A multi-level classification screening method and device for off-site margin financing accounts

Abstract

The present application relates to the field of artificial intelligence technology, in particular to a multi-level classification screening method and device for off-site margin accounts, by dividing business modes, aiming at two off-site margin modes of "independent type" and "separate warehouse type", setting general rules, carrying out business scene division, expanding the scope of investigation mode, enhancing the pertinence of screening, adopting a composite index to form an index system instead of a simple index mode, which is more comprehensive, scientific and accurate, using a scientific screening process, comprehensively considering factors such as operation complexity, data size, system computing power, etc., preferentially screening indexes with simple calculation and typical characteristics, screening indexes with complex calculation and low data quality later, comprehensively considering the quality and efficiency of index stratification, improving the operability of the model, improving the recognition accuracy, screening coverage, operation efficiency and screening time range, setting the screening time range, and reducing random errors.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and specifically to a multi-level classification screening method and apparatus for off-exchange margin trading accounts. Background Technology

[0002] Off-exchange margin trading, a term in the securities market, refers to the practice of providing funds in proportion to an account's existing assets for securities trading. It constitutes an illegal off-exchange lending service within the securities market. Common off-exchange margin trading involves lending personal accounts, which are often mixed with ordinary personal securities accounts and managed only through systems outside the stock exchange. This lack of due diligence on participants' qualifications and risk tolerance, coupled with leverage ratios significantly exceeding the trading and risk-bearing capacity of ordinary investors, results in off-exchange margin trading attracting a large number of individual investors with speculative intent. For investors, excessive leverage amplifies reasonable returns during market upturns, fueling speculative impulses; conversely, during market downturns, leverage amplifies losses and ultimately forces forced liquidation, causing investors to suffer losses beyond their means. This risk is inherently hidden. When off-exchange margin trading reaches a certain level, it will increase stock market volatility.

[0003] Regarding methods for identifying off-exchange margin trading accounts, existing technologies have explored machine learning-based approaches. These methods learn from the characteristics of sample transactions, positions, funds, terminals, and market data to form corresponding feature thresholds. Currently, some commercial products use this method. However, regarding sample size, it is extremely difficult to obtain a sufficient number of off-exchange margin trading accounts for accurate classification and judgment. A small sample size can easily lead to overfitting, resulting in inaccurate predictions. Off-exchange margin trading accounts can only be accurately identified through case handling. Samples obtained from other channels lack sufficient verification methods, have inconsistent methods, and varying judgment criteria, and can only be classified as suspected off-exchange margin trading accounts. Forcibly using such inaccurate accounts as training samples can easily lead to the model becoming increasingly inaccurate. Therefore, this alternative approach has been ruled out. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-level classification screening method and apparatus for off-exchange margin trading accounts, which screens securities accounts that are highly suspected of being used for off-exchange margin trading in the securities market.

[0005] According to a first aspect of the present invention, the present invention claims protection for a multi-level classification screening method for off-exchange margin trading accounts, characterized in that:

[0006] Determine the screening time and the scope of investor accounts to be screened, and decide on the screening strategy;

[0007] Obtain the parameters to be screened for the investor account to be screened, perform margin financing screening on the parameters to be screened according to the corresponding screening strategy, output the margin financing identification result and judgment type of the investor account to be screened, and store the feature values ​​of each composite indicator of the account and the comprehensive score.

[0008] Based on the results of the margin trading identification and the judgment type, a comprehensive analysis is performed, and the front-end charts of the probe analysis system are displayed, supporting the analysis of investigation clues.

[0009] Furthermore, determining the screening time and the scope of investor accounts to be screened, and deciding on the screening strategy, also includes:

[0010] The scope of the investor account screening includes both market-wide screening and specific account screening.

[0011] When the scope of the screening of investor accounts is the entire market, the first screening strategy shall be adopted;

[0012] When the scope of the screening of investor accounts is limited to specific account screening, a second screening strategy shall be adopted.

[0013] Furthermore, the first screening strategy also includes:

[0014] The entire market accounts to be screened are input into the screening margin trading model, and the entire market accounts are subjected to sample reduction, behavior recognition and pattern recognition to obtain the first screening result;

[0015] Based on the first screening result, multi-source comprehensive analysis and reverse lookup of the financing platform are performed to obtain the first financing identification result.

[0016] Furthermore, the second screening strategy also includes:

[0017] Obtain a list of specific accounts for screening by focusing on key platforms and / or online public opinion;

[0018] The specific account list is verified by platform basic information verification, website code verification, transaction software configuration file verification, mobile software packet capture verification, mobile software decompilation verification, and credit-related account verification to obtain the specific verification account list;

[0019] The specific account list is screened by key terminals, phone numbers, and reverse securities account information to obtain a second specific account list;

[0020] Input the second specific account list into the screening margin trading model, and perform sample reduction, behavior recognition and pattern recognition on the second specific account list to obtain the second margin trading recognition result.

[0021] Furthermore, the process of obtaining the parameters to be screened for the investor accounts, performing margin financing screening on the parameters to be screened according to the corresponding screening strategy, outputting the margin financing identification results and judgment type of the investor accounts, and storing the scores of each composite indicator feature value and the comprehensive score of the account, specifically includes:

[0022] Based on the corresponding screening strategy, the parameters to be screened for the investor account are input into the screening margin trading model to reduce the screening sample and obtain the first candidate screening account;

[0023] The first candidate screening account is subjected to account behavior identification to obtain the candidate margin trading pattern identification result;

[0024] Based on the candidate margin trading pattern recognition results, the first candidate screening account is subjected to margin trading pattern recognition to obtain the margin trading pattern recognition results.

