Intelligent private placement net value management system

Abstract

The invention belongs to the technical field of private placement net value management, and particularly relates to an intelligent private placement net value management system, which comprises a product creation module used for retrieving a created product after product information is input so as to avoid repeated input and management; the file acquisition module is used for capturing the mails from the mailbox and matching the mails to the products for analysis; the data verification is used for verifying the data so as to verify whether the data has the phenomena of field missing, data inconsistency and data missing or not; the data processing module is used for carrying out weight recovery net value calculation, frequency reduction net value calculation and performance deduction net value calculation on the data; and the early warning module is used for carrying out early warning according to the calculated indexes after the indexes are calculated. According to the method, file matching and file analysis are automatically completed, most of net value files with different formats in the market are intelligently compatible, and the efficiency of obtaining the original net value is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of private placement net value management, and specifically to an intelligent private placement net value management system. Background Art

[0002] Private placement net value management refers to the process of managing and monitoring the net value of private placement funds. The private placement net value refers to the total net assets remaining after subtracting the total liabilities from the total assets of a private placement fund on a specific date, and the net value per share of each fund is calculated by dividing the total net assets of each fund by the total number of fund shares, that is, the unit net value of the private placement fund.

[0003] However, the existing private placement net value management has the phenomenon of low intelligence level. For example, it cannot automatically match and parse files, which will affect the acquisition efficiency of the original net value to a certain extent. Therefore, an intelligent private placement net value management system is invented. Summary of the Invention

[0004] To solve the above technical problems, according to one aspect of the present invention, the following technical solutions are provided:

[0005] An intelligent private placement net value management system, which includes:

[0006] A product creation module, which is used to retrieve the created products after entering the product information, so as to avoid repeated entry and management;

[0007] A file acquisition module, which is used to grab emails from the mailbox and match them to products for parsing;

[0008] Data verification, which is used to check the data to see if there are phenomena such as missing fields, inconsistent data, and missing data;

[0009] A data processing module, which is used to calculate the restored net value, downsampled net value, and net value after deducting performance fees of the data;

[0010] An early warning module, which is used to give an early warning according to the calculated indicators after calculating the indicators.

[0011] As a preferred solution of the intelligent private placement net value management system described in the present invention, among them: the product creation module includes:

[0012] An entry module, which is used to enter the basic information of the product, and the basic information includes the filing code, product name, manager, and establishment date;

[0013] A retrieval module, which is used to give priority to retrieving the previously created products when creating a new product. If the retrieval exists, it will be given priority to use, so as to avoid repeated entry and management.

[0014] As a preferred solution of the intelligent private placement net value management system described in the present invention, wherein: the file acquisition module includes:

[0015] A scraping module, which is used to automatically scrape the titles, bodies and attachments of emails from the mailbox, and at the same time support manual upload by users;

[0016] A matching module, which is used to match files to products;

[0017] A file parsing module, which is used to parse files.

[0018] As a preferred solution of the intelligent private placement net value management system described in the present invention, wherein: the specific steps of the scraping module are as follows:

[0019] S11: Support users to select which folder in the personal mailbox to obtain;

[0020] S12: Support users to set the automatic synchronization time, and also support manual triggering of synchronization;

[0021] S13: Identify the truly needed files and truly private files through a large model.

[0022] As a preferred solution of the intelligent private placement net value management system described in the present invention, wherein: the specific steps of S13 are as follows:

[0023] S131: Collect the most comprehensive net value files on the market as training data;

[0024] S132: Train a multi-modal large model to enable it to have the ability to identify the specific data location according to the file;

[0025] S133: According to the trained large model, automatically judge the newly arrived files to judge whether the file is a single date or multiple dates. If it is judged to be multiple dates, it will judge whether it is horizontally arranged or vertically arranged, and find the location of the corresponding data.

[0026] As a preferred solution of the intelligent private placement net value management system described in the present invention, wherein: the specific steps of the matching module are as follows:

[0027] S21: Support users to set rule matching, including email subject, file title, sender;

[0028] S22: Support large model matching, and automatically identify and match according to the file name, email content and product name;

[0029] The specific steps of S22 are as follows:

[0030] S221: Input few-shot prompt words to the model;

[0031] S222: Let the large model automatically match the input information with the product information in the database to find the corresponding product.

