Annuity investment manager evaluation method

Abstract

According to the annuity investor evaluation method provided by the invention, the accuracy and efficiency of the exhaustion regulation are improved by structuring the exhaustion regulation template, splitting the exhaustion regulation groups and the exhaustion regulation dimensions and automatically generating the exhaustion regulation qualitative index library. According to the method, multi-dimensional evaluation indexes such as quantitative and qualitative evaluation indexes and style and label evaluation indexes are formed, deep portraits of the investor, the investment manager and the pension product are constructed, and the selection standard of the pension investor is formed.

Description

Technical Field

[0001] The present invention relates to the technical field of investment management, and in particular to a method for evaluating annuity investment managers. Background Art

[0003] At present, in the process of entrusted management of pension annuities, the selection of suitable investment managers (hereinafter referred to as "investment managers") and the management capabilities and styles of investment managers will directly determine the investment performance of the annuity.

[0004] However, the existing methods for selecting investment managers and investment managers by trustees mainly evaluate them based on the dimensions of return and risk, which cannot fully reflect the real situation of investment managers. At the same time, traditional due diligence methods often rely on manual data collection and analysis, which are inefficient and highly subjective. In addition, there are also problems such as the lack of unified standards for the management of evaluation indicators and the incompleteness of the underlying data used for calculations. Summary of the invention

[0005] The purpose of this invention is to build a scientific, comprehensive and dynamic evaluation system for investment managers, investment managers and pension products through data support, comprehensive evaluation and efficient management.

[0006] In order to achieve the above-mentioned purpose of the invention, the technical solution of the present invention provides a method for evaluating annuity investment managers, comprising the following steps:

[0007] Due diligence documents are collected through questionnaires, interviews, etc., and systematically decomposed in multiple dimensions to form structured due diligence dimension metadata. The due diligence dimension metadata includes nine groups: basic company information, annuity investment business management and structure, annuity investment decision-making process, annuity investment business model and investment support, risk management, pension product style labels, investment manager information, fixed income investment manager performance and style, and equity investment manager performance and style;

[0008] The metadata group distribution characteristics of due diligence metadata are analyzed based on three dimensions: distribution uniformity, existence of extreme values, and the degree of influence of each metadata point in the due diligence metadata on the overall results;

[0009] When the metadata points are evenly distributed and have no extreme values, the arithmetic mean method is used to process multiple first metadata groups to form qualitative indicators; when the metadata points are discretely distributed and some metadata points have a greater impact on the overall result, the weighted average method is used to process multiple first metadata groups and second metadata groups to form qualitative indicators; when the third metadata group contains the first metadata group and the second metadata, and there is inequality in the third metadata group, the weighted summation method is used to process the metadata groups to form qualitative indicators; the metadata points in the first metadata group are evenly distributed, and the metadata points in the second metadata group are discretely distributed.

[0010] According to the basic data of the net value of daily planned portfolio position transactions, the basic data indicator library is parsed and managed from the attributes, dimensions, and application indicator classification tree. The attributes include business attributes and technical attributes, and the dimensions include calculation objects, application objects, calculation levels, time intervals, indicator units, data types, engine sources, and indicator types. The application indicator classification tree includes application object layer, indicator classification layer, and indicator detail classification layer.

[0011] Establish an indicator dimension matrix, generate an engine indicator set based on the basic engine indicators registered to the indicator platform by each computing engine, generate an application indicator set based on the application indicators with business significance derived from the engine indicators, and form a customized indicator plan based on the indicator set packaged by the application indicators of the indicator platform. Perform functional tests on the indicator plan that has been created to ensure the correctness of the results returned by the created indicator interface. Select portfolios / plans / investment managers / investment managers and their corresponding level indicators for flexible data analysis, including but not limited to analysis of detailed assets within the portfolio, and implement detailed level indicator analysis by selecting corresponding portfolios and detailed assets, as well as detailed related indicators. Unify indicator generation rules and business calibers to automatically generate quantitative indicators and verify them through the indicator browser. Indicators formed based on daily data, monthly frequency data, comparative indicators, benchmarking indexes, composite and average calculation results, indicators formed by interval correlation coefficients, indicators formed by similar rankings, and indicators formed by scenario analysis are processed to form quantitative indicators that meet business requirements;

[0012] Based on the combination calculation between qualitative and quantitative indicators, the calculation results are labeled according to the weights to generate multi-quantile ranking labels, weight value comparison labels, fuzzy matching labels, and investment manager style label libraries;

[0013] Based on the relationship between qualitative and quantitative indicators and the investment manager's style label library, the investment manager's basic situation, return characteristics, asset allocation capabilities, fixed income investment capabilities, equity investment capabilities, and risk management investment capabilities are evaluated through preset scoring rules and scoring templates to obtain the annuity investment manager evaluation results.

[0014] Preferably, when analyzing the due diligence dimension metadata, if corresponding information can be found in the due diligence table and the matching degree is greater than 90%, the due diligence information is used as a positioning qualitative indicator, and the positioning qualitative indicator is used to generate a qualitative indicator library.