[0025] Furthermore, the step of inputting the parameters to be screened from the investor accounts into the screening model according to the corresponding screening strategy to reduce the screening sample and obtain the first candidate screening accounts also includes:

[0026] Set parameters such as the start and end dates of the study;

[0027] Based on investor classification information, identify accounts whose stock holdings meet a set threshold.

[0028] Based on the results of the previous step, match the identifiers of all algorithmic trading accounts in the market, and retain the ordinary accounts and margin accounts that have been removed from the algorithmic trading accounts;

[0029] Based on the results of the previous step, match the securities trading information and extract the daily transaction data of the accounts to be screened;

[0030] Based on the results of the previous step, accounts with transaction amounts exceeding the set threshold, transaction days exceeding the set threshold, and transaction frequency exceeding the set threshold during the observation period will be retained.

[0031] Based on the results of the previous step, accounts that have a higher percentage of T+0 trading days, T+0 trading frequency, or T+0 trading amount during the observation period will be removed.

[0032] Based on the results of the previous step, the subscription and redemption information of ETFs in the entire market is matched to obtain a list of accounts whose proportion of the number of days participating in ETF arbitrage during the observation period is higher than the set threshold, which are then used as the first candidate screening accounts.

[0033] Furthermore, the step of performing account behavior identification on the first candidate screening account to obtain candidate margin trading pattern identification results also includes:

[0034] Based on the first candidate screening account, securities holding information and transaction calendar information are matched to calculate the total market value of the holdings of the first candidate screening account on a daily basis.

[0035] Based on the first candidate screening account, securities transaction information and transaction calendar information are matched to calculate the daily total transaction amount and number of transactions for the first candidate screening account;

[0036] Based on the first candidate screening account, match bank-securities transfer information, securities account and fund account correspondence information, and transaction calendar information to calculate the daily bank-securities transfer-in amount, daily bank-securities transfer-out amount, and fund account balance of the first candidate screening account;

[0037] Based on the first candidate screening account, match bank-securities transfer information, securities account and fund account correspondence information, transaction calendar information, and fund settlement information of ordinary account and credit account to calculate the details of the number of times the first candidate screening account is used by different investors on various terminal devices.

[0038] Based on the results of the previous step, the number of times each of the various terminal devices involved in the first candidate screening account was used by different investors was counted.

[0039] Furthermore, the step of performing account behavior identification on the first candidate screening account to obtain candidate margin trading pattern identification results also includes:

[0040] The first candidate screening account is identified for both independent margin trading pattern and sub-account margin trading pattern.

[0041] The identification of the independent margin trading pattern includes:

[0042] Based on the daily total market value of holdings and the daily amount transferred from bank to securities accounts of the first candidate screening account, a score is calculated for any single-direction change in funds from day T to day T+1 during the observation period that exceeds the set threshold.

[0043] Based on the daily total market value of the first candidate screening account and the daily amount of bank-securities transfer, the scores are calculated based on the holdings on day T and day T+1 during the observation period being higher than the set threshold, and the funds changing in a single direction from day T to day T+2.

[0044] Based on the rapid asset changes and rapid liquidation of the first candidate screening accounts, a list of accounts with rapid asset changes is compiled.

[0045] Based on rapidly changing account information and the daily transaction information of the first candidate screening account, obtain the daily transaction information of the account with rapidly changing information;

[0046] Based on the results of the previous step, count the number of transactions in the observation period whose maximum and minimum transaction amounts exceed the set threshold.

[0047] Based on the results of rapidly changing accounts, the number of times the transaction amount difference exceeds a set threshold during the observation period;

[0048] Based on the daily transaction information of the first candidate screening account, the total daily transaction amount of the account is calculated.

[0049] Based on the results of the previous step, calculate the number of times the difference in intraday transaction amount exceeds the set threshold during the observation period;

[0050] Based on the daily bank-securities transfer amount of the first candidate screening account, calculate the number of times the intraday transaction amount difference exceeds the set threshold during the observation period;

[0051] Based on the individual cases where the scores of the accounts are higher than the set threshold, the accounts with stage scores that are higher than the set threshold are used as the first candidate financing pattern identification results.

[0052] The identification of the sub-account type margin trading model also includes:

[0053] Based on the account list of the first candidate screening account, the transaction flow information is matched to obtain the transaction flow information of the first candidate screening account;

[0054] Based on the transaction amount information of the first candidate screening account, calculate the number of days during the observation period when the difference in the intraday transaction amount of the target is higher than the set threshold, the maximum daily transaction frequency, and the maximum number of targets traded in a single day;

[0055] Based on the transaction history of the first candidate screening account, calculate the number of days during the observation period where the frequency of small-amount transactions exceeds a set threshold;

[0056] Based on the transaction history of the first candidate screening account, calculate the number of times the number of transactions exceeds a set threshold within a set time slice during the observation period;

[0057] Based on the first candidate screening account, the holding information is matched to obtain the holding information of the first candidate screening account;

[0058] Based on the holding information of the first candidate screening account, calculate the number of days during the observation period when the number of small-amount holdings of the first candidate screening account exceeds a set threshold.

[0059] Based on the holding information of the first candidate screening account, calculate the number of days during which the maximum and minimum holding ratios of the first candidate screening account were higher than a set threshold within the respective observation period;

[0060] Based on the individual cases where the scores exceed the set threshold, accounts with scores exceeding the set threshold in the aggregation stage are used as the second candidate financing pattern identification results.

