CN109118041A - Stock market's investor sentiment index back-and-forth method based on cluster with maximum entropy delta ratio - Google Patents
Stock market's investor sentiment index back-and-forth method based on cluster with maximum entropy delta ratio Download PDFInfo
- Publication number
- CN109118041A CN109118041A CN201810701748.2A CN201810701748A CN109118041A CN 109118041 A CN109118041 A CN 109118041A CN 201810701748 A CN201810701748 A CN 201810701748A CN 109118041 A CN109118041 A CN 109118041A
- Authority
- CN
- China
- Prior art keywords
- index
- entropy
- stock market
- cluster
- maximum entropy
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/06—Asset management; Financial planning or analysis
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Accounting & Taxation (AREA)
- Marketing (AREA)
- Educational Administration (AREA)
- Finance (AREA)
- Game Theory and Decision Science (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Technology Law (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention is a kind of stock market's investor sentiment index back-and-forth method based on cluster with maximum entropy delta ratio, is related to stock market's investor sentiment and estimates field, comprising: index mass-election, any index related with stock market or investor can bring alternative index storehouse into；Data non-availability or the discontinuous invalid index of data are rejected in primary election；Screening rejects unrelated index by unrelated analysis；It selects, the index that amount of redundant information is greater than threshold value is deleted by correlation analysis and clustering；It is preferred that calculating sensitive factor, i.e. comentropy is stayed in conjunction with clustering map than mean entropy and maximum entropy than the ratio between mean entropy and selects the highest index of conspicuousness.The open selective goal of the present invention, design rationally, by a series of quantitative analysis, obtain it is more scientific, more contain, wider investor sentiment is estimated acts on behalf of index.
Description
Technical field
The invention belongs to investor sentiments to estimate field more particularly to a kind of stock based on cluster with maximum entropy delta ratio
City's investor sentiment index back-and-forth method.
Background technique
The stabilization of maintenance stock market depends on the stabilization of investor sentiment, and many behavior Financial Research are all built at this stage
It stands on the basis of investor sentiment, wherein estimating for investor sentiment is critical issue in investor sentiment research, science
Reasonably selecting investor sentiment and acting on behalf of index is the basis estimated.
Research about mood measurement index mainly has following two feature:
First, measurement index is numerous, and theoretical foundation and method for normalizing there is no to carry out preferably these indexs.Currently about investor
Measurement Method of the connotation of mood, the mechanism of action of the emotion processes of Investor's Decision, investor sentiment etc. none can be with
Control global systematicness to discuss, estimate about mood and do not propose its theoretical foundation specially, scholars from each Microscopic Angle,
Establish individual event mood measurement index with their own characteristics --- it up to up to a hundred (lot of documents repeats no more), enriches this and grinds
Study carefully, but these indexs are scattered at random, it is unable to decide which is right, it is very different, mood measurement index for how more to standardize is chosen, still
Lack science, uniformly, the method for standard.
Second, the selection of measurement index has subjectivity and belongingness.Numerous mood measurement indexs that scholars construct, which
Need to reject, which is more important, which is more effective, need to come by the method for specification preferred.Investor sentiment is estimated not
Only subjective should optionally be partial to some or several index factors, and should using certain method first to all indexs into
Preferably, devegetation deposits essence to row, eliminates the false and retains the true, and the method that educational circles currently lacks specification a set of so just.
To sum up, it needs to improve the basic theory estimated about investor sentiment, creates a set of investor sentiment and estimate
The preferred method for normalizing of index, to obtain better mood measurement index.This will be established for the Synthetic Measurement of investor sentiment can
It by basis, and by the theory of abundant investor sentiment, while being also the supervision of financial regulation agencies in reality and determining for investor
Plan provides reference frame.
Summary of the invention
(1) the technical issues of solving
The technical problem to be solved by the present invention is to overcome drawbacks described above, provide a kind of based on cluster and maximum entropy delta ratio
Stock market's investor sentiment index back-and-forth method, provide a set of scientific, open, feasible system preferred method so that selection
It is more effective, easier to act on behalf of index, establishes solid foundation for the research of investor sentiment Synthetic Measurement, while also in reality
The supervision of financial regulation agencies and the decision of investor provide reference frame.