[0015] Preferably, when analyzing the due diligence dimension metadata, if the due diligence information cannot be used directly as a qualitative indicator, the indicator value is qualitatively determined by manual judgment to obtain an artificial qualitative indicator, and the artificial qualitative indicator is used to generate a qualitative indicator library.

[0016] Preferably, when performing the distribution characteristic analysis, one or more of the frequency distribution histogram, median statistical indicators, variance and standard deviation are used for judgment. If the histogram presents a rectangular shape, the difference between the mean and the median is not large, and the variance or standard deviation is small, then the distribution is considered to be uniform, otherwise it is considered to be discrete.

[0017] Preferably, the preset scoring rules include: equivalence method, tiered scoring method, tiered weighted scoring method, ranking scoring method, reverse percentile ranking method, tiered + linear interpolation method, ranking + linear interpolation method, tiered / ranking + multi-linear interpolation method.

[0018] Preferably, the method further comprises comparing the quantitative indicator with the market index benchmark by adding a corresponding market index as a benchmark, and further analyzing the difference with the general market benchmark.

[0019] Preferably, the scoring template is designed according to time interval and object type.

[0020] The technical solution of the present invention proposes a method for evaluating annuity investment managers. Through a structured due diligence template, due diligence groups and due diligence dimensions are split, and a due diligence qualitative indicator library is automatically generated to improve the accuracy and efficiency of due diligence. This method forms a deep portrait of investment managers, investment managers and pension products by forming multi-dimensional evaluation indicators such as "quantitative + qualitative" and "style + label", forming a "selection standard" for pension investment managers. DETAILED DESCRIPTION

[0021] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0022] The embodiment of the present invention provides a method for constructing an annuity investment manager evaluation system, comprising the following steps:

[0023] Due diligence documents are collected through questionnaires, interviews, etc., and systematically decomposed into multiple dimensions to form structured due diligence dimension metadata. The due diligence dimension metadata includes nine major groups, including basic company information, annuity investment business management and structure, annuity investment decision-making process, annuity investment business model and investment support, risk management, pension product style labels, investment manager information, fixed income investment manager performance and style, equity investment manager performance and style, as well as several due diligence dimensions.

[0024] Analyze the distribution characteristics of due diligence metadata to determine whether it is evenly distributed, whether there are extreme values, and the degree of influence of each metadata point on the overall result. When the metadata values ​​are evenly distributed and there are no extreme values, the arithmetic mean method is used to process the metadata to form qualitative indicators; when the metadata values ​​are discrete and some data points have a greater degree of influence on the overall result, the weighted average method is used to process the metadata to form qualitative indicators; when the analysis dimension needs to be summarized by a group of metadata, and there is inequality in the group of metadata, the weighted summation method is used to process the metadata to form qualitative indicators.

[0025] (1) Weighted average: When the analysis indicator depends on a set of metadata (A n ), and (A n ) have large differences in the corresponding values, and need to be combined with another set of metadata (B n ) Use the weighted average processing method to calculate the average value, and use the average value as the corresponding index value. The formula is Such as A n = {average management scale of equity investment managers, average management scale of fixed income investment managers, average management scale of mixed investment managers}, corresponding representative B n ={number of equity managers, number of fixed income managers, number of mixed managers}, then the average managed portfolio size indicator = SUM(average management scale of equity investment managers*number of equity managers + average management scale of fixed income investment managers*number of fixed income managers + average management scale of mixed investment managers*number of mixed managers) / (number of equity managers + number of fixed income managers + number of mixed managers).

[0026] (2) Arithmetic mean: When the analysis indicator depends on a set of metadata (A n ), and (A n ) The corresponding numerical distribution is relatively uniform, and the average value can be calculated by arithmetic mean processing method, and the average value is used as the corresponding index value. The formula is For example, the average years of experience of investment research personnel in investment work = SUM (average years of experience of equity investment managers + average years of experience of fixed income investment managers + average years of experience of equity investment assistants + average years of experience of fixed income investment assistants + average years of experience of researchers) / 5.

[0027] (3) Weighted sum: When the analysis indicator depends on a set of metadata (An ) are summarized, and (A n ) are not equal, and the corresponding weight value (a n ), use the weighted summation method to calculate the result, and use the result as the corresponding index value. The formula is For example, the number of investment research personnel of an investment manager = number of investment managers*0.5+assistant investment managers*0.3+researchers*0.2.

[0028] (4) In addition to the above methods, positioning processing and manual processing methods can also be used. Positioning processing is applicable to cases where the corresponding information can be found in the due diligence table and the matching degree is greater than 90%, and the due diligence information can be directly used as the qualitative indicator value; manual processing is applicable to cases where the corresponding information can be found in the due diligence table and the matching degree is less than 90%, and the due diligence information cannot be directly used as the qualitative indicator value, and the indicator value is qualitatively determined by manual judgment.

[0029] According to the basic data of the net value of daily planned portfolio holdings, the basic data indicator library is parsed and managed from the attributes, dimensions, and application indicator classification tree.