[0061] Furthermore, the step of identifying the financing pattern of the first candidate screening account based on the candidate financing pattern identification result to obtain the financing pattern identification result also includes:

[0062] Summarize the identification results of the first candidate margin trading pattern and the second candidate margin trading pattern, and add independent and sub-account type identifiers;

[0063] Based on the results of the previous step, the corresponding fund accounts for each account are obtained by matching the information corresponding to the securities account and the fund account.

[0064] Based on the results of the previous step, match bank-securities transfer information, transaction calendar information, and fund settlement information of ordinary accounts and credit accounts to obtain terminal usage information during the observation period;

[0065] Based on the results of the previous step, the number of IP and MAC terminal devices used for transactions and bank-securities transfers during the observation period was counted, and transaction and transfer identifiers were marked.

[0066] Based on the results of the previous step, we categorized and obtained the MAC terminal devices with the highest frequency of transactions and bank-securities transfers during the account review period, and compared their consistency.

[0067] Based on the user information of various terminals of the first candidate screening account, statistics were compiled on the cases where multiple accounts shared the MAC addresses used for transactions and bank-securities transfers during the observation period.

[0068] Based on the various terminal usage information of the first candidate screening account, obtain the location information of the IP terminal used;

[0069] Based on the results of the previous step, we categorized and obtained the IP terminal devices with the highest frequency of transactions and bank-securities transfers during the account review period, and compared their consistency.

[0070] Based on the user information of various terminals of the first candidate screening account, statistics were compiled on the cases where multiple accounts shared the IP address used for transactions and bank-securities transfers during the observation period;

[0071] Based on the terminal usage information of the first candidate screening account, the percentage of MAC devices with the highest transaction frequency is obtained;

[0072] Based on the terminal usage information of the first candidate screening account, the percentage of IP devices with the highest transaction frequency is obtained;

[0073] Based on the terminal usage information of the first candidate screening account and the list of key IPs and MACs, obtain the accounts that have used the key IPs and MACs;

[0074] Based on the identification strategies for independent and sub-account margin trading accounts, the corresponding comprehensive scores are calculated to obtain the margin trading pattern identification results.

[0075] According to a second aspect of the present invention, the present invention claims protection for a multi-level classification screening device for off-exchange margin trading accounts, characterized in that it comprises:

[0076] The screening strategy determination module determines the screening time and the scope of investor accounts to be screened, and makes decisions on the screening strategy.

[0077] The classification and screening module obtains the parameters to be screened for the investor accounts to be screened, performs margin financing screening on the parameters to be screened according to the corresponding screening strategy, outputs the margin financing identification results and judgment type of the investor accounts to be screened, and stores the feature values ​​of each composite indicator of the account and the comprehensive score.

[0078] The comprehensive analysis module performs a comprehensive analysis based on the financing identification results and judgment type, displays the front-end charts of the probe analysis system, and supports the analysis of investigation clues;

[0079] The aforementioned multi-level classification screening device for off-exchange margin trading accounts is used to execute the aforementioned multi-level classification screening method for off-exchange margin trading accounts.

[0080] This invention relates to the field of artificial intelligence technology, specifically to a multi-level classification screening method and device for off-exchange margin trading accounts. By dividing business models into "independent" and "segmented" off-exchange margin trading models, it classifies them according to business scenarios based on general rules, expanding the scope of the examination model and enhancing the targeting of the screening. It adopts a composite indicator system, using a combination of indicators rather than simple indicator models, making it more comprehensive, scientific, and accurate. A scientific screening process is used, comprehensively considering factors such as computational complexity, data scale, and system computing power. Indicators with simple calculations and typical characteristics are prioritized for screening, while those with complex calculations and low data quality are screened later. This comprehensive consideration of the quality and efficiency of indicator stratification improves the model's operability, resulting in high identification accuracy, high screening coverage, high computational efficiency, and a long screening time range (approximately six months) to reduce random errors. Attached Figure Description

[0081] Figure 1 This is a flowchart illustrating a multi-level classification screening method for off-exchange margin trading accounts as described in this invention.

[0082] Figure 2 This is a second workflow diagram of a multi-level classification screening method for off-exchange margin trading accounts involved in the present invention;

[0083] Figure 3 This is a structural module diagram of a multi-level classification screening device for off-exchange margin trading accounts involved in the present invention. Detailed Implementation

[0084] According to a first embodiment of the present invention, referring to the appendix Figure 1 This invention claims protection for a multi-level classification screening method for off-exchange margin trading accounts, characterized by comprising the following steps:

[0085] S1. Determine the screening time and the scope of investor accounts to be screened, and decide on the screening strategy;

[0086] S2. Obtain the parameters to be screened for the investor account to be screened, perform the margin financing screening of the parameters to be screened according to the corresponding screening strategy, output the margin financing identification result and judgment type of the investor account to be screened, and store the feature values ​​of each composite indicator of the account and the comprehensive score.

[0087] S3. Based on the financing identification results and judgment type, a comprehensive analysis is performed to display the front-end charts of the probe analysis system and support the analysis of investigation clues.

[0088] Furthermore, step S1 also includes:

[0089] The scope of the investor account screening includes both market-wide screening and specific account screening.

[0090] When the scope of the screening of investor accounts is the entire market, the first screening strategy shall be adopted;

[0091] When the scope of the screening of investor accounts is limited to specific account screening, a second screening strategy shall be adopted.

[0092] Further, refer to the appendix. Figure 2 The first screening strategy further includes:

[0093] The entire market accounts to be screened are input into the screening margin trading model, and the entire market accounts are subjected to sample reduction, behavior recognition and pattern recognition to obtain the first screening result;

[0094] Based on the first screening result, multi-source comprehensive analysis and reverse lookup of the financing platform are performed to obtain the first financing identification result.