[0032] As a preferred solution of the intelligent private placement net value management system described in the present invention, wherein: the specific steps of the parsing file module are as follows:

[0033] S31: For PDF and picture files, identify them as excel files through OCR;

[0034] S32: For excel files, use rules to judge whether they are net value date, unit net value, cumulative net value, virtual net value, and share according to the situation of words such as date, net value, cumulative, virtual, and share included in the fields in the file;

[0035] S33: For excel files, use the large model to judge whether the fields are net value date, unit net value, cumulative net value, virtual net value, and share according to the fields in the file;

[0036] S34: For excel files, manually directly set the corresponding relationship between the fields in the file and the net value date, unit net value, cumulative net value, virtual net value, and share, and judge whether the set corresponding relationship can be applied to the automatic recognition of other subsequent excel files of this product, or only applied to the current excel file;

[0037] S35: For excel files, read the corresponding data in the excel file according to the positions of the corresponding fields of the found net value date, unit net value, cumulative net value, virtual net value, and share;

[0038] The specific steps of S33 are as follows:

[0039] S331: The large model tells the model the content of each cell and the fields to be extracted through few-shot prompting engineering;

[0040] S332: Let the model judge the specific meaning corresponding to the cell.

[0041] As a preferred solution of the intelligent private placement net value management system described in the present invention, wherein: the data verification includes:

[0042] Field missing module, used to check whether there is a phenomenon of missing fields in the data;

[0043] Data inconsistency module, used to check whether there is a phenomenon of data inconsistency in the data;

[0044] Data missing module, used to check whether there is a phenomenon of data missing in the data;

[0045] The specific steps of the missing field module are as follows:

[0046] S41: For the file to be recognized as the net value of an ordinary manager, when there are fields in the date, unit net value, and cumulative net value that are not recognized, prompt for missing fields;

[0047] S42: For the file to be recognized as the virtual net value, when the date, virtual net value, and share are empty, prompt for missing fields;

[0048] S43: For those with missing fields, manually directly set the corresponding relationships between the fields in the file and the net value date, unit net value, cumulative net value, virtual net value, and share;

[0049] The specific steps of the data inconsistency module are as follows:

[0050] S51: Compare the unit net value, cumulative net value, virtual net value, and share read from the current excel with the unit net value, cumulative net value, virtual net value, and share on the same date that were once confirmed historically. When there are inconsistencies or some dates are empty, prompt for anomalies;

[0051] S52: When there are inconsistencies, automatically prompt the user for anomalies;

[0052] S53: Preview the file, and at the same time modify, add, delete, and correct data errors;

[0053] The specific steps of the data missing module are as follows:

[0054] S61: Set the data missing judgment rules, daily, weekly, monthly disclosures, and the delayed disclosure time;

[0055] S62: According to the current date and the disclosure rules, prompt for missing net values.

[0056] As a preferred solution of the intelligent private placement net value management system described in the present invention, wherein: the data processing module includes:

[0057] The restored net value calculation module is used for the calculation of the restored net value;

[0058] The downsampled net value calculation is used for the calculation of the downsampled net value;

[0059] The net value calculation after deducting performance fees is used for the calculation of the net value after deducting performance fees;

[0060] The specific steps of the restored net value calculation module are as follows:

[0061] S71: Perform retroactive dividend calculation;

[0062] S72: Use the deduced dividends to calculate the restored net value;

[0063] The calculation steps of S71 are as follows:

[0064] S711: Among them, cumnav t is the cumulative net value, and nav t is the unit net value;

[0065] S712: cumD 2,t = cumnav t - nav t * cumS 2,t ;

[0066] S713: If cumS 2,t is not 1, and if there is no problem with the manager's data, it means that there has been a dividend or split. At this time, if cumS 2,t of the previous data and the next data are both 1, then cumS 2,t will directly use the next cumS 2,t for filling, and cumD 2,t will also directly use the next cumD 2,t for filling;

[0067] S714: According to the filled data, judge whether cumS 2,t are all 1. If so, it means that there has only been a dividend and no split;

[0068] S715: The dividend D 2,t on the day = cumD 2,t - cumD 2,t-1 ;

[0069] The calculation steps of S72 are as follows:

[0070] S721: The restoration factor

[0071] S722: The cumulative restoration factor cumf t = ∏f t ;

[0072] S723: The restored net value

[0073] The specific steps of the downsampled net value calculation are as follows:

[0074] S81: Downsample the daily frequency to the weekly frequency to enable the use of data from this Wednesday to next Tuesday while being compatible with data not announced on Friday, and retain the data closest to Friday;

[0075] S82: Reduce the daily frequency to weekly frequency to back-calculate the missing net value based on the return rate of the same-type strategy index for the missing net value dates.