[0030] (1) Define indicator attributes:

[0031] Indicator attributes are divided into business attributes and technical attributes;

[0032] Business attributes include: indicator number, indicator name, calculation object, application object, calculation level, indicator business classification, indicator formula, indicator definition, indicator unit, data precision, and display format;

[0033] Technical attributes include: data type, indicator parameters, indicator technical classification, indicator type, system to which it belongs, engine source, engine interface, engine indicator code, indicator link, and indicator verification.

[0034] (2) Define indicator dimensions:

[0035] The indicator dimensions are divided into calculation object, application object, calculation level, time interval, indicator unit, data type, engine source, and indicator type;

[0036] The calculation objects include: portfolios, composite portfolios, investment managers, investment managers, pension products, and trustees;

[0037] Application objects include: portfolios, plans, investment managers, investment managers, pension products, MOM portfolios;

[0038] The calculation levels include: combination, grouping, combination details, and grouping details;

[0039] Time intervals include: since this week, since this month, since this quarter, since this year, since last year, since the year before last, in the past week, in the past two weeks, in the past month, in the past three months, in the past six months, in the past year, in the past two years, in the past three years, in the past five years, today, and since establishment;

[0040] Indicator units include: default, percent, yuan, ten thousand yuan, million, ten million, hundred million yuan, day, month, year;

[0041] Data types include: numeric type, integer type, date type, and text type;

[0042] Engine sources include: default engine, custom engine, and third-party engine;

[0043] The indicator types include: engine indicators, extended indicators, expression indicators, and custom indicators.

[0044] (3) Define the application indicator classification tree:

[0045] The application indicators are divided into three layers: the first layer is the application object, the second layer is the indicator classification, and the third layer is the indicator detailed classification;

[0046] The application objects are divided into: pension products, trustees, investment managers, investment managers, plans, portfolios, and benchmarks;

[0047] Pension product indicators are divided into: basic information, performance, and performance evaluation;

[0048] Trustee indicators are divided into: statistics from the Ministry of Human Resources and Social Security;

[0049] The investment manager indicators are divided into: basic information, performance, performance evaluation, and statistics from the Ministry of Human Resources and Social Security;

[0050] Investment manager indicators are divided into: basic information, performance, performance evaluation, performance attribution, portfolio holdings, transaction statistics, market risk, and concentration risk;

[0051] The planned indicators are divided into: basic information, performance, performance evaluation, performance attribution, portfolio holdings, transaction statistics, market risk, and concentration risk;

[0052] The portfolio indicators are divided into: basic information, performance, performance evaluation, performance attribution, portfolio holdings, transaction statistics, market risk, and concentration risk;

[0053] Benchmark indicators are divided into: performance and performance evaluation.

[0054] Establish an indicator dimension matrix of "atomic indicators → basic indicators → derived indicators → application indicators → indicator schemes", unify indicator generation rules and business caliber to automatically generate quantitative indicators.

[0055] (1) Generate engine indicators. Each computing engine is registered to the basic engine indicators of the indicator platform;

[0056] (2) Generate application indicators, which are derived from engine indicators to form application indicators with business significance;

[0057] (3) Generate an indicator plan, and form a customized indicator plan through the indicator set packaged by the application indicators of the indicator platform;

[0058] (4) Indicator solution browser: performs functional testing on the created indicator solution to ensure the correctness of the return results of the created interface.

[0059] (5) Object indicator browser: select portfolio / plan / investment manager / investment manager and their corresponding level indicators for flexible data analysis.

[0060] (6) Detailed indicator browser: analyzes the detailed assets within the portfolio and implements detailed indicator analysis by selecting the corresponding portfolio and detailed assets, as well as detailed related indicators.

[0061] Based on the above standardized definition of indicators and after verification by the indicator browser, quantitative indicators of complex business requirements can be formed. Application indicators (quantitative indicators) = calculation object + application object + calculation level + time interval + indicator unit + data type. In actual use, the application object, calculation level, and time interval are combined with multiple fields and attributes. That is, assuming there are m fields, the first field has n_1 attribute values, the second field has n_2 attribute values, and the mth field has n_m attribute values, then the final indicators are arranged and combined, and the total number is n_1*n_2*...*n_m.

[0062] The quantitative indicator generation rules are described as follows:

[0063] (1) It is necessary to process based on daily data. For example, the basic indicator [Rate of Return under Trust] generates corresponding application indicators according to the rules: the calculation object is the investment manager, the application object is the MOM portfolio (including all portfolios, equity portfolios, job-year equity portfolios, enterprise-year equity portfolios, fixed income portfolios, job-year fixed income portfolios, and enterprise-year fixed income portfolios), the calculation level is the portfolio layer and the grouping layer, the time interval includes short-term (since this year, since this month) and long-term (nearly one year, nearly three years), and the indicator unit and data type are assigned according to the actual situation. According to the permutation and combination rules, the basic indicator [Rate of Return under Trust] can include 7*2*4 application indicator combinations, namely [Rate of Return of All Portfolios under Trust of Investment Manager since this year], [Rate of All Portfolios under Trust of Investment Manager since this month], ..., [Rate of Grouping Layer of Enterprise-year Fixed Income Portfolios under Trust of Investment Manager in the Past Three Years]. The interval rate of return is obtained by adding one to the daily rate of return and then subtracting one, that is, (1+r1)(1+r2)…(1+rn )-1, where r1, ...r n The daily rate of return.