[0095] The second screening strategy also includes:

[0096] Obtain a list of specific accounts for screening by focusing on key platforms and / or online public opinion;

[0097] The specific account list is verified by platform basic information verification, website code verification, transaction software configuration file verification, mobile software packet capture verification, mobile software decompilation verification, and credit-related account verification to obtain the specific verification account list;

[0098] The specific account list is screened by key terminals, phone numbers, and reverse securities account information to obtain a second specific account list;

[0099] Input the second specific account list into the screening margin trading model, and perform sample reduction, behavior recognition and pattern recognition on the second specific account list to obtain the second margin trading recognition result.

[0100] In this embodiment, the key aspects of the platform account screening include:

[0101] I. Sources of Clues: Clues regarding margin trading platforms mainly come from two sources: firstly, platforms under close monitoring; and secondly, key platforms identified through online public opinion analysis. This embodiment uses the "Analysis Method for Obtaining Platform Lists Through Online Public Opinion Analysis" to obtain a list of key platforms under close monitoring.

[0102] 1. Platform List Compilation

[0103] The system identifies platform characteristics to determine if it includes modules for registration, deposits, margin trading, transactions, and settlement. It extracts platform keywords, assessing their relevance to the keyword target database, noting that different keywords have different weights. Platform popularity is calculated and ranked accordingly. Combining the above characteristic weights, keyword weights, and popularity information with thresholds, the system identifies violations, filters and reduces non-margin trading website links, and removes duplicate website domains, retaining a total of 300 newly identified suspected off-exchange margin trading platforms.

[0104] 2. Business Status Verification

[0105] In response to the suspected continued operation of margin trading platform operators, a comprehensive review was conducted using databases, credit information, and other data to examine the business registration status, operator names, corporate credit codes, and registered addresses of the 300 platforms' corresponding operating companies.

[0106] 3. Platform content verification

[0107] For the margin trading platforms that were still in operation in the previous step, a web crawler was used to comprehensively investigate, categorize, and verify these platforms, including margin trading-related websites, non-margin trading platforms, and inaccessible websites. Finally, an automated, delayed access method was used to save screenshots of the website homepages for manual content verification. A total of 22 key margin trading platforms were verified.

[0108] 4. Main body fine screening

[0109] Based on the intellectual property rights and personnel information of the entities being verified, the list of platforms has been further narrowed down to 12.

[0110] II. Analytical Methods

[0111] 1. Platform basic information verification

[0112] By analyzing the information on the websites / apps of these 12 platforms, we obtained relevant company information, bank account information, platform operation models, financing models, traffic, and financing scale, and preserved screenshots of the corresponding interfaces as evidence.

[0113] 2. Website code verification

[0114] By analyzing the code of these 12 platform websites, we were able to find information about the transaction server terminals.

[0115] 3. Verification of computer trading software configuration files

[0116] By analyzing the configuration file code of the computer-based trading software, information about the trading server terminal and the parent account was found and further verified in the database.

[0117] 4. Mobile phone and website packet capture verification

[0118] By using packet capture analysis software, Windows Resource Monitor, and setting up proxy servers, we can obtain relevant terminal information of the margin trading platform's APP and website, and verify in the database whether there are any accounts that use this terminal information for transactions.

[0119] 5. Software decompilation verification

[0120] The installation package of the trading platform's software was decompiled, and the source files of the software were analyzed using regular expressions and key strings to obtain suspicious terminal information. The information was then verified in the central database to see if any accounts had used this terminal information for trading.

[0121] 6. Verification of Credit-Related Accounts

[0122] We verified the deletion of securities accounts by relevant company personnel through databases and the ZhengTong Credit System, and verified whether the accounts exhibited characteristics of margin trading.

[0123] III. Collection of key terminal and number information

[0124] Based on the six analysis methods and techniques described in the previous step, various terminal information corresponding to the platforms was obtained and summarized for use in the next stage of reverse lookup. This analysis and reverse lookup of 12 platforms yielded terminal information from 2 servers.

[0125] IV. Reverse Investigation of Margin Trading Accounts

[0126] Based on the terminal information list from the previous step, the terminal information lists corresponding to the two platforms were matched and compared with the total market account trading terminals and bank-securities transfer terminals to obtain the securities accounts using these terminals. After comparison, only one platform, Platform N, could be matched with securities accounts. Platform N involves one key server IP address. A reverse lookup revealed that two investors used the terminal: Mr. Niu (2 securities accounts) and Mr. Du (2 securities accounts). A total of 33 accounts were associated with the terminals of these two investors.

[0127] The resulting account list was screened and analyzed in three stages: sample reduction, basic feature identification, and margin trading pattern identification. The analysis covered a specific range of accounts.

[0128] Furthermore, step S2 specifically includes:

[0129] S2.1. Based on the corresponding screening strategy, input the parameters to be screened for the investor account to be screened into the screening margin trading model to reduce the screening sample and obtain the first candidate screening account;

[0130] S2.2. Perform account behavior identification on the first candidate screening account to obtain the candidate margin trading pattern identification result;

[0131] S2.3. Based on the candidate financing pattern recognition results, the first candidate screening account is subjected to financing pattern recognition to obtain financing pattern recognition results.