[0076] S83: Reduce the daily frequency to monthly frequency to use the data from the 25th of each month to the 3rd of the next month, and retain the last one closest to the end of the month.

[0077] S84: Reduce the daily frequency to monthly frequency to back-calculate the net value of the missing dates based on the return rate of the same-type strategy index for the missing net value dates.

[0078] The specific steps of the above S82 are as follows:

[0079] S821: Assume that the consecutive net values of the fund to be filled are missing for t = 1, 2, and there are net values for t = 0, 3.

[0080] S822: Assume that the return rate of the index to which the fund belongs is The total return is The total return of the 3 weeks of the fund to be filled is Then the return of the first week of the fund to be filled

[0081] The specific steps of the above S84 are as follows:

[0082] S841: Assume that the consecutive net values of the fund to be filled are missing for t = 1, 2, and there are net values for t = 0, 3.

[0083] S842: Assume that the return rate of the index to which the fund belongs is The total return is The total return of the 3 months of the fund to be filled is Then the return of the first month of the fund to be filled

[0084] The specific steps of the net value calculation for deducting performance fees are as follows:

[0085] S91: Automatically maintain the net value using the virtual net value extracted from the email.

[0086] S92: Automatically maintain the net value using the virtual net value set for the fee deduction.

[0087] The specific steps of the above S92 are as follows:

[0088] S921: Calculate the return rate using the adjusted net value data

[0089]

[0090] P t is the adjusted net value of the current day;

[0091] P0 is the net value on the first day of the stage where the deduction net value needs to be calculated;

[0092] S922: Performance compensation is accrued according to the annualized return rate gradient:

[0093] Calculate the return rate using the restored net value data:

[0094]

[0095] The calculation method for calculating the annualized return rate:

[0096]

[0097] P t is the daily unit restored net value;

[0098] P 0 is the unit restored net value on the first day of the stage where the deduction net value needs to be calculated;

[0099] Among them,

[0100] is the first-tier rate;

[0101] is the second-tier rate;

[0102] S923: Performance compensation is accrued according to the excess annualized return rate gradient:

[0103] Repeat the above steps and use the excess annualized return rate to judge during this process.

[0104] As a preferred solution of the intelligent private placement net value management system described in the present invention, where: The warning module includes:

[0105] An index calculation module for calculating the index;

[0106] An index scoring module for manually setting interval grading according to the calculated index value, or sorting within a certain range to determine the score, or sorting first and then scoring to make the index into a more comparable index value;

[0107] An index combination module for combining multiple different indexes with weights to form a more comprehensive index;

[0108] A net value / style warning module for using the calculated index, setting a threshold according to the original index value or the scored value, or the combined index, and giving a prompt when the warning is triggered to monitor the obvious fluctuations and abnormal changes in the net value.

[0109] Compared with the prior art:

[0110] 1. Automatically complete file matching and file parsing, and intelligently compatible with most net value files of various formats in the market, greatly improving the efficiency of obtaining the original net value;

[0111] 2. Automatically implement data verification in multiple ways, significantly improving the data quality;

[0112] 3. Realize data restoration rights through the reverse deduction method, and give the restored net value without relying on separate dividend split data;

[0113] 4. Complete the net value by referring to the trend of the same strategy index, improving the integrity of the data on the premise of ensuring data rationality;

[0114] 5. Automatically deduct fees from the net value, support automatic calculation of performance fees under multiple performance fee calculation rules, and the data more accurately reflects the post-investment performance after deducting performance fees;

[0115] 6. Automatically give early warnings for the net value to prevent abnormal performance and style drift, and use many exclusive indicators. BRIEF DESCRIPTION OF THE DRAWINGS

[0116] Figure 1 It is a schematic flow chart of the present invention.

[0117] Figure 2 It is the decomposition of the schematic flow chart of the present invention Figure 1 。

[0118] Figure 3 It is the decomposition of the schematic flow chart of the present invention Figure 2 。

[0119] Figure 4 It is the decomposition of the schematic flow chart of the present invention Figure 3 。 DETAILED DESCRIPTION OF THE INVENTION

[0120] To make the objectives, technical solutions and advantages of the present invention clearer, the following will further describe the embodiments of the present invention in detail with reference to the accompanying drawings.

