样本对预先构建的LSTM神经网络模型进行训练得到无灾难预测子模型，依次将每一类所述有灾难交易数据集分别作为训练样本对每一个预先构建的LSTM神经网络模型进行训练得到有灾难预测子模型，将所述无灾难预测子模型和若干个有灾难预测模型共同构成总预测模型；
价格预测模块，用于将实时获取到的待预测数据输入到所述总预测模型中，得到金融衍生品的价格预测序列结果。
优选地，所述金融衍生品交易历史数据集包括金融市场行情历史数据和灾难历史数据，所述数据分割模块包括：
排序单元，用于将各条金融市场行情历史数据和灾难历史数据都加上时间戳并按时间顺序进行排序；
筛选单元，用于按照时间先后顺序，筛选出非空集的金融市场行情历史数据和灾难历史数据；
组合单元，用于将非空集的金融市场行情历史数据和灾难历史数据按照灾难类型组合成无灾难交易数据集、地震灾难交易数据集、洪水灾难交易数据集、热带风暴灾难交易数据集。
优选地，所述实时获取到的待预测数据包括金融衍生品特征数据和灾难特征数据，所述价格预测模块包括：
判断单元，用于按照时间先后顺序，依次判断在同一时间上金融衍生品特征数据是否附带有灾难特征数据；
第一输入单元，用于当同一时间上金融衍生品特征数据附带有灾难特征数据时，将所述金融衍生品特征数据和灾难特征数据同时输入到与灾难特征数据的灾难类型相同的有灾难预测子模型中，得到预测值；
第二输入单元，用于当同一时间上金融衍生品特征数据附带有灾难特征数据时，将所述金融衍生品特征数据输入到无灾难预测子模型中，得到预测值；
拼接单元，用于将得到的各个预测值按照时间先后顺序进行排序得到价格预测序列结果。
本申请还提供了一种计算机可读存储介质，所述计算机可读存储介质存储有金融衍生品价格的预测程序，所述金融衍生品价格的预测程序被处理器执行时实现上述的金融衍生品价格的预测方法。
本申请还提供了一种计算机设备，所述计算机设备包括计算机可读存储介质、处理器和存储在所述计算机可读存储介质中的金融衍生品价格的预测程序，所述金融衍生品价格的预测程序被处理器执行时实现上述的金融衍生品价格的预测方法。
 有益效果
本发明公开了一种金融衍生品价格的预测方法、预测装置、存储介质和设备，相对于现有技术，具有如下技术效果：
该方法先对原始数据集进行拆分，分别训练单个预测模型，再整合得到总预测模型，有效地解决了灾难数据稀缺性的问题，保留了原始数据的更多信息，有效融合来自不同灾难对金融衍生品的影响，在突发灾难下金融衍生品价格预测实现全面提升。
 附图说明
图1为本发明的实施例一的金融衍生品价格的预测方法的流程图；
图2为本发明的实施例一的金融衍生品交易历史数据集的分割流程图；
图3为本发明的实施例一的利用测试数据集进行预测的过程图；
图4为本发明的实施例二的金融衍生品价格的预测装置的原理框图；
图5为本发明的实施例四的计算机设备示意图。
 本发明的实施方式
为了使本发明的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本发明进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