(2) technical solution
To solve the above problems, the technical scheme adopted by the invention is that: the stock based on cluster with maximum entropy delta ratio
City's investor sentiment index back-and-forth method, which is characterized in that the described method includes:
Mass-election module, any index related with stock market or investor can bring alternative index storehouse into；
Primary election module rejects data non-availability or the discontinuous invalid index of data；
Screening module rejects unrelated index by unrelated analysis, calculates each index and stock market's deep bid index and dividend yield
The related coefficient of rate, available unrelated degree and significance test value reject the poor index of conspicuousness；
Module is selected, the index that amount of redundant information is greater than threshold value is deleted by correlation analysis and clustering, calculates each finger
Related coefficient between mark sorts out the index that related coefficient is greater than threshold value, deletes information overlap index；
Preferred module, by calculating sensitive factor, i.e., comentropy than mean entropy and maximum entropy than the ratio between mean entropy, in conjunction with poly-
The tree-shaped classification map and index quantity complexity of alanysis, finally stay and select the highest index of conspicuousness.
Preferably, any index of mass-election module can be current index, it is also possible to the derivative index of p phase in advance,
As { An,An-1,An-2...An-p, wherein p is the positive integer greater than 1.
Preferably, the unrelated degree relative coefficient of screening module is 0.3 and following.
Preferably, screening module significance value is t statistic, which obeys the t distribution of n-2 freedom degree, generally
Rate boundary may be selected 1% to 5%.
Preferably, the threshold value for selecting module is 0.9 or more.
Preferably, sensitive factor can be used in the conspicuousness of preferred module, i.e., comentropy is than mean entropy and maximum entropy ratio
The ratio between mean entropy differentiates
Z=ln (m), (3)
In formula, piFor the probability of discrete type system mode, i is system mode number, and m is system mode maximum value, and is defined
When probability is 0, entropy zero；H is the information entropy of index；For mean entropy；Z is maximum entropy；D is sensitive factor, i.e. information
Entropy is than mean entropy and maximum entropy than the ratio between mean entropy.
Preferably, preferred module complexity can be stayed according to artificial experience and demand selects index, it should be in clustering figure
In spectrum, classified by major class to group, in each category equilibrium selection index, the quantity of classification can be taking human as decision.
(3) beneficial effect
Stock market's investor sentiment index back-and-forth method based on cluster with maximum entropy delta ratio that the present invention provides a kind of, with
The prior art is compared, have it is following the utility model has the advantages that
The present invention has rational design, open selective goal, the range of choice of further expansion index, improves index and selects generation
Table, to obtain more containing, the index of acting on behalf of that wider investor sentiment is estimated carries out basis.
By a series of unrelated analysis, correlation analysis, clustering, significance analysis, the choosing of index is advanced optimized
It selects, Quantitatively Selecting standard, avoids the information overlap and synteny in previous index screening, more balancedly selective goal, so that
It stays and selects index redundancy less, reduce the complexity and blindness of photometry system；Meanwhile it is the selection course of index is complete
Visualization, black box transparent procedures so that preferred process definitely, it is more scientific；For grinding for investor sentiment Synthetic Measurement
Study carefully and establish solid foundation, while also providing reference frame for the supervision of financial regulation agencies in reality and the decision of investor.
Detailed description of the invention
The stock market Fig. 1 investor sentiment, which is estimated, acts on behalf of index selection flow chart
Fig. 2 screens 22 index Dendrograms spectrum, related coefficient and sensitive factor figure
Specific embodiment
Technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
It is a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
The time in January, 2005 in December, 2015 is had chosen as sample interval, had collected 60 potential indexs altogether.
Mass-election module: collection obtains 60 indexs, obtains 60x132 rank matrix；
Primary election module: in view of the availability of information search cost and data, continuity, the visitor of index is allowed also for
The property seen remains 33 indexs, obtains a 33x132 rank after rejecting data non-availability or the discontinuous invalid index of data
Original matrix；
Screening module: calculating 33 indexs and upper card composite related coefficient and followed probability, then calculates and receive with upper card composite
The related coefficient and followed probability of beneficial rate can be used SPSS software and correlation coefficient charts are calculated, as shown in table 1 below:
Table 1
It is greater than 1% standard with followed probability according to 0.3 or less related coefficient and rejects the poor index of conspicuousness, can integrates
It stays and selects 22 indexs, delete 11 unrelated indexs.
It selects module: to delete the index that amount of redundant information is greater than threshold value, being analyzed by relevant cluster, obtain system tree
Map specifies the close and distant relation of 22 indexs, such as Fig. 2 according to map.The correlation coefficient threshold for selecting module is selected as 1, rejects
The maximum index of redundancy, and by related coefficient label in 22 indexs cluster map, phase relation ordered series of numbers as shown in Figure 2.It needs
It is noted that related coefficient is 1, it is meant that its status is identical, can mutually represent, and need to only stay and select 1 index.22 indexs
In, having 6 related coefficients is 1, therefore is reduced to 16 indexs.In order to which the technical program is better described, still pressing in next step
It carries out calculating sensitive factor analysis according to 22 indexs.