[0064] (2) Based on the monthly frequency data, such as the basic indicator [equity timing ability], according to the application indicator rules, the application indicator [equity timing ability of all equity portfolios entrusted by the investment manager in the past three years] can be generated. This long-term indicator is different from the yield indicator. It uses monthly frequency data instead, that is, the average net equity position of all equity portfolios under the trust W(i) (monthly frequency) = the proportion of equity asset positions after hedging, the average net equity position of the equity portfolio of investment manager j in the same period W(i,j) (monthly frequency) = the proportion of equity asset positions after hedging, and the yield of the CSI 800 Index in the same period R(i), then the equity timing ability of the investment manager in the past three years R(j) = (W(i,j)-W(i))*R(i) is calculated.

[0065] (3) Indicators formed by comparing two sets of data, such as the [Similarity of Heavily-weighted Stocks] indicator, are formed by comparing two pairs of weighted combinations under the same investment manager (the number of comparable combinations is no less than 3). The top ten heavily-weighted stocks appearing at the same time are counted once. The cumulative sum is taken and the average of the two pairs of combinations is taken, which is then divided by 10 to obtain the similarity percentage value.

[0066] (4) Indicators formed based on composite and average calculation results, such as the [Investment Manager’s Equity Asset Ratio] indicator. Assume that the investment manager holds three portfolios, A, B, and C. The equity assets and total assets of portfolio A are RMB 100,000 and RMB 1 million, the equity assets and total assets of portfolio B are RMB 150,000 and RMB 4 million, and the equity assets and total assets of portfolio C are RMB 300,000 and RMB 6 million. If the average calculation method is adopted, then [the proportion of equity assets of the investment manager] = (equity assets of portfolio A / total assets of portfolio A + equity assets of portfolio B / total assets of portfolio B + equity assets of portfolio C / total assets of portfolio C) / number of portfolios = (10 / 100+15 / 400+30 / 600) / 3 = 6.25%; if the compound calculation method is adopted, then [the proportion of equity assets of the investment manager] = (equity assets of portfolio A + equity assets of portfolio B + equity assets of portfolio C) / (total assets of portfolio A + total assets of portfolio B + total assets of portfolio C) = (10+15+30 / 100+400+600) = 5%.

[0067] (5) Indicators based on interval correlation coefficients are mainly used for the correlation between the target indicator and the target index. The calculation formula is the sample covariance of the daily value of the target indicator and the daily value of the benchmark within the specified interval / (sample standard deviation of the daily value of the target indicator * sample standard deviation of the daily value of the target index). For example, the correlation coefficient of the growth style preference of pension product A in the past year is calculated using the CSI 800 Growth Index. The correlation coefficient of the growth style preference of pension product A in the past year ρ P,B = cov(RP ,R B ) / (σ P ·σ B ), where R P = Daily net value sample of pension product A in the past year, R B = Daily net value sample of CSI 800 Growth Index in the past year, cov(R P ,R B ) = sample covariance of the daily net value of pension product A and the daily net value of CSI 800 Growth Index in the past year, σ P = Standard deviation of the daily net value sample of pension product A in the past year, σ B =The standard deviation of the daily net value sample of CSI 800 Growth Index in the past year.

[0068] (6) Indicators based on rankings in the same category can more objectively judge the capabilities of investment managers and investment managers. The formula for calculating the ranking percentage = ranking among all samples / total number of samples * 100%. For example, for the indicator [investment manager’s equity-based return rate this year, the equity-based return rate of investment manager A this year = 15%, ranking 7th among all 22 investment managers. According to the standard calculation formula, the equity-based return rate of investment manager A this year = 7 / 22 * 100% = 31.82%.

[0069] (7) Indicators formed based on scenario analysis are used to screen out products with suitable returns and risk control under various market scenarios, and make analysis and screening. The return trend within a period of time can be defined as function y = f(x). If there is a fixed neighborhood of x0 where f(x0) ≥ f(x), and there is a fixed neighborhood of x1 where f(x1) ≤ f(x), then within the fixed interval, x0 is the maximum point of function f(x), and then it will enter the bear market interval; x1 is the minimum point of function f(x), and then it will enter the bull market interval; if it is currently at a non-stage extreme point, it will continue the bull-bear state of the previous day. For example, if the [Technology Industry Bull-Bear Market Scenario Indicator] is used as the judgment benchmark, if the current point is at the lowest or highest point of the previous three months, if it is currently at the stage minimum point, it will then enter the bull market interval; if it is currently at the stage maximum point, it will then enter the bear market interval; if it is currently at a non-stage extreme point, it will continue the bull-bear state of the previous day.

[0070] Generate a style label library for investment managers based on qualitative and quantitative indicators.