在详细描述本申请的各个实施例之前，首先简单描述本申请的发明构思：现有技术在利用与金融交易行情相关的灾难数据对机器学习模型进行训练时，为了克服灾难数据的稀疏性问题，往往直接将灾难数据直接处理成零一变量，或者将预测结果处理成分类变量，但是这两种处理方式都将原始数据进行了简化，导致丢失很多信息，为此，本申请提供了一种金融衍生品价格的预测方法，按照不同灾难类型将原始数据分割多类数据集，同时未对原始数据进行零一化处理，接着利用各个数据集单独训练LSTM神经网络模型，得到相应的预测子模型，各个预测子模型共同构成总预测模型，在实际应用时将各条待预测数据输入到相匹配的预测子模型中，最后将各个预测结果按照时间顺序排列得到完整的价格预测序列，本方法通过拆分后形成的各个数据集单独训练预测模型，可以避免数据的稀疏性，同时由于原始数据未进行简化处理，可以保留更多信息。
具体地，如图1所示，本实施例一的金融衍生品价格的预测方法包括如下步骤：
步骤S10：将获取到的金融衍生品交易历史数据集按照时间顺序和灾难类型分割成无灾难交易数据集和若干类有灾难交易数据集，其中无灾难交易数据集包括金融衍生品特征数据和价格数据，每类有灾难数据集包括灾难特征数据、金融衍生品特征数据和价格数据；
步骤S20：将无灾难交易数据集作为训练样本对预先构建的LSTM神经网络模型进行训练得到无灾难预测子模型，依次将每一类有灾难交易数据集分别作为训练样本对每一个预先构建的LSTM神经网络模型进行训练得到有灾难预测子模型，将无灾难预测子模型和若干个有灾难预测模型共同构成总预测模型；
步骤S30：将实时获取到的待预测数据输入到总预测模型中，得到金融衍生品的价格预测序列结果。
具体地，在步骤S10中，金融衍生品交易历史数据集包括金融市场行情历史数据和灾难历史数据，金融市场行情历史数据即为金融衍生品特征数据和对应的价格数据。针对灾难历史数据，利用网络爬虫技术，从灾难新闻中爬取相应的财产损失金额、死亡人数等信息，形成灾难历史数据库。其中灾难新闻来源端可以是全球各国国家的新闻网站，并将灾难历史数据库存储在本地数据库中。进一步地，步骤S10包括如下步骤：
步骤S101、将各条金融市场行情历史数据和灾难历史数据都加上时间戳并按时间顺序进行排序；
步骤S102、按照时间先后顺序，筛选出非空集的金融市场行情历史数据和灾难历史数据；
步骤S103、将非空集的金融市场行情历史数据和灾难历史数据按照灾难类型组合成无灾难交易数据集、地震灾难交易数据集、洪水灾难交易数据集、热带风暴灾难交易数据集。
示例性地，如图2所示，将获取金融市场行情和灾难的历史数据后，将每条数据都加上时间戳，并按时间戳合并，按时间排序并拆分成训练集和测试集。对训练集和测试集分别进行相同的处理，即按灾难类型进行拆分。首先，判断数据是否为空集，若为空集则跳过该条数据，否则获取第一条数据；接着，判断该第一条条数据是否附带有地震灾难特征数据，若有则加入地震灾难交易数据集；然后，判断该条数据是否附带有洪水灾难特征数据，若有，则加入洪水灾难交易数据集；接着，判断该条数据是否附带有热带风暴灾难特征数据，若有则加入热带风暴灾难交易数据集；然后，若有其他灾难类型，接着判断，并将数据加入相应的数据集。最后若以上判断结果均为否，则加入无灾难交易数据集。一条数据判断完成，进行下一条数据判断，直到所有数据集中所有数据判断完成，至此将金融衍生品交易历史数据集分割形成无灾难交易数据集、地震灾难交易数据集、洪水灾难交易数据集、热带风暴灾难交易数据集，这里的无灾难交易数据集包括金融衍生品特征数据价格数据；地震灾难交易数据集包括金融衍生品特征数据、价格数据和地震灾难特征数据；洪水灾难交易数据集包括金融衍生品特征数据、价格数据和洪水灾难特征数据；热带风暴灾难交易数据集包括金融衍生品特征数据、价格数据和热带风暴灾难特征数据。
