CN109559042A - A kind of fund manager's scoring algorithm based on multidimensional index regression analysis - Google Patents
A kind of fund manager's scoring algorithm based on multidimensional index regression analysis Download PDFInfo
- Publication number
- CN109559042A CN109559042A CN201811449984.6A CN201811449984A CN109559042A CN 109559042 A CN109559042 A CN 109559042A CN 201811449984 A CN201811449984 A CN 201811449984A CN 109559042 A CN109559042 A CN 109559042A
- Authority
- CN
- China
- Prior art keywords
- fund
- fund manager
- manager
- nearly
- score
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06398—Performance of employee with respect to a job function
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Educational Administration (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The method of the present invention is mainly used for the scoring of fund manager.By having managed fund performance performance since holding a post to fund manager, personal index is respectively created by Fund Type first in method;It then is to calculate basis with personal exponent data, the quantization for doing different dimensions to performance of the fund manager on certain type fund is investigated, and is investigated dimension and is included profit point, Sharp point, stability score, anti-risk score, debt ability of selecting stocks, selects 9 dimensions such as Shi Nengli, empirical value, the scoring of benchmark tracking ability, excess earnings power scoring;Each dimension of the weighting evaluation fund manager in the short, medium and long phase later is based on linear regression model (LRM) finally by previous fund data and carries out parameter estimation, and export the comprehensive score of each fund manager.Fund score rank can recommend outstanding fund manager and and its corresponding managed fund for user.
Description
Technical field
The present invention relates to fund data digging technology fields, more particularly, to a kind of base based on multidimensional index regression analysis
Jin handles scoring algorithm.
Background technique
Security investment fund is a kind of specialized investment way, it is by way of floatation of funds share, so that the public
The fund of investor's dispersion is gathered in the legal form for closing rule as a security investment fund.It is general in security investment fund
Trusteeship party of the bank as fund, fund management company carry out the operation of fund as fund manager, and the fund investors are then
Its respective investment return is obtained in a manner of joint adventure.
The characteristics of investment funds: investment funds can turn parts into the whole, and reduce risk by professional means, while can enjoy
Offshore fund by preferential tax revenue and is more absorbed extensively.Investment funds can increase the diversity of investment combination, and investor can be with
It is open with the overall risk that this reduces investment combination.
Fund manager generally requires the financial above education background of relevant speciality master, have good Fundamentals of Mathematics and
Sturdy economic theory grounding in basic skills then will be more competitive if any experience or the acquisition CFA certificate of studying abroad.Every kind of fund is equal
The combination and investment tactics for being responsible for determining the fund are gone by a manager or one group of manager, investment combination is according to fund specification
The investment objective go to select, and go to determine by the investment tactics of the fund manager.Warfighting capabilities, that is, investment performance of investment is determined
Determine its influence power and emolument return in the industry.
The ranking of fund manager is a kind of assessment of the mechanism for the continuous exhibition ability of fund manager, and investor is selected
Fund manager is selected with reference significance, directly concerning the pecuniary benefit of investor.If establish a science, it is objective, have
Fund manager's ranking (performance appraisal) system of effect, then for the investor of Fund Market, so that it may more from side
Understand the risk status data of the fund of its management, and then science preferably is carried out to Market Fund and is effectively selected.
Summary of the invention
The method of the present invention is mainly used for the scoring of fund manager.Method since holding a post to fund manager first by having been managed
Fund performance performance is crossed, personal index is respectively created by Fund Type;It then is to calculate basis with personal exponent data, to fund
The quantization investigation of different dimensions is done in performance of the manager on certain type fund, and investigating dimension includes that profit is divided, Sharp divides, stability
Score, anti-risk score, debt ability of selecting stocks, select Shi Nengli, empirical value, the scoring of benchmark tracking ability, excess earnings power comments
Divide equal 9 dimensions;Each dimension of the weighting evaluation fund manager in the short, medium and long phase later, finally by previous fund data base
Parameter estimation is carried out in linear regression model (LRM), and exports the comprehensive score of each fund manager.Fund score rank can be use
Recommend outstanding fund manager and and its corresponding managed fund in family.