[0121] The present invention provides an intelligent private equity net value management system. Please refer to Figures 1 to 4 ;

[0122] It includes: a product creation module, which can retrieve the created products after entering product information to avoid duplicate entry and management; a file acquisition module, which is used to grab emails from the mailbox and match them to products for parsing; data verification, which is used to check the data to see if there are phenomena such as missing fields, inconsistent data, and missing data; a data processing module, which is used to calculate the restored net value, downsampled net value, and net value after deducting performance compensation of the data; and a warning module, which can give warnings based on the calculated indicators after calculating the indicators.

[0123] The product creation module includes: an entry module, which is used to enter the basic information of the product, and the basic information includes the filing code, product name, manager, and establishment date; a retrieval module, which gives priority to retrieving the previously created products when creating a new product. If a retrieval result exists, it will be used preferentially to avoid duplicate entry and management.

[0124] The file acquisition module includes: a grabbing module, which is used to automatically grab the subject, body, and attachments of emails from the mailbox and also supports manual upload by the user; a matching module, which is used to match the files to products; and a file parsing module, which is used to parse the files.

[0125] The specific steps of the grabbing module are as follows:

[0126] S11: Support the user to select which folder in the personal mailbox to obtain files from;

[0127] S12: Support the user to set the automatic synchronization time and also support manual triggering of synchronization;

[0128] S13: Identify the truly needed files and truly private files through a large model.

[0129] The specific steps of S13 are as follows:

[0130] S131: Collect the most comprehensive net value files in the market as training data;

[0131] S132: Train a multi-modal large model to enable it to have the ability to identify the specific data location according to the file;

[0132] S133: Automatically judge the newly arrived files according to the trained large model to determine whether the file is for a single date or multiple dates. If it is judged to be for multiple dates, it will further determine whether it is horizontally arranged or vertically arranged and find the corresponding data location.

[0133] The specific steps of the matching module are as follows:

[0134] S21: Support the user to set rule matching, including email subject, file title, and sender;

[0135] S22: Support large model matching, and automatically identify and match according to the file name, email content, and product name;

[0136] The specific steps of S22 are as follows:

[0137] S221: Input few-shot prompt words into the model;

[0138] S222: Let the large model automatically correspond the input information to the product information in the database and find the corresponding product.

[0139] The specific steps of the file parsing module are as follows:

[0140] S31: For PDF and picture files, identify them as excel files through OCR;

[0141] S32: For excel files (excel files obtained from OCR recognition or directly), use rules to judge whether they are net worth dates, unit net worths, cumulative net worths, virtual net worths, and shares according to the situation of words such as date, net worth, cumulative, virtual, and share in the fields of the file (find the position of the row where the net worth date is located, extract all fields in this row, and then loop through each field to judge the positions of the rows and columns where the date, unit net worth, and cumulative net worth are located);

[0142] S33: For excel files, use the large model to judge whether the fields are net worth dates, unit net worths, cumulative net worths, virtual net worths, and shares according to the fields in the file;

[0143] The specific steps of S33 are as follows:

[0144] S331: Through few-shot prompt word engineering, the large model tells the model the content of each cell and the fields to be extracted;

[0145] S332: Let the model judge the specific meaning corresponding to the cell;

[0146] S34: For excel files, manually directly set the corresponding relationship between the fields in the file and the net worth date, unit net worth, cumulative net worth, virtual net worth, and share, and set whether the corresponding relationship can be applied to the automatic recognition of other subsequent excel files of this product, or only applied to the current excel file;

[0147] S35: For excel files, read the corresponding data in the excel file according to the positions of the corresponding fields of the found net worth date, unit net worth, cumulative net worth, virtual net worth, and share;

[0148] Data verification includes: a field missing module for checking whether there is a phenomenon of missing fields in the data; a data inconsistency module for checking whether there is a phenomenon of data inconsistency in the data; and a data missing module for checking whether there is a phenomenon of data missing in the data.