[0132] Furthermore, step S2.1 also includes:

[0133] S2.1(1) Set parameters such as the start and end dates of the investigation;

[0134] S2.1(2) Based on investor classification information, obtain accounts whose stock market value meets the set threshold;

[0135] S2.1(3) Based on the results of the previous step, match the identifiers of all algorithmic trading accounts in the market and retain the ordinary accounts and margin accounts that have been removed from the algorithmic trading accounts;

[0136] S2.1(4) Based on the results of the previous step, match the securities transaction information and extract the daily transaction volume of the account to be screened;

[0137] S2.1(5) Based on the results of the previous step, retain accounts whose transaction amount during the observation period is higher than the set threshold, whose number of transaction days is greater than the set threshold, and whose transaction frequency is higher than the set threshold.

[0138] S2.1(6) Based on the results of the previous step, remove accounts whose T+0 transaction days account for more than the set threshold, or whose T+0 transaction frequency account for more than the set threshold, or whose T+0 transaction amount account for more than the set threshold during the observation period;

[0139] S2.1(7) Based on the results of the previous step, match the ETF subscription and redemption information of the entire market to obtain a list of accounts whose proportion of the number of days participating in ETF arbitrage during the observation period is higher than the set threshold. A total of 37 accounts are retained after this step.

[0140] Furthermore, step S2.2 also includes:

[0141] S2.2(1) Based on the account list in step S2.1(7), match the securities holding information and transaction calendar information to calculate the total market value of the holdings of these 37 accounts on a daily basis.

[0142] S2.2(2) Based on the account list in step S2.1(7), match the securities transaction information and transaction calendar information to calculate the total daily transaction amount and number of transactions for these 37 accounts.

[0143] S2.2(3) Based on the account list in step S2.1(7), match the bank-securities transfer information, the correspondence between securities accounts and fund accounts, and the transaction calendar information to calculate the daily bank-securities transfer in amount, daily bank-securities transfer out amount, and fund account balance for these 37 accounts.

[0144] S2.2(4) Based on the account list in step S2.1(7), match the bank-securities transfer information, the correspondence between securities accounts and fund accounts, the transaction calendar information, and the fund settlement information of ordinary accounts and credit accounts to calculate the details of the number of times these 37 accounts are used by different investors on various terminal devices.

[0145] S2.2(5) Based on the results of the previous step, count the number of times each of the various terminal devices involved in these 37 accounts was used by different investors.

[0146] Furthermore, step S2.2 also includes:

[0147] The first candidate screening account is subjected to independent margin trading pattern identification step S2.2.1 and sub-account margin trading pattern identification step S2.2.2;

[0148] Step S2.2.1 includes:

[0149] S2.2.1(1) Based on the total market value of the holdings of account S2.2(1) and the amount of funds transferred into the securities account S2.2(3) daily, calculate the score for any single-direction change in funds from T to T+1 during the observation period that exceeds the set threshold.

[0150] S2.2.1(2) Based on the total market value of the daily holdings of the S2.2(1) account and the daily amount of bank-securities transfers of the S2.2(3) account, calculate the points for holdings on day T and day T+1 during the observation period that are higher than the set threshold and for funds to change in a single direction from day T to day T+2.

[0151] S2.2.1(3), Based on S2.2.1(1) rapid asset changes and S2.2.1(2) rapid liquidation statistics, a list of accounts with rapid asset changes is compiled;

[0152] S2.2.1(4), Based on the rapidly changing account information in S2.2.1(3) and the daily transaction information of the account in S2.2(2), obtain the daily transaction information of the account with rapid changes;

[0153] S2.2.1(5) Based on the results of the previous step, count the number of transactions in the observation period whose maximum and minimum transaction amounts are higher than the set threshold.

[0154] S2.2.1(6) Based on the results of S2.2.1(4), count the number of times the transaction amount difference exceeds the set threshold during the observation period;

[0155] S2.2.1(7) Based on the daily transaction information of the account in S2.2(2), the total transaction amount of the account is calculated daily;

[0156] S2.2.1(8) Based on the results of the previous step, calculate the number of times the difference in intraday transaction amount exceeds the set threshold during the observation period;

[0157] S2.2.1(9) Based on the daily bank-securities transfer amount of 37 accounts in S2.2(2), calculate the number of times the difference in intraday transaction amount exceeds the set threshold during the observation period;

[0158] S2.2.1(10), based on the cases where S2.2.1(5), S2.2.1(6), S2.2.1(8), and S2.2.1(9) are all above the set threshold, accounts with scores above the set threshold in the summary stage are identified. Six accounts were screened out of 37 accounts; the specific indicators involved can be found in Table 1.

[0159] Table 1. Screening Model Indicators for Sub-account Type (1:N) Off-exchange Margin Trading Accounts

[0160]

[0161]

[0162] Step S2.2.2 further includes:

[0163] S2.2.2(1) Based on the account list in step S2.1(7), match the transaction flow information to obtain the transaction flow information of 37 accounts;

[0164] S2.2.2(2) Based on the transaction amount information in S2.2(2), calculate the number of days during the observation period when the difference in the intraday transaction amount of the target is higher than the set threshold, the maximum daily transaction frequency, and the maximum number of targets traded in a single day;

[0165] S2.2.2(3) Based on the account transaction records in S2.2.2(1), calculate the number of days during the observation period when the frequency of small-amount transactions per day exceeds the set threshold;

[0166] S2.2.2(4) Based on the account transaction flow in S2.2.2(1), calculate the number of times the number of transactions in a set time slice during the observation period exceeds the set threshold;

[0167] S2.2.2(5) Based on the account list in step S2.1(7), match the holding information to obtain the holding information of 37 accounts;

[0168] S2.2.2(6) Based on the account holding information in S2.2.2(5), calculate the number of days during the observation period when the number of small-amount holdings of each of the 37 accounts exceeded the set threshold;

[0169] S2.2.2(7) Based on the account holding information in S2.2.2(5), calculate the number of days during which the maximum and minimum holding ratios of each of the 37 accounts were higher than the set threshold within the observation period;

[0170] S2.2.2(8), based on the cases where S2.2.2(2), S2.2.2(3), S2.2.2(4), S2.2.2(6), and S2.2.2(7) are all above the set threshold, accounts with scores above the set threshold are summarized. 8 accounts were screened out of 37 accounts; the specific indicators involved can be found in Table 2.