[0071] Define labels. By combining and calculating the indicators, the calculation results are labeled according to the weights to generate corresponding label items. The label generation rules are as follows:

[0072] Multi - quantile ranking - type label, used to represent the ranking information of the investment manager's ability in a certain field. Define N different values of a certain index as a data set, and its elements are arranged in ascending order. If the data set is divided into k equal parts (i.e., k quantiles), the position Q of the i - th quantile i can be calculated by the following formula: Q i = N*(i / k), where i is the index of the quantile, ranging from 1 to k. If the size N of the data set is not an integer multiple of k, then the calculated position needs to be rounded to determine the specific quantile value. In practical applications, the more commonly used quantile rankings are the 3 - quantile and 4 - quantile methods. If the 3 - quantile method is adopted, the first quantile (Q1, i.e., 33%),

[0073] the second quantile (Q2, i.e., 67%); if the 4 - quantile method is adopted, the first quantile (Q1, i.e., 25%),

[0074] the second quantile (Q2, i.e., 50%), and the third quantile (Q3, i.e., 75%). For example, for the label item

Investment Manager's Fixed - Income Investment Ability

Annualized Yield of Fixed - Income Assets in the Past 3 Years under Entrustment

[0075] Weight - value comparison - type label, representing the label information of the investment manager in a certain field based on the weight value. Use a membership function μ(x), whose value range is [0, 1], to represent the ability range of x in the field. When μ(x) is close to 1, it means that x has better ability; when μ(x) is close to 0, it means that x has weaker ability. Usually, during use, if the value range is not [0, 1], then normalization processing is required. The data is scaled between 0 and 1 by subtracting the minimum value of the data and dividing by the range of the data (the maximum value minus the minimum value). The formula is: X nor =(X - X min ) / (X max - X min ), where X is the original value, X nor is the normalized value, X min is the minimum value in the data set, and X max is the maximum value in the data set summary. For example, for the label item

Investment Manager Assessment Mechanism - Short - Term Performance Weight (Equity)

Short - Term Weight

Short - Term Weight

Short - Term Weight

Short - Term Weight

Short - Term Weight

[0076] Fuzzy matching tags can be assigned tags through fuzzy matching for some qualitative indicators in the due diligence table. The matching method can be loose matching or strict matching. Loose matching allows synonyms or near-synonyms of keywords and ignores the order between keywords; strict matching only accepts completely identical keywords and strictly requires the order between keywords. For example, the [Stop Loss Mechanism Setting] tag under the asset allocation dimension performs fuzzy matching through the results of the answers in the due diligence information. If the results contain keywords such as "has" or "exists", the tag value is [has]; if the results contain keywords such as "does not have" or "does not exist", the tag value is [does not have].

[0077] Based on the indicator library and label library, the investment manager evaluation standards are formed.

[0078] The evaluation criteria for investment managers are assessed from six dimensions: basic situation, return characteristics, asset allocation capabilities, fixed income investment capabilities, equity investment capabilities, and risk management investment capabilities.

[0079] The basic situation includes: list of company shareholders, company operating data, information on company executives and key decision makers, company department structure and human resources, annuity business management structure, annuity business personnel, annuity investment core personnel, investment manager turnover, performance evaluation mechanism, pension product analysis, the overall situation of the company's issued pension products, etc. The above information comes from due diligence information.

[0080] The income characteristics include: enterprise annuity management data of the Ministry of Human Resources and Social Security, entrusted management data - scale overview, entrusted management data - performance, performance stability. The above information is obtained through indicator calculation. The indicators cover the Ministry of Human Resources and Social Security's yield rate in different intervals, entrusted yield rate in different intervals, management scale, benchmark yield rate, etc.

[0081] Asset allocation capability includes: (1) annuity investment philosophy, investment decision-making mechanism, annuity investment business model, and annuity investment business positions. The above information comes from due diligence information; (2) asset allocation capability, allocation ratios of major asset classes, and market index trends. The above information is obtained through indicator calculations, which cover equity timing capability, equity position consistency, similarity of heavily weighted stocks, the proportion of issued stock / hybrid funds, the proportion of issued bond funds, and the proportion of various asset holdings.

[0082] Fixed income investment capabilities include: (1) entrusted performance - absolute performance, cumulative return trend of fixed income investment, entrusted performance - relative performance, risk analysis, monthly return performance of fixed income assets, performance in rising / falling markets, performance of fixed income investment managers, Campisi attribution analysis, interest rate bond investment, credit bond investment, fixed income + investment, cost allocation strategy, leverage strategy, transaction analysis, etc. The above information is obtained through indicator calculation, which covers yields in different intervals, annualized yields in different intervals, maximum drawdown, annualized volatility, annualized Sharpe ratio, Calmar ratio, benchmark interval yield, average monthly yield, average monthly excess yield, monthly winning rate, average monthly yield in rising markets, and average monthly excess yield in rising markets. , the winning rate of the rising market next month, Campisi attribution analysis, the maturity distribution of interest rate bonds, the duration of the portfolio, the number of interest rate bond transactions, the scale of interest rate bond transactions, the internal rating of the issuer, the internal rating of the bond, the proportion of credit bonds below BBB-level of the issuer, the proportion of credit bonds below BBB-level of the bond, the proportion of municipal investment bonds, the proportion of industrial bonds, the regional distribution of municipal investment bonds, convertible bond positions, convertible bond conversion premium rate, fixed income productization rate, trust product type, debt plan type, PPN and private placement bonds, leverage ratio, annualized turnover rate, etc.; (2) the style distribution of fixed income investment managers (in the past year). The above information is obtained through style tags, which include interest rate investment strategy, credit investment strategy, fixed income + investment strategy, leverage strategy, product investment strategy, and trading style investment strategy.