[0149] The specific steps of the field missing module are as follows:

[0150] S41: For a file to be recognized as the net value of an ordinary manager, when there are fields that cannot be recognized in the date, unit net value, and cumulative net value, a field missing prompt is given;

[0151] S42: For a file to be recognized as a virtual net value, when the date, virtual net value, and share are empty, a field missing prompt is given;

[0152] S43: For those with missing fields, manually directly set the corresponding relationship between the fields in the file and the net value date, unit net value, cumulative net value, virtual net value, and share;

[0153] The specific steps of the data inconsistency module are as follows:

[0154] S51: Compare the unit net value, cumulative net value, virtual net value, and share read from the current excel with the unit net value, cumulative net value, virtual net value, and share on the same date that were previously confirmed. When there is an inconsistency or when some dates are empty, an abnormality prompt is given;

[0155] S52: When there is an inconsistency, automatically prompt the user of the abnormality;

[0156] S53: Preview the file and at the same time modify, add, delete, and correct data errors;

[0157] The specific steps of the data missing module are as follows:

[0158] S61: Set data missing judgment rules, daily, weekly, monthly disclosures, and delayed disclosure times;

[0159] S62: According to the current date and disclosure rules, prompt for missing net values; if the set rule is daily, refer to the trading day data of the Shanghai Stock Exchange. When trading on the exchange but there is no net value on the current day, a prompt is given; if the set delay is three days for disclosure, a prompt is given three days later, and the missing within three days is regarded as normal; if the set rule is weekly, then check whether there is a net value on the last trading day of each week on the Shanghai Stock Exchange. If not, an abnormality prompt is given; if the set rule is monthly, then check whether there is a net value on the last trading day of each month on the Shanghai Stock Exchange. If not, an abnormality prompt is given.

[0160] The data processing module includes: a restored net value calculation module for calculating the restored net value; a downsampled net value calculation module for calculating the downsampled net value; and a net value calculation after deducting performance fees for calculating the net value after deducting performance fees.

[0161] The specific steps of the restored net value calculation module are as follows:

[0162] S71: Perform retroactive dividend calculation;

[0163] The calculation steps of S71 are as follows:

[0164] S711: Among them, cumnav t is the cumulative net value, and nav t is the unit net value (Purpose: Calculate the cumulative split coefficient; at the same time, determine whether there has been a dividend split in this item. If it is not 1, it means a dividend split has occurred; at the same time, also determine whether the next item is still affected. If the next item is not 1, that is, if it is still affected, then it is a split. If it is no longer affected, it is a dividend);

[0165] S712: cumD 2,t = cumnav t - nav t * cumS 2,t (Purpose: Calculate the cumulative dividend coefficient);

[0166] S713: If cumS 2,t is not 1, and if there is no problem with the manager's data, it means that there has been a dividend or split. At this time, if the cumS 2,t of the previous data and the next data are both 1, then cumS 2,t will directly use the next cumS 2,t for filling, and cumD 2,t will also directly use the next cumD 2,t for filling;

[0167] S714: According to the filled data, judge whether cumS 2,t is all 1. If so, it means that only dividends have occurred and no splits have occurred;

[0168] S715: The dividend D 2,t on the day = cumD 2,t - cumD 2,t-1 ;

[0169] S72: Use the deduced dividend to calculate the restored net value;

[0170] The calculation steps of S72 are as follows:

[0171] S721: The restoration factor

[0172] S722: Cumulative adjusted factor cumf t = ∏f t ;

[0173] S723: Adjusted net value

[0174] The specific steps for calculating the downsampled net value are as follows:

[0175] S81: Downsample the daily frequency to weekly frequency to enable the use of data from this Wednesday to next Tuesday and retain the data closest to Friday when the compatible data is not announced on Friday;

[0176] S82: Downsample the daily frequency to weekly frequency to infer the missing net value based on the return rate of the same - type strategy index for net value dates with missing values;

[0177] The specific steps of S82 are as follows:

[0178] S821: Assume that the consecutive net values of the fund to be filled are missing for t = 1, 2 and there are net values for t = 0, 3;

[0179] S822: Assume that the return rate of the index to which the fund belongs is The total return is The total return of the 3 - week fund to be filled is Then the return of the first week of the fund to be filled For ease of understanding, the above is the logarithmic return. Of course, ordinary return can also be used;

[0180] S83: Downsample the daily frequency to monthly frequency to use the data from the 25th of each month to the 3rd of the next month and retain the closest one to the end of the month;

[0181] S84: Downsample the daily frequency to monthly frequency to infer the net value of the missing date based on the return rate of the same - type strategy index for net value dates with missing values;

[0182] The specific steps of S84 are as follows:

[0183] S841: Assume that the consecutive net values of the fund to be filled are missing for t = 1, 2 and there are net values for t = 0, 3;

[0184] S842: Assume that the return rate of the index to which the fund belongs is The total return is The total return of the 3 - month fund to be filled is Then the return of the first month of the fund to be filled The above is the logarithmic return. Of course, ordinary return can also be used;

[0185] The specific steps for calculating the net performance compensation are as follows:

[0186] S91: Automatically maintain the net value using the virtual net value extracted from the email;

[0187] S92: Automatically maintain the net value using the virtual net value set for the deduction;

[0188] The specific steps of S92 are as follows:

[0189] S921: Calculate the rate of return using the post - adjusted net value data

[0190]

[0191]

[0192] P t is the post - adjusted net value of the current day;

[0193] P 0 is the net value on the first day of the stage for which the net value for deduction needs to be calculated.