[0171] Table 2. Indicators for Screening Model of Independent (1:1) Off-exchange Margin Trading Accounts

[0172]

[0173]

[0174] Furthermore, step S2.3 also includes:

[0175] S2.3(1) Summarize the two types of result account tables in steps S2.2.1 and S2.2.2, and add independent type and sub-warehouse type labels;

[0176] S2.3(2) Based on the results of the previous step, match the securities account and fund account information to obtain the corresponding fund account for each account;

[0177] S2.3(3) Based on the results of the previous step, match the bank-securities transfer information, transaction calendar information, and fund settlement information of ordinary accounts and credit accounts to obtain terminal usage information during the observation period;

[0178] S2.3(4) Based on the results of the previous step, count the number of IP and MAC terminal devices used for transactions and bank-securities transfers during the observation period, and mark the transaction and transfer identifiers;

[0179] S2.3(5) Based on the results of the previous step, classify and obtain the MAC terminal devices with the highest frequency of use for transactions and bank-securities transfers during the account review period, and compare their consistency.

[0180] S2.3(6) Based on the terminal user information in S2.2(4), statistics were compiled on the situation where multiple accounts shared the MAC address used for transactions and bank-securities transfers during the observation period;

[0181] S2.3(7) Based on the terminal usage information in S2.3(3), obtain the location information of the IP terminal being used;

[0182] S2.3(8) Based on the results of the previous step, classify and obtain the IP terminal devices with the highest frequency of use for transactions and bank-securities transfers during the account review period, and compare their consistency.

[0183] S2.3(9) Based on the terminal user information in S2.2(4), statistics were compiled on the situation where multiple accounts shared the IP used for transactions and bank-securities transfers during the observation period;

[0184] S2.3(10) Based on the terminal usage information of S2.3(3), obtain the proportion of MAC devices with the highest transaction frequency;

[0185] S2.3(11) Based on the terminal usage information of S2.3(3), obtain the proportion of IP devices with the highest transaction frequency;

[0186] S2.3(12), Based on the terminal usage information and the list of key IPs and MACs in S2.3(3), obtain the accounts that have used the key IPs and MACs;

[0187] Independent margin trading account identification results:

[0188] Based on the cases where S2.2.1(10), S2.3(4), S2.3(5), S2.3(6), S2.3(8), and S2.3(9) each exceed the set threshold, accounts with scores exceeding the set threshold in the summary stage are identified. 0 accounts were found out from the 6 accounts screened out.

[0189] Results of identifying sub-account type margin trading accounts:

[0190] Based on the cases where S2.2.2(8), S2.3(4), S2.3(5), S2.3(6), S2.3(8), S2.3(9), S2.3(10), S2.3(11), and S2.3(12) each exceed the set threshold, accounts with scores exceeding the set threshold in the summary stage are identified. Two accounts belonging to Niu and one account belonging to Du were screened out from the eight accounts.

[0191] Comprehensive Analysis:

[0192] The results of the screening of key platform accounts (Section 1) and the screening results of specific account range (Section 2) were combined and analyzed. It was found that two accounts belonging to Niu (an operator associated with Platform N) and one account belonging to Du (another operator associated with Platform N) frequently used key server IP addresses for trading during the investigation period. The corresponding MAC addresses were highly suspected to be the addresses of the servers issuing trading orders. During the investigation period, the three accounts used this IP to trade 1345 stocks, with 18656 transactions, totaling 1.054 billion yuan in purchases and 1.044 billion yuan in sales. A large number of frequent transactions occurred 16 times within a set time slice, highly suspected to be the parent account of a sub-account type off-exchange margin trading account. Another account belonging to Du is suspected to be a test account of Platform N. The above analysis results will be reported and provided to the investigation and violation screening for analysis and reference.

[0193] According to a second embodiment of the present invention, referring to Figure 3 This invention claims protection for a multi-level classification screening device for off-exchange margin trading accounts, characterized in that it comprises:

[0194] The screening strategy determination module determines the screening time and the scope of investor accounts to be screened, and makes decisions on the screening strategy.

[0195] The classification and screening module obtains the parameters to be screened for the investor accounts to be screened, performs margin financing screening on the parameters to be screened according to the corresponding screening strategy, outputs the margin financing identification results and judgment type of the investor accounts to be screened, and stores the feature values ​​of each composite indicator of the account and the comprehensive score.

[0196] The comprehensive analysis module performs a comprehensive analysis based on the financing identification results and judgment type, displays the front-end charts of the probe analysis system, and supports the analysis of investigation clues;

[0197] The aforementioned multi-level classification screening device for off-exchange margin trading accounts is used to execute the aforementioned multi-level classification screening method for off-exchange margin trading accounts.

[0198] Those skilled in the art will understand that the contents disclosed herein can be varied and modified in many ways. For example, the various devices or components described above can be implemented in hardware, or in software, firmware, or a combination of some or all of the three.

[0199] This disclosure uses flowcharts to illustrate the steps of a method according to embodiments of this disclosure. It should be understood that the preceding or following steps are not necessarily performed in exact order. Instead, the steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes.

[0200] Those skilled in the art will understand that all or part of the steps in the above methods can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk. Optionally, all or part of the steps in the above embodiments can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiments can be implemented in hardware or as a software functional module. This disclosure is not limited to any particular combination of hardware and software.