[0083] Equity investment capabilities include: entrusted performance-absolute performance, cumulative return trend of equity investment, entrusted performance-relative performance, risk analysis, monthly return performance of equity assets, performance in rising / falling markets, performance of equity investment managers, sector allocation, concentration of holdings in the top five industries, concentration of holdings in the top ten individual stocks, top ten heavily held stocks, investment style, growth / value / balanced ratio, holding PE, holding PB, profitability, large / medium / small cap ratio, changes in Hong Kong Stock Connect positions, distribution of Hong Kong stock investments, new stock issuance income statistics, and turnover rate.

[0084] The above information is obtained through indicator calculation, which covers annualized rate of return, cumulative rate of return within the period, maximum drawdown, annualized volatility, annualized Sharpe ratio, Calmar ratio, average monthly rate of return, average monthly excess rate of return, average monthly rate of return in rising market, average monthly excess rate of return in rising market, monthly winning rate in rising market, average monthly rate of return in falling market, average monthly excess rate of return in falling market, monthly winning rate in falling market, Brinson attribution, proportion of CITIC's seven major sectors, proportion of Shenwan's top five industries, concentration of top five industries, concentration of top ten stocks, concentration of top five industries, concentration of top ten stocks, balanced proportion of large, medium and small caps, growth / value / balanced proportion, PE, PB, profitability ROE, proportion of Hong Kong Stock Connect, proportion of Hong Kong Stock Connect funds, proportion of Hong Kong Stock Connect pensions, the number of new shares that can be subscribed offline in the entire market, the number of successful bids, hit rate, break-even rate, contribution of new shares to returns, average return and annualized turnover rate.

[0085] Risk management investment capabilities include: enterprise annuity management data of the Ministry of Human Resources and Social Security, entrusted management data, and risk statistics. The above information is obtained through indicator calculations, which cover annualized volatility, maximum drawdown, annualized Sharpe ratio, maximum drawdown, Calmar ratio, etc.

[0086] Generate scoring results based on the investment manager's scoring model.

[0087] The scoring model defines multiple levels of dimensions and associates indicators / label items, calculates the scoring details of each dimension according to different weights, and finally summarizes the scoring results of the analysis object. The specific implementation process is as follows:

[0088] Define dimensions and weights:

[0089] Index set: X = {x1, x2, ..., x n}, where x i represents the i-th indicator or label score;

[0090] Dimension set: D = {d1, d2, ..., d m}, where d j represents the jth dimension;

[0091] Indicator weight set: W = {w1,w2,...,w n}, where w i represents the scoring weight of the i-th indicator or label in the model, and

[0092] Dimension weight set (optional, if you need to consider the weight differences between dimensions): W D ={w d1 ,w d2 ,...,w dm}, where w djrepresents the weight of the jth dimension, and If the dimension weight does not need to be considered, the indicator weight under each dimension directly constitutes W.

[0093] Calculation based on indicators, dimensions, and weights:

[0094] If the dimension weight is not considered, the weighted average is used as the final scoring result.

[0095] If the dimension weight difference needs to be considered, the weighted average is calculated in each dimension first, and then the weighted average is calculated according to the dimension weight. For the jth dimension d j , which contains the index set X dj , the corresponding weight set is W dj (Here W dj is the corresponding d in W j The jth dimension score is The final score is then calculated based on the dimension weights

[0096] For different evaluation indicators, corresponding scoring rules are set. The indicator scoring rules are described as follows:

[0097]

[0098]

[0099]

[0100] 3. Generate the corresponding scoring template through the definition of the above scoring model. For example, according to the time interval, you can define interval templates such as this year, last year, the past year, the past two years, and the past three years to analyze long-term and short-term evaluation results; according to the object type, you can define equity, fixed income, and mixed templates for evaluation results between different objects.

[0101] The following is an example of the equity investment manager scoring model:

[0102] The first-level dimensions of the model cover management experience (15 points), profitability (20 points), configuration ability (15 points), stock selection ability (15 points), stability ability (15 points), and anti-fall ability (20 points).

[0103] Management experience includes evaluation indicators such as annuity management scale, investment manager experience, social security and basic pension fund investment experience, etc. The indicator processing methods include tiered weighted scoring method and equivalent method. Taking the annuity management scale indicator as an example, the total score of this indicator is 4 points, and the tiered weighted scoring method is used. The tiering rules are as follows: if the scale is greater than 5 billion, the score is 100% of the total score; if the scale is between 3 billion and 5 billion, the score is 90% of the total score; if the scale is between 2 billion and 3 billion, the score is 80% of the total score; if the scale is between 1 billion and 2 billion, the score is 70% of the total score; if the scale is less than 1 billion, the score is 60% of the total score. If the investment manager's annuity management scale is 3.543 billion, which is within the range of 3 billion to 5 billion, the score of the investment manager's annuity management scale indicator is 4*90%=3.6 points.