Specific step is as follows:
(1) the personal index of the be good at Fund Type of fund manager is calculated;
(2) based on personal index, the dimension that fund manager respectively scores is calculated;
(3) dimension scores are calculated based on short, medium and long phase of the weighting evaluation to fund manager;
(4) weight determination is carried out to fund manager's dimensions based on linear regression model (LRM) and calculates final score.
Wherein, the personal index of calculating the be good at Fund Type of fund manager of step (1), refers to and setsIt is the fund
The day amount of increase and amount of decrease of the kth Zhi Jijin of i-th of day of trade management is handled,It is corresponding fund flows size, thenIndicate that i-th hands over
Such fund asset scale easily currently managed day is the weighted average day amount of increase and amount of decrease of weight:
According toDefine the personal index value of j-th of day of trade of fund manager are as follows:
。
Wherein, the dimension of step (2) to be scored based on personal index, calculating fund manager, is referred to and is commented fund manager
Divide 9 dimensions expansion.
Dimension 1: nearly N fund manager profit point.Remember fund manager in i-th of management period of nearly N, that is, refer to last time from
The period that duty is left office to next time,P i Fund manager is indicated in the i-th period of nearly N, (there is managed fund in this section of tenure
Period) personal index return rate sorts from large to small, be then converted into score, i.e. 1-ranking/participate in evaluation and electing fund manager's sum,
Then:, whereinIndicate the personal index set for participating in appraising through comparison fund manager,Indicate the fund
The ranking (descending ranking) of manager,Sum is participated in evaluation and electing in expression (participate in evaluation and electing fund manager's sum).Nearly N manager profit point
Are as follows:。
Dimension 2: nearly N fund manager Sharp point.The Sharpe Ratio of i-th of management period of the nearly N of fund manager are as follows:, whereinRisk free rate is indicated, using domestic term bank fixed-term deposit rate/365;It indicates
Fund manager i-th section of tenure (period of managed fund) of nearly N day amount of increase and amount of decrease standard deviation.It willIt carries out
Ranking obtains the ranking of each fund manager, is then converted to its Sharp point:
, wherein in the periodIndicate the Sharp Ratio set for participating in appraising through comparison fund manager,
Indicate the ranking (descending ranking) of the fund manager,Sum is participated in evaluation and electing in expression (participate in evaluation and electing fund manager's sum).Nearly N
Manager Sharp is divided into:。
Dimension 3: nearly N fund manager's stability score.In nearly N, during i-th of period of fund manager's management,
The earning rate of its v-th of day of trade is denoted as, the period daily earning rate that is averaged is denoted as, true with the weighting of unbiased variance
The stability of the fixed period:,
Wherein, in the periodIndicate the set of the stability score of participation comparation and assessment fund manager,Indicate that the fund passes through
The ranking (descending ranking) of reason,Sum is participated in evaluation and electing in expression (participate in evaluation and electing fund manager's sum).Nearly N fund manager's is steady
Qualitative score are as follows:。
Dimension 4: nearly anti-risk point of fund manager of N.IfFor the net value of the i-th some day period,When indicating i-th
Between maximum net value of the section from starting current point,It is the net value of current point, then the maximum in the i-th period withdraws rate are as follows:, by each fund manager's periodRanking is carried out, obtains the ranking of each fund manager, then turn
Turn to anti-risk point:, wherein in the periodIndicate anti-risk point of collection for participating in appraising through comparison fund manager
It closes,Indicate the ranking (descending ranking) of the fund manager,Indicate the anti-risk fund manager that participates in evaluation and electing that participates in evaluation and electing
Sum.The stability score of nearly N fund manager are as follows:。
Dimension 5: fund manager's Stock Selectivity/select debt ability score.Four weeks time are selected by the method that expert determines
Phase, respectively 2005/07/21-2007/10/16,2008/10/29-2009/08/04,2014/07/22-2015/06/12,
, the profitability scoring of the managed fund in above four periods is calculated, fund manager is found out
Stock Selectivity/select debt ability score are as follows:。
Dimension 6: ability score when fund manager selects.Four time cycles are selected by the method that expert determines, respectively
2007/10/17-2008/10/28、2011/04/23-2012/12/04、2015/06/13-2016/01/27、2018/01/30-
2018/11/27, the profitability scoring of the managed fund in above four periods is calculated, find out fund manager selects Shi Nengli