[0201] Unless otherwise defined, all terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It should also be understood that terms such as those defined in a common dictionary should be interpreted as having a meaning consistent with their meaning in the context of the relevant art, and not as having an idealized or highly formalized meaning, unless expressly defined herein.

[0202] The foregoing description is intended to illustrate the present disclosure and should not be construed as limiting it. While several exemplary embodiments of the present disclosure have been described, those skilled in the art will readily understand that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present disclosure. Therefore, all such modifications are intended to be included within the scope of the present disclosure as defined by the claims. It should be understood that the foregoing description is intended to illustrate the present disclosure and should not be construed as limiting it to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present disclosure is defined by the claims and their equivalents.

[0203] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0204] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A multi-level classification and screening method for off-exchange margin trading accounts, characterized in that: Determine the screening time and the scope of investor accounts to be screened, and decide on the screening strategy; Obtain the parameters to be screened for the investor account to be screened, perform margin financing screening on the parameters to be screened according to the corresponding screening strategy, output the margin financing identification result and judgment type of the investor account to be screened, and store the feature values ​​of each composite indicator of the account and the comprehensive score. Based on the results of the margin trading identification and the judgment type, a comprehensive analysis is performed to display the front-end charts of the probe analysis system and support the analysis of investigation clues; The process of determining the screening time and the scope of investor accounts to be screened, and deciding on the screening strategy, also includes: The scope of the screening of investor accounts includes both market-wide screening and specific account screening; When the scope of the investor account screening is the entire market, the first screening strategy shall be adopted; When the scope of the investor account screening is specific account screening, a second screening strategy is adopted; The first screening strategy further includes: The entire market accounts to be screened are input into the screening margin trading model, and the entire market accounts are subjected to sample reduction, behavior recognition and pattern recognition to obtain the first screening result; Based on the first screening result, multi-source comprehensive analysis and reverse lookup of the financing platform are performed to obtain the first financing identification result; The second screening strategy also includes: Obtain a list of specific accounts for screening by focusing on key platforms and / or online public opinion; The specific account list is verified by platform basic information verification, website code verification, transaction software configuration file verification, mobile software packet capture verification, mobile software decompilation verification, and credit-related account verification to obtain the specific verification account list; The specific account list is screened by key terminals, phone numbers, and reverse securities account information to obtain a second specific account list; Input the second specific account list into the screening margin trading model, and perform sample reduction, behavior recognition and pattern recognition on the second specific account list to obtain the second margin trading recognition result; The process involves obtaining the parameters to be screened for the investor accounts, performing margin financing screening on the parameters according to the corresponding screening strategy, outputting the margin financing identification results and judgment type of the investor accounts, and storing the scores of each composite indicator feature value and the comprehensive score of the account. Specifically, this includes: Based on the corresponding screening strategy, the parameters to be screened for the investor account are input into the screening margin trading model to reduce the screening sample and obtain the first candidate screening account; The first candidate screening account is subjected to account behavior identification to obtain the candidate margin trading pattern identification result; Based on the candidate margin trading pattern recognition results, the first candidate screening account is subjected to margin trading pattern recognition to obtain the margin trading pattern recognition results; The step of identifying account behavior of the first candidate screening account to obtain candidate margin trading pattern identification results also includes: The first candidate screening account is identified for both independent margin trading pattern and sub-account margin trading pattern. The identification of the independent margin trading model includes: Based on the daily total market value of holdings and the daily amount transferred from bank to securities accounts of the first candidate screening account, a score is calculated for any single-direction change in funds from day T to day T+1 during the observation period that exceeds the set threshold. Based on the daily total market value of the first candidate screening account and the daily amount of bank-securities transfer, the scores are calculated based on the holdings on day T and day T+1 during the observation period being higher than the set threshold, and the funds changing in a single direction from day T to day T+2. Based on the rapid asset changes and rapid liquidation of the first candidate screening accounts, a list of accounts with rapid asset changes is compiled. Based on rapidly changing account information and the daily transaction information of the first candidate screening account, obtain the daily transaction information of the account with rapidly changing information; Based on the results of the previous step, count the number of transactions in the observation period whose maximum and minimum transaction amounts exceed the set threshold. Based on the results of rapidly changing accounts, the number of times the transaction amount difference exceeds a set threshold during the observation period; Based on the daily transaction information of the first candidate screening account, the total daily transaction amount of the account is calculated. Based on the results of the previous step, calculate the number of times the difference in intraday transaction amount exceeds the set threshold during the observation period; Based on the daily bank-securities transfer amount of the first candidate screening account, calculate the number of times the intraday transaction amount difference exceeds the set threshold during the observation period; Based on the individual cases where the scores of the accounts are higher than the set threshold, the accounts with stage scores that are higher than the set threshold are used as the first candidate financing pattern identification results. The identification of the sub-account type margin trading model also includes: Based on the account list of the first candidate screening account, the transaction flow information is matched to obtain the transaction flow information of the first candidate screening account; Based on the transaction amount information of the first candidate screening account, calculate the number of days during the observation period when the difference in the intraday transaction amount of the target is higher than the set threshold, the maximum daily transaction frequency, and the maximum number of targets traded in a single day; Based on the transaction history of the first candidate screening account, calculate the number of days during the observation period where the frequency of small-amount transactions exceeds a set threshold; Based on the transaction history of the first candidate screening account, calculate the number of times the number of transactions exceeds a set threshold within a set time slice during the observation period; Based on the first candidate screening account, the holding information is matched to obtain the holding information of the first candidate screening account; Based on the holding information of the first candidate screening account, calculate the number of days during the observation period when the number of small-amount holdings of the first candidate screening account exceeds a set threshold. Based on the holding information of the first candidate screening account, calculate the number of days during which the maximum and minimum holding ratios of the first candidate screening account were higher than a set threshold within the respective observation period; Based on the individual cases where the scores exceed the set threshold, accounts with scores exceeding the set threshold in the aggregation stage are used as the second candidate financing pattern identification results.