进一步地，在步骤S20中，每个LSTM神经网络模型包括三层结构，一层输入层、一层隐含层和一层输出层。维度为N×1的LSTM层作为输入层，将tanh函数作为输入层内部激活函数和输入层与隐含层之间的激活函数，输入层网络节点是N个；Tanh函数作为隐含层内部激活函数，其中隐含层是LSTM结构，隐含层的维度是J×K的，有J个网络节点；Linear函数作为输出层的激活函数，输出节点为1。
进一步地，在训练每个LSTM神经网络模型时，将每一类训练样本的80%作为训练集，20%作为验证集，训练的过程中利用交叉验证的方法来调节参数，选择平均平方误差作为损失函数，选择RMSprop作为随机梯度下降算法训练模型，具体的训练过程为现有技术，在此不进行赘述。具体来说，本实施例一利用无灾难交易数据集训练得到无灾难预测子模型，利用地震灾难交易数据集训练得到地震灾难预测子模型，利用洪水灾难交易数据集训练得到洪水灾难预测子模型，利用热带风暴灾难交易数据集训练得到热带风暴灾难预测子模型。
最后，实时获取到的待预测数据包括金融衍生品特征数据和灾难特征数据，在实际应用模型时，按照时间先后顺序，依次判断在同一时间上金融衍生品特征数据是否附带有灾难特征数据；若有，则将金融衍生品特征数据和灾难特征数据同时输入到与灾难特征数据的灾难类型相同的有灾难预测子模型中，得到预测值；若无，则将金融衍生品特征数据输入到无灾难预测子模型中，得到预测值；将得到的各个预测值按照时间先后顺序进行排序得到价格预测序列。
示例性地，如图3所示，按照时间先后顺序，对于每条金融衍生品特征数据，判断该数据是否附带有地震灾难特征数据，若是，则输入地震灾难预测子模型；若否，判断是否附带有洪水灾难特征数据，若是，则输入洪水灾难预测子模型；然后，判断是否附带有热带风暴灾难特征数据，若是，则输入热带风暴灾难预测子模型；最后，若前面的判断结果均为否，则输入无灾难预测子模型。概括来说，即先判断是当天发生了哪个灾难，投入对应的训练模型，得到一个预测值，最后再将所有得到的预测值按时间戳排列得到一条完整的价格预测序列。
为了验证本实施例一的有效性和先进性，在地震、洪水、热带风暴三个基准数据集上进行了广泛的实验，以评估所提出的方法的性能。采用准确度、召回率、F1值三个方面的评价指标对本方法的效果进行评估，与经典CNN模型、SVM模型以及线性回归模型进行比对，实验结果一致表明，利用本实施例一所提出的预测方法可有效利用稀疏的灾难数据，可以被广泛地应用于金融产品价格的预测，具有十分广阔的应用前景。
本实施例二还公开了一种金融衍生品价格的预测装置，如图4所示，预测装置包括数据分割模块100、模型训练模块200和价格预测模块300。数据分割模块100用于将获取到的金融衍生品交易历史数据集按照时间顺序和灾难类型分割成无灾难交易数据集和若干类有灾难交易数据集，其中无灾难交易数据集包括金融衍生品特征数据和价格数据，每类有灾难数据集包括灾难特征数据、金融衍生品特征数据和价格数据。模型训练模块200用于将无灾难交易数据集作为训练样本对预先构建的LSTM神经网络模型进行训练得到无灾难预测子模型，依次将每一类有灾难交易数据集分别作为训练样本对每一个预先构建的LSTM神经网络模型进行训练得到有灾难预测子模型，将无灾难预测子模型和若干个有灾难预测模型共同构成总预测模型。价格预测模块300用于将实时获取到的待预测数据输入到总预测模型中，得到金融衍生品的价格预测序列结果。