[0104] Profitability includes evaluation indicators such as the return on category assets in the past year and the annualized Sharpe ratio of category assets in the past year. The indicator processing method is ranking + linear interpolation. Taking the return on category assets in the past year as an example, the total score of this indicator is 10 points. The ranking + linear interpolation method is used, and the rules are as follows: if the return rate is in the top 5%, the score is 100% of the total score; if the return rate is in the bottom 5%, the score is 60% of the total score; if the return rate is between the top 5% and the bottom 5%, the linear difference is calculated according to the formula [0.6+0.4*(|AB|) / (|CB|), where A is the return on category assets of the investment manager in the past year, B is the return on category assets of the investment manager ranked in the bottom 5% in the past year, and C is the return on category assets of the investment manager ranked in the top 5% in the past year. If A=0.14, B=0.05, C=0.5, then the investment manager's score for the category asset return index in the past year is 10*[0.6+0.4*(|0.14-0.05|) / (|0.5-0.05|)=6.8 points.

[0105] The allocation capability covers evaluation indicators such as the monthly winning rate in the rising market in the past year, the monthly winning rate in the falling market in the past year; the stock selection capability covers evaluation indicators such as the Brinson attribution stock selection capability of equity assets in the past year, the stock selection capability of equity assets in the past year (HM model) and other evaluation indicators; the stability capability covers indicators such as the monthly winning rate of category assets in the past year, the sustainability of stock selection capability in the past year and other indicators; the anti-fall capability covers indicators such as the downside risk of category assets in the past year, the maximum drawdown of category assets in the past year and other indicators. The processing methods cover the above 12 processing types.

[0106] By using various processing methods on each indicator, the sum of the scores of each level 1 evaluation dimension is obtained, and then the total score of the evaluation object is obtained by adding up each level 1 evaluation dimension.

[0107] Based on the scoring results, multi-scenario, full-coverage screening functions for investment managers, investment managers, and pension products are provided.

[0108] General screening provides quick screening of commonly used indicators, such as interval performance ranking indicators, management scale, number of management combinations, etc.

[0109] Advanced filtering provides customized combination filtering of all indicators and all tags.

[0110] Each screening scheme is composed of multiple screening conditions, and then the final screening conditions are formed through expression logic operations.

[0111] Each filtering condition includes a condition group, an operator, and a condition value. The condition group comes from the indicator library and the label library. The indicator operator supports value, equal, not equal, greater than, greater than or equal to, less than, less than or equal to, and range. The label operator supports value, contain, and not contain. The condition value is the result corresponding to the selected indicator or label.

[0112] The expression supports expression operations for each filter condition. The supported operations include logical AND and logical OR.

[0113] An example of advanced screening: If you want to screen out all investment managers whose yield under trust in the past year is greater than 5% and whose ability to hit new stocks in the past year is in the top 1 / 3 of the comparison group, then you can add two custom screening conditions C1 and C2. The C1 condition group is the [yield under trust in the past year] indicator, the operator is [greater than], and the condition value is 5%; the C2 condition group is the [ability to hit new stocks in the past year] label, the operator is [contains], and the condition value is [top 1 / 3 of the comparison group]; through the logical AND && operator, C1 and C2 are operated on expressions to form an advanced screening solution.

[0114] Bubble screening provides any two-way combination of multiple indicators such as interval returns, annualized volatility, annualized Sharpe ratio, maximum drawdown, entrusted management scale, etc., to form a bubble chart, which more intuitively reflects the capabilities of investment managers and other analysis objects.

[0115] Provide flexible and scalable analysis report download.

[0116] 1. Define components. Generate atomic components of analysis objects through visual configuration. Component types support graphics, tables, and text boxes. Component data sources support custom interfaces, application databases, and indicator solutions.

[0117] 2. Define cards. Form analysis cards by flexibly assembling atomic components.

[0118] 3. Define report views. Based on the card view, a complete analysis report is formed for the six evaluation dimensions of investment managers, investment managers, and pension products. The analysis report supports flexible screening of card content information.

[0119] 4. Complete specific report downloads as needed.

[0120] The advantages of the embodiments of the present invention are as follows:

[0121] 1. Establish a pension annuity finance shared information database. Facing the problem of lack of public information data in the pension industry, we integrated internal business data of pension asset management and external information and due diligence data to form a pension finance shared information database, which provided solid data support for the evaluation and portrait construction of investment managers.

[0122] 2. A multi-dimensional evaluation system with pension management characteristics has been formed. Based on the established pension finance information database, multi-dimensional evaluation indicators such as "quantitative + qualitative", "style + label" have been formed to build an in-depth portrait of investment managers, investment managers and pension products, forming the "selection criteria" for pension investment managers.

[0123] 3. Create a flexible and scalable card-style visualization view. Through component-based layout and modular configuration, it supports custom view configuration through templates. It can build multi-dimensional investment analysis tools according to the needs of different business scenarios, control investment risks, and improve investment efficiency.