Score are as follows:。
Dimension 7: fund manager's empirical value score.By the management Fund Type day of trade total number of days t from more to small sequence, so
After convert component number, i.e., 1-ranking/sum.The empirical value of fund manager are as follows:=。
Dimension 8: nearly N benchmark tracking ability scoring.5 steps: 1. notes are divided into the calculating of tracking ability scoring
Be kth fund in time t(t=1 day) in amount of increase and amount of decrease,On the basis of combination in time t(t=1 day) in amount of increase and amount of decrease,
Obtain indicate kth fund in time t(t=1 day) in tracking irrelevance,;2. calculating tracking to miss
Difference,;3. calculating the tracking error of the t period of nearly N;4. calculating nearly N
The performance tracking error of kth fund in year tracks target error from small by winning the race in same type fund in t-th period
To big Rank scores:;Unit net value based on fund weighs calculating, fund manager's tenure in nearly N again
It manages kth exponential type fund benchmark and tracks score are as follows:;5. being weighted and averaged nearly N with fund flows
The benchmark of interior managed fund tracks scoring:。
Dimension 9: nearly N excess earnings power scoring.Unit net value based on fund weighs calculating again, calculates fund manager and exists
The excess earnings score of the tenure fund in nearly N, ifLess than zero, then it is not involved in ranking, which is scored at zero., then nearly N excess earnings power scoring:。
Wherein, step (3) based on weighting evaluation calculates dimension scores to short, medium and long phase of fund manager, refers to fund
Manager carries out short-term (nearly 1 year) to the same fund manager respectively by the above dimension, while by the method for expert judgments,
Mid-term (nearly 3 years), the investigation of long-term (nearly 5 years), and calculate dataThat is the nearly N score of i-th dimension degree: if the fund passes through
Reason meets 1 annual data, then;If the fund manager meets 3 annual datas,;If
The fund manager meets 5 annual datas, then;The as i-th dimension degree final score of the fund.
Wherein, step (4) to carry out weight to fund manager's dimensions based on linear regression model (LRM) determining and calculate most
Whole score refers to the dimension obtained by data prediction, as input vector and it is desirable that predict real value。
Linear regression model (LRM) shaped like:, passed through by the fund that previous fund data calculates time in the past section
Reason dimension values R and the fund practical manifestation of fund manager's management provide expert analysis mode y, are denoted as training dataset, eachIt is the characteristic measure vector of i-th of data.Pass through minimum
Square law selects coefficient, it is 0 by the way that it is differentiated and enabled to w, obtains unique solution:, then in input vectorOn predicted value are as follows:, the original of fund manager comment
It is divided into:。
Detailed description of the invention
Fig. 1 is the flow chart of fund manager's scoring algorithm based on multidimensional index regression analysis.
Fig. 2 is the partial output results that fund manager's long-term score ranking is calculated based on nearly 5 annual data.
Fig. 3 is the partial output results that fund manager's short-term score ranking is calculated based on nearly 1 annual data.
Specific embodiment
With reference to the accompanying drawing and example, the method for the present invention is described.
The method of the present invention is created by having managed fund performance performance since holding a post to fund manager by Fund Type respectively
Personal index is built, and is based on the premise of market environment identical (i.e. system risk is identical), is to calculate basis with personal exponent data,
The quantization for doing different dimensions to performance of the fund manager on certain type fund is investigated.The Asset Allocation requirement of different type fund
It requires to be all different with management, so the investigation dimension of different type fund is different.Such as fund manager is on stock fund
Investigation dimension include that profitability, the ability to ward off risks, Stock Selectivity, selects Shi Nengli, empirical value at Sharpe Ratio.In exponential type
On investigation dimension be tracking error, excess earnings and empirical value.Model finally calculates fund manager and manages different fund classes
Short-term, the ability scoring of medium and long term of type.The higher management of investment ability for illustrating fund manager that scores is stronger, fund manager
Managerial ability determine the fund performance of fund, in short, the superiority and inferiority for judging fund can be provided for user by model.
The specific steps of the method for the present invention are described as follows.
One, the personal index of the be good at Fund Type of fund manager is calculated.