[0194] S922: The performance compensation is accrued in gradients according to the annualized rate of return:

[0195] Calculate the rate of return using the post - adjusted net value data:

[0196]

[0197] The calculation method (simple interest) for calculating the annualized rate of return:

[0198]

[0199] P t is the unit post - adjusted net value of the current day;

[0200] P 0 is the unit post - adjusted net value on the first day of the stage for which the net value for deduction needs to be calculated;

[0201]

[0202] Among them,

[0203] is the first - tier rate;

[0204] is the second - tier rate;

[0205] S923: The performance compensation is accrued in gradients according to the excess annualized rate of return:

[0206] Repeat the above steps and use the excess annualized rate of return to judge during this process.

[0207] The early warning module includes: an index calculation module for calculating indexes; an index scoring module for manually setting interval grading according to the calculated index values, or sorting within a certain range to determine the score, or sorting first and then scoring to make the indexes into more comparable index values; an index combination module for combining multiple different indexes with weights to form a more comprehensive index; a net value / style early warning module for using the calculated indexes, setting thresholds according to the original index values or scored values, or combined indexes, and giving a prompt when the early warning is triggered to monitor obvious fluctuations and abnormal changes in the net value.

[0208] Among them, assuming the set thresholds are: >, <, <=, >=, ==, in the interval, such as fluctuations or maximum drawdown, a net value early warning module will be formed; assuming the set thresholds are: >, <, <=, >=, ==, in the interval, such as equity exposure, campis style exposure, a style early warning module will be formed.

[0209] Regarding the index calculation module: Commonly used market indexes can be selected (interval return rate, interval excess return rate (arithmetic), interval excess return rate (geometric), annualized return rate, annualized excess return rate (arithmetic), annualized excess return rate (geometric), ratio to the highest net value, number of weeks without a new high, 60-trading-day rolling maximum return, return win rate, excess return win rate, volatility, annualized volatility, upward volatility, annualized upward volatility, downward volatility, annualized downward volatility, maximum single-week decline, ratio of declining weeks, tracking error, annualized tracking error, maximum interval increase, maximum interval drawdown, number of weeks in the maximum drawdown interval, number of weeks to repair the maximum drawdown, historical 5% VaR, historical 5% CvaR, longest consecutive declining weeks, longest consecutive rising weeks, skewness, kurtosis, alpha, annualized alpha, beta, goodness of fit, residual risk, annualized residual risk, information ratio, Sharpe ratio, annualized Sharpe ratio, M2, Sortino ratio, annualized Sortino ratio, Treynor ratio, Jensen index, annualized Jensen index, Calmar ratio, stock selection ability (C-L), market timing ability (C-L), stock selection ability (H-M), market timing ability (H-M), stock selection ability (TM model), market timing ability (TM model), equity style exposure, campisi style exposure), data frequency, calculation interval, index benchmark, parameters before and after deduction, and the net value will be automatically updated as the data is updated.

[0210] Exclusive market indexes can also be selected.

[0211] For example, the investment ability in a bear market represents the fund's investment ability in a poor market environment and can issue a warning when a decline occurs. In actual calculation, the indices of each market are used as the basis for judging the market state, and the median judgment method is adopted. Each time RET_DOWN is calculated, the fund returns for the previous 12 months are selected, the returns in the UP months are recorded as 0, and the cumulative return rate for 12 months is calculated using a 12-month sliding window. Only when there are returns for 8 months are they calculated, that is

[0212]

[0213] Although the present invention has been described with reference to the embodiments above, various improvements can be made to it and its components can be replaced with equivalents without departing from the scope of the present invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments of the present invention can be combined with each other in any way. The exhaustive description of these combinations is not given in this specification only for the sake of saving space and resources. Therefore, the present invention is not limited to the specific embodiments disclosed in the text, but includes all technical solutions falling within the scope of the claims.