其中，数据分割模块100包括排序单元、筛选单元和组合单元。排序单元用于将各条金融市场行情历史数据和灾难历史数据都加上时间戳并按时间顺序进行排序；筛选单元用于按照时间先后顺序，筛选出非空集的金融市场行情历史数据和灾难历史数据；组合单元用于将非空集的金融市场行情历史数据和灾难历史数据按照灾难类型组合成无灾难交易数据集、地震灾难交易数据集、洪水灾难交易数据集、热带风暴灾难交易数据集。其中，排序单元、筛选单元和组合单元的具体工作过程可参见实施例一的相关描述，在此不进行赘述。
进一步地，价格预测模块300包括判断单元、第一输入单元、第二输入单元和拼接单元。判断单元用于按照时间先后顺序，依次判断在同一时间上金融衍生品特征数据是否附带有灾难特征数据；第一输入单元用于当同一时间上金融衍生品特征数据附带有灾难特征数据时，将金融衍生品特征数据和灾难特征数据同时输入到与灾难特征数据的灾难类型相同的有灾难预测子模型中，得到预测值；第二输入单元用于当同一时间上金融衍生品特征数据附带有灾难特征数据时，将金融衍生品特征数据输入到无灾难预测子模型中，得到预测值；拼接单元用于将得到的各个预测值按照时间先后顺序进行排序得到价格预测序列结果。判断单元、第一输入单元、第二输入单元和拼接单元的具体工作过程可参见实施例一的描述，在此不进行赘述。
实施例三还公开了一种计算机可读存储介质，计算机可读存储介质存储有金融衍生品价格的预测程序，金融衍生品价格的预测程序被处理器执行时实现上述的金融衍生品价格的预测方法。
进一步地，实施例四还公开了一种计算机设备，在硬件层面，如图5所示，该计算机设备包括处理器12、内部总线13、网络接口14、计算机可读存储介质11。处理器12从计算机可读存储介质中读取对应的计算机程序然后运行，在逻辑层面上形成请求处理装置。当然，除了软件实现方式之外，本说明书一个或多个实施例并不排除其他实现方式，比如逻辑器件抑或软硬件结合的方式等等，也就是说以下处理流程的执行主体并不限定于各个逻辑单元，也可以是硬件或逻辑器件。计算机可读存储介质11上存储有金融衍生品价格的预测程序，金融衍生品价格的预测程序被处理器执行时实现上述的金融衍生品价格的预测方法。
计算机可读存储介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机可读存储介质的例子包括，但不限于相变内存（PRAM）、静态随机存取存储器（SRAM）、动态随机存取存储器（DRAM）、其他类型的随机存取存储器（RAM）、只读存储器（ROM）、电可擦除可编程只读存储器（EEPROM）、快闪记忆体或其他内存技术、只读光盘只读存储器（CD-ROM）、数字多功能光盘（DVD）或其他光学存储、磁盒式磁带、磁盘存储、量子存储器、基于石墨烯的存储介质或其他磁性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。
上面对本发明的具体实施方式进行了详细描述，虽然已表示和描述了一些实施例，但本领域技术人员应该理解，在不脱离由权利要求及其等同物限定其范围的本发明的原理和精神的情况下，可以对这些实施例进行修改和完善，这些修改和完善也应在本发明的保护范围内。