IfIt is the day amount of increase and amount of decrease of the kth fund managed i-th of day of trade of the fund manager,It is corresponding fund
Scale, thenIndicate that such fund asset scale currently managed i-th day of trade is the weighted average day amount of increase and amount of decrease of weight:
According toDefine the personal index value of j-th of day of trade of fund manager are as follows:
。
Two, based on personal index, the dimension that fund manager respectively scores is calculated.
The scoring of fund manager is calculated from the expansion of following 9 dimensions.In calculating process, have following three it is common
Rule, i.e.,
(1) there is side by side ranking in the dimension of Rank scores involved in model, that is, is worth identical, and ranking is identical;
(2) other fund managers that sequence is participated in involved in model must satisfy the day of trade of scoring the i-th period of fund manager 3/4
It;
(3) fund manager managed total transaction number of days of fund in nearly N, it is necessary to it is more than or equal to the friendship of (3/4 nearly N)
Easy number of days.
The following detailed description of the calculating process of each dimension.
Nearly N fund manager profit point.
The fund profit and loss situation that the seperated existing fund manager of fund manager's profit is managed in nearly N.Remember fund manager in nearly N
I-th of management period in year refers to that last time leaves office the period left office to next time,P i Indicate fund manager in the i-th of nearly N
Section, the personal index return rate of this section of tenure (period for having managed fund) sort from large to small, and are then converted into score, i.e.,
1-ranking/participate in evaluation and electing fund manager's sum, then:
Wherein,Indicate the personal index set for participating in appraising through comparison fund manager,Indicate the ranking (descending of the fund manager
Ranking),Sum is participated in evaluation and electing in expression (participate in evaluation and electing fund manager's sum).
Nearly N manager profit is divided into:
Wherein, m represents the when number of segment of the nearly N managed fund of fund manager;Represent the nearly N of fund manager before i-th day the section appoint
Duration during duty;Represent total transaction number of days of the nearly N managed fund of fund manager.
Nearly N fund manager Sharp point.
Sharpe Ratio, be otherwise known as Sharp Ratio, is fund valuation standardized index, it is therefore an objective to calculate investment combination
One unit overall risk of every receiving, can generate how many excess salaries.Sharp's ratio of i-th of management period of the nearly N of fund manager
Rate are as follows:
Wherein,Risk free rate is indicated, using domestic term bank fixed-term deposit rate/365;Indicate that fund manager exists
I-th section of tenure (period of managed fund) of nearly N day amount of increase and amount of decrease standard deviation.
It willRanking is carried out, the ranking of each fund manager is obtained, is then converted to its Sharp point:
Wherein, in the periodIndicate the Sharp Ratio set for participating in appraising through comparison fund manager,Indicate the fund manager's
Ranking (descending ranking),Sum is participated in evaluation and electing in expression (participate in evaluation and electing fund manager's sum).
Nearly N manager Sharp is divided into:
Wherein, m represents the when number of segment of the nearly N managed fund of fund manager;Represent the nearly N of fund manager before i-th day the section appoint
Duration during duty;Represent total transaction number of days of the nearly N managed fund of fund manager.
Nearly N fund manager's stability score.
In nearly N, during i-th of period of fund manager's management, the earning rate of its v-th of day of trade is denoted as,
The period daily earning rate that is averaged is denoted as, the stability of the period is determined with the weighting of unbiased variance:
。
By each fund manager's periodRanking is carried out, the ranking of each fund manager is obtained, is then converted to them
Stability score:
Wherein, in the periodIndicate the set of the stability score of participation comparation and assessment fund manager,Indicate that the fund passes through
The ranking (descending ranking) of reason,Sum is participated in evaluation and electing in expression (participate in evaluation and electing fund manager's sum).
The stability score of nearly N fund manager are as follows:
Wherein, m represents the when number of segment of the nearly N managed fund of fund manager;Represent the nearly N of fund manager before i-th day the section appoint
Duration during duty;Represent total transaction number of days of the nearly N managed fund of fund manager.
Nearly anti-risk point of fund manager of N.
Value is withdrawn with maximum to estimate the anti-risk score of fund manager.It is any within the selected period go through that maximum, which withdraws value,
For history time point toward pusher, earning rate when product net value goes to minimum point withdraws the maximum value of amplitude.