Claims

1. Intelligent private equity net value management system, characterized by: include: The product creation module is used to search for created products after entering product information, avoiding repeated entry and management; The file acquisition module is used to grab emails from the mailbox and match them to the product for parsing; Data verification is used to check the data to see if there are any missing fields, inconsistent data, or missing data. The data processing module is used to calculate the net value of the data after re-weighting, reducing the frequency of the net value, and calculating the net value after deducting performance compensation; The early warning module is used to issue early warnings based on the calculated indicators after the indicators are calculated.

2. The intelligent private equity net value management system according to claim 1 is characterized in that: The product creation module includes: An input module is used to input basic information of the product, including the registration code, product name, manager, and establishment date; The search module is used to prioritize previously created products when creating new products. If the search exists, it will be used first to avoid duplicate entry and management.

3. The intelligent private equity net value management system according to claim 1 is characterized in that: The file acquisition module includes: The capture module is used to automatically capture the title, body and attachments of emails from the mailbox, and also supports manual upload by users; A matching module, used to match files to products; The file parsing module is used to parse files.

4. The intelligent private equity net value management system according to claim 3 is characterized in that: The specific steps of the grabbing module are as follows: S11: Support users to choose which folder in their personal mailbox to retrieve from; S12: Supports users to set the time for automatic synchronization and also supports manual triggering of synchronization; S13: Use the big model to identify the files that are really needed and the truly private files.

5. The intelligent private equity net value management system according to claim 4 is characterized in that: The specific steps of S13 are as follows: S131: By collecting the most comprehensive net worth documents on the market as training data; S132: Train a large multimodal model to enable it to identify specific data locations based on files; S133: Based on the trained large model, the newly arrived file is automatically judged to determine whether the file is a single date or multiple dates. If it is determined to be multiple dates, it will be determined whether it is arranged horizontally or vertically, and the location of the corresponding data will be found.

6. The intelligent private equity net value management system according to claim 3 is characterized in that: The specific steps of the matching module are as follows: S21: Supports users to set matching rules, including email subject, file title, and sender; S22: Supports large model matching, automatically identifying and matching based on file name, email content and product name; The specific steps of S22 are as follows: S221: input a few-shot prompt word to the model; S222: Let the big model automatically match the input information with the product information in the database to find the corresponding product.

7. The intelligent private equity net value management system according to claim 3 is characterized in that: The specific steps of the file parsing module are as follows: S31: For PDF and image files, they are recognized as Excel files through OCR; S32: For the Excel file, the rules are used to determine whether the fields in the file contain the words date, net value, cumulative, virtual, and share, and whether it is the net value date, unit net value, cumulative net value, virtual net value, or share; S33: For the excel file, based on the fields in the file, use the big model to determine whether the field is the net value date, unit net value, cumulative net value, virtual net value, or share; S34: For Excel files, manually set the corresponding relationship between the fields in the file and the net value date, unit net value, cumulative net value, virtual net value, and share, and set whether the corresponding relationship can be applied to the automatic recognition of other Excel files of this product in the future, or only applied to the current Excel file; S35: For the excel file, according to the positions of the corresponding fields of the net value date, unit net value, cumulative net value, virtual net value, and share found, the corresponding data in the excel file is read; The specific steps of S33 are as follows: S331: The large model uses a few-shot prompt word project to tell the model the content of each cell and the fields that need to be extracted; S332: Let the model determine the specific meaning of the cell.

8. The intelligent private equity net value management system according to claim 1 is characterized in that: The data verification includes: Missing field module, used to check whether there are missing fields in the data; The data inconsistency module is used to check whether there is any inconsistency in the data; Missing data module, used to check whether there is missing data in the data; The specific steps of the field missing module are as follows: S41: For the file to be identified as the general manager's net worth, when there are unidentified fields for date, unit net worth, and cumulative net worth, a prompt will be displayed that the field is missing; S42: For a file to be identified as virtual net value, when the date, virtual net value, and share are empty, it is indicated that the field is missing; S43: For missing fields, manually set the corresponding relationship between the fields in the file and the net value date, unit net value, cumulative net value, virtual net value, and share; The specific steps of the data inconsistency module are as follows: S51: Compare the unit net value, cumulative net value, virtual net value, and share currently read from Excel with the unit net value, cumulative net value, virtual net value, and share of the same date that have been confirmed in history. If there is inconsistency or some dates are empty, an exception is prompted; S52: When inconsistency occurs, the user is automatically prompted with the abnormality; S53: Preview files, modify, add, delete, and correct data errors; The specific steps of the data missing module are as follows: S61: Set data missing judgment rules, daily, weekly, monthly disclosure, and delayed disclosure time; S62: Based on today’s date and disclosure rules, the missing net value is indicated.