2. The multi-level classification screening method for off-exchange margin trading accounts as described in claim 1, characterized in that, The step of inputting the parameters to be screened from the investor accounts into the screening model according to the corresponding screening strategy to reduce the screening sample and obtain the first candidate screening accounts also includes: Set the start and end dates for the study; Based on investor classification information, identify accounts whose stock holdings meet a set threshold. Based on the results of the previous step, match the identifiers of all algorithmic trading accounts in the market, and retain the ordinary accounts and margin accounts that have been removed from the algorithmic trading accounts; Based on the results of the previous step, match the securities trading information and extract the daily transaction data of the accounts to be screened; Based on the results of the previous step, accounts with transaction amounts exceeding the set threshold, transaction days exceeding the set threshold, and transaction frequency exceeding the set threshold during the observation period will be retained. Based on the results of the previous step, accounts that have a higher percentage of T+0 trading days, T+0 trading frequency, or T+0 trading amount during the observation period will be removed. Based on the results of the previous step, the subscription and redemption information of ETFs in the entire market is matched to obtain a list of accounts whose proportion of the number of days participating in ETF arbitrage during the observation period is higher than the set threshold, which are then used as the first candidate screening accounts.

3. The multi-level classification screening method for off-exchange margin trading accounts as described in claim 2, characterized in that, The step of identifying account behavior of the first candidate screening account to obtain candidate margin trading pattern identification results also includes: Based on the first candidate screening account, securities holding information and transaction calendar information are matched to calculate the total market value of the holdings of the first candidate screening account on a daily basis. Based on the first candidate screening account, securities transaction information and transaction calendar information are matched to calculate the daily total transaction amount and number of transactions for the first candidate screening account; Based on the first candidate screening account, match bank-securities transfer information, securities account and fund account correspondence information, and transaction calendar information to calculate the daily bank-securities transfer-in amount, daily bank-securities transfer-out amount, and fund account balance of the first candidate screening account; Based on the first candidate screening account, match bank-securities transfer information, securities account and fund account correspondence information, transaction calendar information, and fund settlement information of ordinary account and credit account to calculate the details of the number of times the first candidate screening account is used by different investors on various terminal devices. Based on the results of the previous step, the number of times each of the various terminal devices involved in the first candidate screening account was used by different investors was counted.

4. The multi-level classification screening method for off-exchange margin trading accounts as described in claim 3, characterized in that, The step of identifying the financing pattern of the first candidate screening account based on the candidate financing pattern identification result to obtain the financing pattern identification result further includes: Summarize the identification results of the first candidate margin trading pattern and the second candidate margin trading pattern, and add independent and sub-account type identifiers; Based on the results of the previous step, the corresponding fund accounts for each account are obtained by matching the information corresponding to the securities account and the fund account. Based on the results of the previous step, match bank-securities transfer information, transaction calendar information, and fund settlement information of ordinary accounts and credit accounts to obtain terminal usage information during the observation period; Based on the results of the previous step, the number of IP and MAC terminal devices used for transactions and bank-securities transfers during the observation period was counted, and transaction and transfer identifiers were marked. Based on the results of the previous step, we categorized and obtained the MAC terminal devices with the highest frequency of transactions and bank-securities transfers during the account review period, and compared their consistency. Based on the user information of various terminals of the first candidate screening account, statistics were compiled on the cases where multiple accounts shared the MAC addresses used for transactions and bank-securities transfers during the observation period. Based on the various terminal usage information of the first candidate screening account, obtain the location information of the IP terminal used; Based on the results of the previous step, we categorized and obtained the IP terminal devices with the highest frequency of transactions and bank-securities transfers during the account review period, and compared their consistency. Based on the user information of various terminals of the first candidate screening account, statistics were compiled on the cases where multiple accounts shared the IP address used for transactions and bank-securities transfers during the observation period; Based on the terminal usage information of the first candidate screening account, the percentage of MAC devices with the highest transaction frequency is obtained; Based on the terminal usage information of the first candidate screening account, the percentage of IP devices with the highest transaction frequency is obtained; Based on the terminal usage information of the first candidate screening account and the list of key IPs and MACs, obtain the accounts that have used the key IPs and MACs; Based on the identification strategies for independent and sub-account margin trading accounts, the corresponding comprehensive scores are calculated to obtain the margin trading pattern identification results.

5. A multi-level classification screening device for off-exchange margin trading accounts, characterized in that, include: The screening strategy determination module determines the screening time and the scope of investor accounts to be screened, and makes decisions on the screening strategy. The classification and screening module obtains the parameters to be screened for the investor accounts to be screened, performs margin financing screening on the parameters to be screened according to the corresponding screening strategy, outputs the margin financing identification results and judgment type of the investor accounts to be screened, and stores the feature values ​​of each composite indicator of the account and the comprehensive score. The comprehensive analysis module performs a comprehensive analysis based on the financing identification results and judgment type, displays the front-end charts of the probe analysis system, and supports the analysis of investigation clues; The multi-level classification screening device for off-exchange margin trading accounts is used to perform the multi-level classification screening method for off-exchange margin trading accounts as described in any one of claims 1-4.

Images