For the net value of the i-th some day period,Indicate maximum net value of i-th period from starting current point,
It is the net value of current point, then the maximum in the i-th period withdraws rate are as follows:
。
By each fund manager's periodRanking is carried out, the ranking of each fund manager is obtained, is then converted to wind resistance
Danger point:
Wherein, in the periodIndicate anti-risk point of set for participating in appraising through comparison fund manager,Indicate the fund manager
Ranking (descending ranking),Indicate the anti-risk fund manager's sum that participates in evaluation and electing that participates in evaluation and electing.
The stability score of nearly N fund manager are as follows:
Wherein, m represents the when number of segment of the nearly N managed fund of fund manager;Represent the nearly N of fund manager before i-th day the section appoint
Duration during duty;Represent total transaction number of days of the nearly N managed fund of fund manager.
Fund manager's Stock Selectivity/select debt ability score.
By analyzing equity fund market and tendency in the past 20 years, the current generation is had selected by the method that expert determines
Four time cycles are respectively 2005/07/21-2007/10/16,2008/10/29-2009/08/04,2014/07/22-
2015/06/12,2016/01/28-2018/01/29 determines the Stock Selectivity of the fund manager to participate in evaluation and electing/select debt ability, simultaneously
The period is supported to increase and modify, and can be changed at any time.
Calculate profitability scoring (the profit energy with fund manager's diagnostic model of the managed fund in above four periods
Power scoring calculating is similar, only the difference of period, i=1,2,3,4) it obtains, find out the Stock Selectivity of fund manager/select debt
Ability score are as follows:
Wherein,Represent the duration of the nearly N of fund manager this section of tenure before i-th day.
Ability score when fund manager selects.
By analyzing equity fund market and tendency in the past 20 years, the current generation is had selected by the method that expert determines
Four time cycles are respectively 2007/10/17-2008/10/28,2011/04/23-2012/12/04,2015/06/13-
2016/01/27,2018/01/30-2018/11/27, to determine the Shi Nengli that selects of the fund manager to participate in evaluation and electing, while the period branch
Increase and modification are held, can be changed at any time.
Calculate profitability scoring (the profit energy with fund manager's diagnostic model of the managed fund in above four periods
Power scoring calculating is similar, only the difference of period, i=1,2,3,4) it obtains, find out ability score when selecting of fund manager
Are as follows:
Wherein,Represent the duration of the nearly N of fund manager this section of tenure before i-th day.
Fund manager's empirical value score.
Its empirical value in industry is determined according to the time of fund manager's actual management fund.By the management Fund Type
The day of trade total number of days t from more to small sequence, is then converted into score, i.e., 1-ranking/sum.The empirical value of fund manager are as follows:
=
Wherein, in the periodIndicate the set of the empirical value of participation comparation and assessment fund manager,Indicate the row of the fund manager
Name (descending ranking),Indicate participate in evaluation and electing experience similar fund sum.
Nearly N benchmark tracking ability scoring.
Following 5 steps are divided into the calculating of tracking ability scoring.
(1) irrelevance is tracked.NoteBe kth fund in time t(t=1 day) in amount of increase and amount of decrease,On the basis of group
Close in time t(t=1 day) in amount of increase and amount of decrease, obtain indicate kth fund in time t(t=1 day) in tracking irrelevance,。
(2) tracking error.WhereinIndicate the tracking error of kth fund;Indicate the sample average of the tracking irrelevance of kth fund;N is sample number.Tracking error is bigger, illustrates the net value of fund
Rate and the difference of benchmark portfolio yield asked are bigger, and the risk that fund manager actively invests is bigger.It has been generally acknowledged that tracking
Error means that comparison in difference is significant 2% or more.
(3) tracking error of the t period of nearly N is calculated。Sample number n takes the nearly N manager to manage
Data, and meet 3/4 tracking error of 300 day of trade of Shanghai and Shenzhen.