Claims (9)

  1. 一种金融衍生品价格的预测方法，其特征在于，所述预测方法包括：
    将获取到的金融衍生品交易历史数据集按照时间顺序和灾难类型分割成无灾难交易数据集和若干类有灾难交易数据集，其中所述无灾难交易数据集包括金融衍生品特征数据和价格数据，每类所述有灾难数据集包括灾难特征数据、金融衍生品特征数据和价格数据；
    将无灾难交易数据集作为训练样本对预先构建的LSTM神经网络模型进行训练得到无灾难预测子模型，依次将每一类所述有灾难交易数据集分别作为训练样本对每一个预先构建的LSTM神经网络模型进行训练得到有灾难预测子模型，将所述无灾难预测子模型和若干个有灾难预测模型共同构成总预测模型；
    将实时获取到的待预测数据输入到所述总预测模型中，得到金融衍生品的价格预测序列结果。
  2. 根据权利要求1所述的金融衍生品价格的预测方法，其特征在于，所述金融衍生品交易历史数据集包括金融市场行情历史数据和灾难历史数据，将获取到的金融衍生品交易历史数据集按照时间顺序和灾难类型分割成无灾难交易数据集和若干类有灾难交易数据集的方法包括：
    将各条金融市场行情历史数据和灾难历史数据都加上时间戳并按时间顺序进行排序；
    按照时间先后顺序，筛选出非空集的金融市场行情历史数据和灾难历史数据；
    将非空集的金融市场行情历史数据和灾难历史数据按照灾难类型组合成无灾难交易数据集、地震灾难交易数据集、洪水灾难交易数据集、热带风暴灾难交易数据集。
  3. 根据权利要求2所述的金融衍生品价格的预测方法，其特征在于，若干个有灾难预测子模型分别是地震灾难预测子模型、洪水灾难预测子模型、热带风暴灾难预测子模型。
  4. 根据权利要求3所述的金融衍生品价格的预测方法，其特征在于，所述实时获取到的待预测数据包括金融衍生品特征数据和灾难特征数据，所述将实时获取到待预测数据输入到所述总预测模型中，得到金融衍生品的价格预测序列结果的方法包括：
    按照时间先后顺序，依次判断在同一时间上金融衍生品特征数据是否附带有灾难特征数据；
    若有，则将所述金融衍生品特征数据和灾难特征数据同时输入到与灾难特征数据的灾难类型相同的有灾难预测子模型中，得到预测值；
    若无，则将所述金融衍生品特征数据输入到无灾难预测子模型中，得到预测值；
    将得到的各个预测值按照时间先后顺序进行排序得到价格预测序列结果。
  5. 一种金融衍生品价格的预测装置，其特征在于，所述预测装置包括：
    数据分割模块，用于将获取到的金融衍生品交易历史数据集按照时间顺序和灾难类型分割成无灾难交易数据集和若干类有灾难交易数据集，其中所述无灾难交易数据集包括金融衍生品特征数据和价格数据，每类所述有灾难数据集包括灾难特征数据、金融衍生品特征数据和价格数据；
    模型训练模块，用于将无灾难交易数据集作为训练样本对预先构建的LSTM神经网络模型进行训练得到无灾难预测子模型，依次将每一类所述有灾难交易数据集分别作为训练样本对每一个预先构建的LSTM神经网络模型进行训练得到有灾难预测子模型，将所述无灾难预测子模型和若干个有灾难预测模型共同构成总预测模型；
    价格预测模块，用于将实时获取到的待预测数据输入到所述总预测模型中，得到金融衍生品的价格预测序列结果。
  6. 根据权利要求5所述的金融衍生品价格的预测装置，其特征在于，所述金融衍生品交易历史数据集包括金融市场行情历史数据和灾难历史数据，所述数据分割模块包括：
    排序单元，用于将各条金融市场行情历史数据和灾难历史数据都加上时间戳并按时间顺序进行排序；
    筛选单元，用于按照时间先后顺序，筛选出非空集的金融市场行情历史数据和灾难历史数据；
    组合单元，用于将非空集的金融市场行情历史数据和灾难历史数据按照灾难类型组合成无灾难交易数据集、地震灾难交易数据集、洪水灾难交易数据集、热带风暴灾难交易数据集。
  7. 根据权利要求6所述的金融衍生品价格的预测装置，其特征在于，所述实时获取到的待预测数据包括金融衍生品特征数据和灾难特征数据，所述价格预测模块包括：
    判断单元，用于按照时间先后顺序，依次判断在同一时间上金融衍生品特征数据是否附带有灾难特征数据；
    第一输入单元，用于当同一时间上金融衍生品特征数据附带有灾难特征数据时，将所述金融衍生品特征数据和灾难特征数据同时输入到与灾难特征数据的灾难类型相同的有灾难预测子模型中，得到预测值；
    第二输入单元，用于当同一时间上金融衍生品特征数据附带有灾难特征数据时，将所述金融衍生品特征数据输入到无灾难预测子模型中，得到预测值；
    拼接单元，用于将得到的各个预测值按照时间先后顺序进行排序得到价格预测序列结果。
  8. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质存储有金融衍生品价格的预测程序，所述金融衍生品价格的预测程序被处理器执行时实现权利要求1至4任一项所述的金融衍生品价格的预测方法。
  9. 一种计算机设备，其特征在于，所述计算机设备包括计算机可读存储介质、处理器和存储在所述计算机可读存储介质中的金融衍生品价格的预测程序，所述金融衍生品价格的预测程序被处理器执行时实现权利要求1至4任一项所述的金融衍生品价格的预测方法。
PCT/CN2022/138142 2022-03-29 2022-12-09 金融衍生品价格的预测方法、预测装置、存储介质和设备 Ceased WO2023185089A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210324409.3 2022-03-29
CN202210324409.3A CN114820199A (zh) 2022-03-29 2022-03-29 金融衍生品价格的预测方法、预测装置、存储介质和设备