[0124] 4. A multi-level unified indicator management system has been established. In combination with the trustee MOM management business scenarios, the analysis indicators are hierarchically divided and standardized according to thematic scenarios such as investment managers, investment managers, and pension products. An indicator dimension matrix of "atomic indicators → basic indicators → derivative indicators → application indicators → indicator solutions" is established to build a unified indicator management system, which has achieved standardization of indicator definitions, quantifiable indicator calibers, flexible composite indicators, and interface-based indicator display. At the same time, it supports online publishing and version management of indicator and dimension definitions, and conducts regular reviews based on the frequency of use and business value of indicators.

Claims

1. A method for evaluating annuity investment managers, characterized in that: The following steps are involved: Due diligence documents are collected through questionnaires, interviews, etc., and systematically decomposed in multiple dimensions to form structured due diligence dimension metadata. The due diligence dimension metadata includes nine groups: basic company information, annuity investment business management and structure, annuity investment decision-making process, annuity investment business model and investment support, risk management, pension product style labels, investment manager information, fixed income investment manager performance and style, and equity investment manager performance and style; Analyze the metadata group distribution characteristics of due diligence metadata based on distribution uniformity, existence of extreme values, and the degree of influence of each metadata point in the due diligence metadata on the overall results; When the metadata points are evenly distributed and have no extreme values, the arithmetic mean method is used to process multiple first metadata groups to form qualitative indicators; When the metadata points are distributed discretely and some metadata points have a greater impact on the overall result, a weighted average method is used to process multiple first metadata groups and second metadata groups to form qualitative indicators; When the third metadata group includes the first metadata group and the second metadata, and there is inequality in the third metadata group, the metadata group is processed by weighted summation method to form a qualitative index; the metadata points in the first metadata group are evenly distributed, and the metadata points in the second metadata group are discretely distributed; According to the basic data of the net value of daily planned portfolio position transactions, the basic data indicator library is parsed and managed from the attributes, dimensions, and application indicator classification tree. The attributes include business attributes and technical attributes, and the dimensions include calculation objects, application objects, calculation levels, time intervals, indicator units, data types, engine sources, and indicator types. The application indicator classification tree includes application object layer, indicator classification layer, and indicator detail classification layer. Establish an indicator dimension matrix, generate an engine indicator set based on the basic engine indicators registered to the indicator platform by each computing engine, generate an application indicator set based on the application indicators with business significance derived from the engine indicators, form a customized indicator plan based on the indicator set packaged by the application indicators of the indicator platform, perform functional tests on the indicator plan that has been created to ensure the correctness of the return results of the created indicator interface, select portfolios / plans / investment managers / investment managers and their corresponding level indicators for flexible data analysis, analyze the detailed assets in the portfolio, and implement detailed level indicator analysis by selecting the corresponding portfolios and detailed assets, as well as detailed related indicators, automatically generate quantitative indicators based on unified indicator generation rules and business calibers, and verify them through the indicator browser, process indicators based on daily data, monthly frequency data, comparison indicators, benchmarking indexes, composite and average calculation results, indicators formed by interval correlation coefficients, indicators formed by similar rankings, and indicators formed by scenario analysis to form quantitative indicators that meet business requirements; Based on the combination of qualitative and quantitative indicators, the calculation results are labeled according to the weights to generate multi-quantile ranking labels, weight value comparison labels, and fuzzy matching labels to generate the investment manager style label library; Based on the relationship between qualitative and quantitative indicators and the investment manager's style label library, the investment manager's basic situation, return characteristics, asset allocation capabilities, fixed income investment capabilities, equity investment capabilities, and risk management investment capabilities are evaluated through preset scoring rules and scoring templates to obtain the annuity investment manager evaluation results.

2. The method for evaluating annuity investment managers according to claim 1, characterized in that: When analyzing the due diligence dimension metadata, if corresponding information can be found in the due diligence table and the matching degree is greater than 90%, the due diligence information is used as a positioning qualitative indicator, and the positioning qualitative indicator is used to generate a qualitative indicator library.

3. The method for evaluating annuity investment managers according to claim 1, characterized in that: When analyzing the due diligence dimension metadata, if the due diligence information cannot be used directly as a qualitative indicator, the indicator value is qualitatively determined by manual judgment to obtain an artificial qualitative indicator, and the artificial qualitative indicator is used to generate a qualitative indicator library.

4. The method for evaluating annuity investment managers according to claim 1, characterized in that: When performing the distribution characteristic analysis, one or more of the frequency distribution histogram, median statistical indicators, variance and standard deviation are used for judgment. If the histogram appears to be rectangular, the difference between the mean and the median is not large, and the variance or standard deviation is small, then the distribution is considered to be uniform, otherwise it is considered to be discrete.

5. The method for evaluating annuity investment managers according to claim 1, characterized in that: The preset scoring rules include: equivalence method, tiered scoring method, tiered weighted scoring method, ranking scoring method, reverse percentile ranking method, tiered + linear interpolation method, ranking + linear interpolation method, tiered / ranking + multi-linear interpolation method.

6. The method for evaluating annuity investment managers according to claim 1, characterized in that: The method also includes comparing the quantitative indicator with the market index benchmark by adding a corresponding market index as a benchmark, and further analyzing the difference with the market common benchmark.

7. The method for evaluating annuity investment managers according to claim 1, characterized in that: The scoring template is designed according to time interval and object type.

Images