(4) in nearly N kth fund performance tracking error.In t-th period in same type fund by win the race with
Track target error Rank scores from small to large:
Wherein, in the periodIndicate the set of the tracking error of the same type fund of participation comparation and assessment,Indicate the base
Gold is won the race in same type fund tracks target error,Indicate the similar fund for participating in evaluation and electing anti-risk sum.Based on base
The unit net value of gold weighs calculating again, and fund manager's job management kth exponential type fund benchmark in nearly N tracks score are as follows:
。
(5) the benchmark tracking scoring of managed fund in nearly N is weighted and averaged with fund flows:
Wherein,Indicate the management scale of kth fund tenure,Indicate all funds of management in the nearly N of fund manager
The total scale of tenure.
Nearly N excess earnings power scoring.
Unit net value based on fund weighs calculating again, calculates the excess earnings of fund manager's tenure fund in nearly N
Score, ifLess than zero, then it is not involved in ranking, which is scored at zero,
Wherein,Representative function, whenBe equal to 1, other when be 0.
Then nearly N excess earnings power scoring:
。
Three, dimension scores are calculated based on short, medium and long phase of the weighting evaluation to fund manager.
Fund manager respectively carries out the same fund manager by the above dimension, while by the method for expert judgments
(nearly 1 year) in short term, mid-term (nearly 3 years), the investigation of long-term (nearly 5 years), and calculate dataThat is the nearly N score of i-th dimension degree.
If the fund manager meets 1 annual data,;
If the fund manager meets 3 annual datas,;
If the fund manager meets 5 annual datas,;
The as i-th dimension degree final score of the fund.
Four, weight determination is carried out to fund manager's dimensions based on linear regression model (LRM) and calculates final score.
The dimension obtained by data prediction, as input vector and it is desirable that predict real value。
Using linear regression model (LRM) shaped like:
Fund manager's dimension values R of time in the past section and the fund of fund manager's management are calculated by previous fund data
Practical manifestation provides expert analysis mode y, is denoted as training dataset, eachIt is the characteristic measure vector of i-th of data.
Coefficient is selected by least square method, with minimization residual sum of squares (RSS):
It is written as vector matrix form are as follows:
It is 0 by the way that it is differentiated and enabled to w, obtains unique solution:
Then in input vectorOn predicted value are as follows:
The original scoring of fund manager are as follows:
For fund manager's managed fund duration, 3/4 day of trade number of days of year and a day does not score then.
In conclusion the invention discloses a kind of fund manager's scoring algorithms based on multidimensional index regression analysis.Method
It is to calculate basis with personal exponent data, the quantization for doing different dimensions to performance of the fund manager on certain type fund is investigated,
Score of the weighting evaluation fund manager in the short, medium and long phase later;Finally building linear regression model (LRM) carries out parameter estimation and exports
The score and its ranking results of fund manager, to recommend outstanding fund manager and and its corresponding managed fund for user.
The method of the present invention is similarly applied to the data that security class has similar operations ontology, such as privately-offered fund, futures, stock
The administrator of ticket etc. scores.Therefore, although disclosing specific embodiments of the present invention and attached drawing for the purpose of illustration, its object is to
Help understands the contents of the present invention and implements accordingly, but it will be appreciated by those skilled in the art that: do not depart from the present invention and
In the spirit and scope of the attached claims, various substitutions, changes and modifications are all impossible.Therefore, the present invention does not answer
It is confined to most preferred embodiment and attached drawing disclosure of that.Presently disclosed embodiment should be understood illustrative in all respects
Rather than limitation to its claimed range.
Claims (7)
1. a kind of fund manager's scoring algorithm based on multidimensional index regression analysis, it is characterised in that the method includes walking as follows
It is rapid:
(1) the personal index of the be good at Fund Type of fund manager is calculated;
(2) based on personal index, the dimension that fund manager respectively scores is calculated;
(3) dimension scores are calculated based on short, medium and long phase of the weighting evaluation to fund manager;
(4) weight determination is carried out to fund manager's dimensions based on linear regression model (LRM) and calculates final score.