Publications (1)

Publication Number Publication Date
WO2023185089A1 true WO2023185089A1 (zh) 2023-10-05

Family

ID=82532388

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/138142 Ceased WO2023185089A1 (zh) 2022-03-29 2022-12-09 金融衍生品价格的预测方法、预测装置、存储介质和设备

Country Status (2)

Country Link
CN (1) CN114820199A (zh)
WO (1) WO2023185089A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114820199A (zh) * 2022-03-29 2022-07-29 深圳先进技术研究院 金融衍生品价格的预测方法、预测装置、存储介质和设备

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106991506A (zh) * 2017-05-16 2017-07-28 深圳先进技术研究院 智能终端及其基于lstm的股票趋势预测方法
CN107481048A (zh) * 2017-08-08 2017-12-15 哈尔滨工业大学深圳研究生院 一种基于混合模型的金融品种价格预测方法及系统
WO2019214143A1 (zh) * 2018-05-10 2019-11-14 平安科技（深圳）有限公司 服务器、金融时序数据的处理方法及存储介质
US20200167869A1 (en) * 2016-04-16 2020-05-28 Overbond Ltd. Real-time predictive analytics engine
CN111291925A (zh) * 2020-01-18 2020-06-16 武汉盛信鸿通科技有限公司 一种基于人工智能的金融市场预测及决策的系统、方法
CN112613909A (zh) * 2020-12-21 2021-04-06 山东麦港数据系统有限公司 一种基于改进lstm模型的农产品短期价格预测方法及装置
CN114820199A (zh) * 2022-03-29 2022-07-29 深圳先进技术研究院 金融衍生品价格的预测方法、预测装置、存储介质和设备

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200167869A1 (en) * 2016-04-16 2020-05-28 Overbond Ltd. Real-time predictive analytics engine
CN106991506A (zh) * 2017-05-16 2017-07-28 深圳先进技术研究院 智能终端及其基于lstm的股票趋势预测方法
CN107481048A (zh) * 2017-08-08 2017-12-15 哈尔滨工业大学深圳研究生院 一种基于混合模型的金融品种价格预测方法及系统
WO2019214143A1 (zh) * 2018-05-10 2019-11-14 平安科技（深圳）有限公司 服务器、金融时序数据的处理方法及存储介质
CN111291925A (zh) * 2020-01-18 2020-06-16 武汉盛信鸿通科技有限公司 一种基于人工智能的金融市场预测及决策的系统、方法
CN112613909A (zh) * 2020-12-21 2021-04-06 山东麦港数据系统有限公司 一种基于改进lstm模型的农产品短期价格预测方法及装置
CN114820199A (zh) * 2022-03-29 2022-07-29 深圳先进技术研究院 金融衍生品价格的预测方法、预测装置、存储介质和设备

Also Published As

Publication number Publication date
CN114820199A (zh) 2022-07-29

Similar Documents

Publication Publication Date Title
CN113240187B (zh) 预测模型生成方法、系统、设备、存储介质及预测方法
CN107194803A (zh) 一种p2p网贷借款人信用风险评估的装置
CN114492978B (zh) 一种基于多层注意力机制的时空序列预测方法及设备
Dobrea et al. Machine Learning algorithms for air pollutants forecasting
CN114168575A (zh) 一种金融领域舆情分析方法及系统
CN105260794A (zh) 一种云数据中心的负载预测方法
CN114757432A (zh) 基于流程日志和多任务学习的未来执行活动及时间预测方法及系统
CN109636017A (zh) 一种金融交易价格预测方法、装置、介质及设备
CN117034143A (zh) 一种基于机器学习的分布式系统故障诊断方法及装置
Du et al. Design and implementation of China financial risk monitoring and early warning system based on deep learning
CN115756922A (zh) 一种故障预测诊断方法、装置、电子设备及存储介质
CN118643467A (zh) 基于多特征融合和二维时间卷积网络的日径流预测方法
CN115660855A (zh) 一种融合新闻数据的股票收盘价预测方法
CN103942251A (zh) 基于多种质控方法的高空气象资料入库方法和入库系统
CN118611052A (zh) 基于图结构网络的分布式光伏功率协同预测方法及系统
CN111680712A (zh) 基于日内相似时刻的变压器油温预测方法、装置及系统
WO2023185089A1 (zh) 金融衍生品价格的预测方法、预测装置、存储介质和设备
CN111985845B (zh) 一种异构Spark集群的节点优先级调优方法
CN114492967A (zh) 基于ceemdan和blstm组合模型的城轨车站客流预测方法和介质
CN112766814A (zh) 一种信用风险压力测试模型的训练方法、装置及设备
CN119359053B (zh) 一种海外高铁城区隧道建设风险评估方法及系统
CN114971022A (zh) 基于D-value-LSTM神经网络模型的风电功率预测方法
CN114418638A (zh) 一种基于数据与ai驱动的产品价格预测方法
CN118644329A (zh) 交易量预测模型的训练方法及装置、交易量预测方法及装置、电子设备及计算机可读存储介质
CN117709539A (zh) 一种现金流预测方法、系统、设备及介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22934906

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22934906

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 22934906

Country of ref document: EP

Kind code of ref document: A1