9. The intelligent private equity net value management system according to claim 1, characterized in that: The data processing module includes: The right-adjusted net value calculation module is used to calculate the right-adjusted net value; Frequency reduction net value calculation, used for frequency reduction net value calculation; Calculation of net value after deduction of performance remuneration, used for calculation of net value after deduction of performance remuneration; The specific steps of the weighted net value calculation module are as follows: S71: Perform reverse dividend calculation; S72: Calculate the adjusted net value using the derived dividends; The calculation steps of S71 are as follows: S711: Among them, cumnav t is the cumulative net value, nav t is the unit net worth; S712:cumD 2,t =cumnav t -nav t *cumS 2,t ; S713: If cumS 2,t If it is not 1, if there is no problem with the manager's data, it means that a dividend or split has occurred. At this time, if the cumS of the previous data and the next data 2,t When both are 1, cumS 2,t Use the next cumS directly 2,t Fill, cumD 2,t Also directly use the next cumD 2,t filling; S714: Determine cumS according to the data after filling 2,t Are all 1s? If so, it means that only dividends have occurred, and no splits have occurred; S715: Dividend D for the day 2,t =cumD 2,t -cumD 2,t-1 ; The calculation steps of S72 are as follows: S721: Adjustment Factor S722: Cumulative weighting factor cumf t =∏f t ; S723: Adjusted net value The specific steps of calculating the net frequency reduction value are as follows: S81: Reduce the daily frequency to a weekly frequency so that when compatible data is not released on Friday, the data from Wednesday to Tuesday of this week will be used, and the data closest to Friday will be retained; S82: Reduce the daily frequency to weekly frequency, so that for the missing net value dates, the missing net value can be inferred based on the return rate of the index of similar strategies; S83: Reduce the daily frequency to a monthly frequency to use the data from the 25th of each month to the 3rd of the next month to keep the one closest to the end of the month; S84: Reduce the daily frequency to monthly frequency, so that for the missing net value date, the net value of the missing date can be inferred based on the return rate of the index of similar strategies; The specific steps of S82 are as follows: S821: Assume that the net value of the fund to be filled is missing at t=1,2 and needs to be filled, and has a net value at t=0,3; S822: Assume that the rate of return of the index to which the fund belongs is, The total revenue is The total return of the fund to be filled in the past three weeks is Then the first week's income of the fund to be filled The specific steps of S84 are as follows: S841: Assume that the fund to be filled has a missing net value at t=1,2 and needs to be filled, and has a net value at t=0,3; S842: Assume that the rate of return of the index to which the fund belongs is, The total revenue is The total return of the fund to be filled in March is Then the first month's income of the fund to be filled The specific steps for calculating the net value after deducting performance compensation are as follows: S91: Automatically maintain net value using virtual net value extracted from emails; S92: Automatically maintain the net value by using the virtual net value set by the deduction; The specific steps of S92 are as follows: S921: Calculate the rate of return using the adjusted net value data P t The adjusted net value for the day; P0 is the net value on the first day of the net value deduction period; S922: Performance compensation is calculated based on the annualized rate of return: Calculate the rate of return using the adjusted net value data: The calculation method of annualized rate of return is: P t The adjusted net value of the unit on that day; P0 is the unit adjusted net value on the first day of the period for which the net value deduction needs to be calculated; in, The first-tier rate; The second tier rate; S923: Performance compensation is calculated based on the excess annualized rate of return: Repeat the above steps and use the excess annualized rate of return to make judgments in the process.

10. The intelligent private equity net value management system according to claim 1, characterized in that: The early warning module comprises: An indicator calculation module, used to calculate the indicator; The indicator scoring module is used to manually set intervals according to the calculated indicator values, or to sort within a certain range to determine the score, or to sort first and then score, so as to make the indicator more comparable. The indicator combination module is used to combine the weights of multiple different indicators to form a more comprehensive indicator; The net value / style warning module is used to use calculated indicators to set thresholds based on the original values ​​or scoring values ​​of the indicators, or combined indicators, and to give prompts when warnings are triggered to monitor obvious fluctuations and abnormal changes in the net value.

Images