2. a kind of fund manager's scoring algorithm based on multidimensional index regression analysis according to claim 1, feature exist
In, the calculating process of the personal index of the be good at Fund Type of fund manager, ifIt is that the fund manager manages i-th of day of trade
The day amount of increase and amount of decrease of the kth fund of reason,It is corresponding fund flows size, thenIndicate currently to manage i-th day of trade is somebody's turn to do
Class fund asset scale is the weighted average day amount of increase and amount of decrease of weight:
,
According toDefine the personal index value of j-th of day of trade of fund manager are as follows:
。
3. a kind of fund manager's scoring algorithm based on multidimensional index regression analysis according to claim 1, feature exist
In setting and corresponding calculating process to the dimensions of fund manager calculate the scoring of fund manager from 9 dimension exhibitions
It opens, is respectively as follows: nearly N fund manager profit point, nearly N fund manager Sharp point, nearly N fund manager stability score, nearly N
Ability score, fund warp when year anti-risk point of fund manager, fund manager's Stock Selectivity/debt ability score, fund manager is selected to select
Manage empirical value score, the scoring of nearly N benchmark tracking ability, the scoring of nearly N excess earnings power.
4. the dimensions of fund manager according to claim 3 calculate nearly N fund manager profit point, feature exists
In, note fund manager refers to that last time leaves office the period left office to next time in i-th of management period of nearly N,P i Indicate fund
Manager arranges from big to small in the i-th period of nearly N, the personal index return rate of this section of tenure (period for having managed fund)
Sequence is then converted into score, i.e. the 1- ranking/fund manager that participates in evaluation and electing is total, then, whereinIt indicates to participate in commenting
Than the personal index set of fund manager,Indicate the ranking (descending ranking) of the fund manager,Expression is participated in evaluation and electing
Sum (participate in evaluation and electing fund manager's sum);Nearly N manager profit is divided into:
Wherein, m represents the when number of segment of the nearly N managed fund of fund manager;Represent the nearly N of fund manager this section tenure before i-th day
The duration of period;Represent total transaction number of days of the nearly N managed fund of fund manager.
5. the dimensions of fund manager according to claim 3, calculating fund manager's Stock Selectivity/debt ability is selected to obtain
Point, which is characterized in that having selected four time cycles by the method that expert determines is respectively 2005/07/21-2007/10/
16,2008/10/29-2009/08/04,2014/07/22-2015/06/12,2016/01/28-2018/01/29 join to determine
The Stock Selectivity of the fund manager commented/select debt ability calculates the profitability scoring of the managed fund in above four periods, asks
The Stock Selectivity of fund manager/select debt ability score out are as follows:
。
6. the dimensions of fund manager according to claim 3 calculate ability score when selecting of fund manager, feature
It is, having selected four time cycles by the method that expert determines is respectively 2007/10/17-2008/10/28,2011/04/
, 2015/06/13-2016/01/27,2018/01/30-2018/11/27, come determine participate in evaluation and electing fund warp
Reason selects Shi Nengli, calculates the profitability scoring of the managed fund in above four periods, finds out energy when selecting of fund manager
Power score are as follows:
。
7. the dimensions of fund manager according to claim 3 calculate the benchmark tracking ability scoring of fund manager,
It is characterized in that, following 5 steps: 1. notes is divided into the calculating of tracking ability scoringIt is kth fund in time t(t=1
Day) in amount of increase and amount of decrease,On the basis of combination in time t(t=1 day) in amount of increase and amount of decrease, obtain indicate kth fund in the time
T(t=1 day) in tracking irrelevance,;2. tracking error is calculated,;3. calculating the tracking error of the t period of nearly N;4. calculating in nearly N
The performance tracking error of kth fund tracks target error from small to large by winning the race in same type fund in t-th of period
Rank scores:;Unit net value based on fund weighs calculating, fund manager's job management in nearly N again
Kth exponential type fund benchmark tracks score are as follows:;5. being weighted and averaged nearly N inner tube with fund flows
The benchmark for managing fund tracks scoring:。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811449984.6A CN109559042A (en) | 2018-11-30 | 2018-11-30 | A kind of fund manager's scoring algorithm based on multidimensional index regression analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811449984.6A CN109559042A (en) | 2018-11-30 | 2018-11-30 | A kind of fund manager's scoring algorithm based on multidimensional index regression analysis |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109559042A true CN109559042A (en) | 2019-04-02 |
Family
ID=65868093
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811449984.6A Pending CN109559042A (en) | 2018-11-30 | 2018-11-30 | A kind of fund manager's scoring algorithm based on multidimensional index regression analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109559042A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111539832A (en) * | 2020-04-21 | 2020-08-14 | 林树 | Multidimensional fund evaluation system and method |
CN113205408A (en) * | 2021-05-28 | 2021-08-03 | 中国工商银行股份有限公司 | Customer manager capacity map generation method and device |
CN113435746A (en) * | 2021-06-28 | 2021-09-24 | 平安银行股份有限公司 | User workload scoring method and device, electronic equipment and storage medium |
TWI775028B (en) * | 2019-12-20 | 2022-08-21 | 國立臺北商業大學 | financial community system |
CN116227958A (en) * | 2022-11-24 | 2023-06-06 | 北京汇成基金销售有限公司 | Method and system for dynamically and quantitatively evaluating offset fund manager based on holding bin and net value |
-
2018
- 2018-11-30 CN CN201811449984.6A patent/CN109559042A/en active Pending
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI775028B (en) * | 2019-12-20 | 2022-08-21 | 國立臺北商業大學 | financial community system |
CN111539832A (en) * | 2020-04-21 | 2020-08-14 | 林树 | Multidimensional fund evaluation system and method |
CN113205408A (en) * | 2021-05-28 | 2021-08-03 | 中国工商银行股份有限公司 | Customer manager capacity map generation method and device |
CN113435746A (en) * | 2021-06-28 | 2021-09-24 | 平安银行股份有限公司 | User workload scoring method and device, electronic equipment and storage medium |
CN116227958A (en) * | 2022-11-24 | 2023-06-06 | 北京汇成基金销售有限公司 | Method and system for dynamically and quantitatively evaluating offset fund manager based on holding bin and net value |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109559042A (en) | A kind of fund manager's scoring algorithm based on multidimensional index regression analysis | |
Bartelsman et al. | Understanding productivity: Lessons from longitudinal microdata | |
McMillan et al. | University efficiency: A comparison and consolidation of results from stochastic and non‐stochastic methods | |
Önder et al. | Forecasting macroeconomic variables using artificial neural network and traditional smoothing techniques | |
Sonje et al. | Efficiency of public expenditure on education: comparing Croatia with other NMS | |
CN108537436A (en) | Securities business Performance Evaluation Methods, system, equipment, storage medium | |
Plotnikov et al. | Model for Assessing the Human Capitalof an Enterprise | |
Lu et al. | Exploring the benchmarks of the Taiwanese investment trust corporations: Management and investment efficiency perspectives | |
Wichmann et al. | Auctioning risky conservation contracts | |
Sharma | Evaluation of Basel III revision of quantitative standards for implementation of internal models for market risk | |
Basri et al. | The efficiency of Islamic banks in Malaysia: Based on DEA and Malmquist productivity index | |
Wanke et al. | Revisiting camels rating system and the performance of Asean banks: a comprehensive mcdm/z-numbers approach | |
Panić et al. | Analysis of higher education indicators coherency in central and eastern Europe | |
RT Ferreira | Financial volatility and its economic effects | |
JPH10275177A (en) | Device and method for evaluating performance of investment trust | |
Sood et al. | Deep Reinforcement Learning for Optimal Portfolio Allocation: A Comparative Study with Mean-Variance Optimization | |
Kimeu | The effect of portfolio composition on the financial performance of investment companies listed in Nairobi securities exchange | |
Nuroğlu | Estimating and forecasting trade flows by panel data analysis and neural networks | |
Maslak et al. | Application of the Rasch method of evaluating latent variables in management and administration | |
Lian et al. | Machine learning and time series models for VNQ market predictions | |
Suvorov et al. | Approaches to measuring the dynamics and structure of human capital and assessing its accumulated impact on economic growth | |
CN111402041A (en) | Scientific plate simulated marketing enterprise valuation model based on principal component analysis method and Q cluster analysis | |
Nesvetailov et al. | Assessment of sustainable development levels in russian organizations using fuzzy sets theory | |
Gordienko et al. | Fiscal and investment indicators for evaluating professional sports facilities | |
Zheng | Efficiency measurement. A methodological comparison of parametric and non-parametric approaches. |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190402 |