Preferred module: the comentropy of calculating 22 indexs to be selected of sensitive factor is than mean entropy and maximum entropy than the ratio between mean entropy
Z=ln (m), (3)
In formula, piFor the probability of discrete type system mode, i is system mode number, and m is system mode maximum value, and is defined
When probability is 0, entropy zero；H is the information entropy of index；For mean entropy；Z is maximum entropy；D is sensitive factor, i.e. information
Entropy is than mean entropy and maximum entropy than the ratio between mean entropy.By sensitive factor label in 22 indexs to be selected, such as table 2, table 3, Fig. 2.
Comentropy is bigger, then illustrates that the information content of index is bigger, has more significant information characteristics, has stronger generation
Table.Conversely, comentropy is smaller, then indication information content is smaller, or even does not have representativeness.In selective goal, Ying Youxin
Cease the lower bound limit value of content.The lower limit value of comentropy should be greater than the average value under the system mode, can be only achieved satisfied information
Content can have stronger representativeness.Lower bound limit value herein is average information entropy.
Mean entropy, the size of maximum entropy are only related with the maximum value of system mode, and the comentropy size with other indexs
It is not related.For open index selection method of the invention, undoubtedly optimal selection.No matter how many index to be selected into
Enter index storehouse, all will not influence the identification of information content feature.Sensitive factor can be apparent from the ratio of information delta, into
And make the across comparison performance boost of index system, the sensibility of lifting feature identification.Therefore, comentropy than mean entropy with most
Big entropy is practicable as the conspicuousness differentiation of preferred module than the ratio between mean entropy.
By clustering map combining information entropy than mean entropy and maximum entropy than the ratio between mean entropy, in divided major class
In, it selects sensitive factor maximum as staying and selects index.It stays and selects number that can determine in conjunction with experience and complexity.In fact,
22 indexs can be divided into 2 classes, 3 classes, 4 classes even 22 classes.It stays and selects index number more, the accuracy for describing mood can be corresponding
Increase, and meets marginal decreasing effect, but system complexity also can be promoted accordingly.
As shown in Fig. 2 cluster map right half part, from right to left, categorical measure gradually increases, dotted line and map number of hits
Amount gradually increases, wherein number of intersections is quantity of classifying.Herein according to point 3 major class and 6 major class for example, in order to make table
It states and is more clear, in Fig. 2,2 classification dotted lines have been made.
If 22 indexs are divided into 3 major class, A1, A2, A3 can be divided by rightmost side dotted line in Fig. 2 and be total to three classes.Wherein 22
The classification ownership situation of a index is as shown in table 2, Fig. 2.
In A1, A2, A3 are all kinds of, sensitive factor Maximum Index is selected respectively as staying and selects index.Such as at 18 of A1 class
In index, sensitive factor maximum " investment index number CICSI " should be selected as such stay and select index.In 1 index of A2 class,
Sensitive factor maximum " Consumer Prices index " should be selected as such stay and select index.In 3 indexs of A3 class, Ying Xuan
Sensitive factor maximum " hand-off single order rate difference " is selected as such stay and selects index.Finally, it in the case of dividing 3 class, can obtain
Index is acted on behalf of to 3 investor sentiments.
Table 2
If 22 indexs are divided into 6 major class, B1, B2, B3, B4, B5, B6 totally six class can be divided by right side dotted line in Fig. 2.
Wherein the classification ownership situation of 22 indexs is as shown in table 3, Fig. 2.
In B1, B2, B3, B4, B5, B6 class, sensitive factor Maximum Index is selected respectively as staying and selects index.Such as in B1 class
5 indexs in, sensitive factor maximum " investment index number CICSI " should be selected as such stay and select index.In 2 indexs of B2 class
In, sensitive factor maximum " average return of IPO circulation number of share of stock weighting " should be selected as such stay and select index.In B3 class
In 11 indexs, sensitive factor maximum " last month open an account several logarithms " should be selected as such stay and select index.In 1 finger of B4 class
In mark, sensitive factor maximum " Consumer Prices index " should be selected as such stay and select index.In 2 indexs of B5 class
In, sensitive factor maximum " turnover rate first-order difference " should be selected as such stay and select index.In 1 index of B6 class, answer
Sensitive factor maximum " above demonstrate,proving composite earning rate " is selected to select index as such stay.It finally, can be in the case of point 6 class
It obtains 6 investor sentiments and acts on behalf of index.
Table 3
It is in fact possible to the quantity for acting on behalf of index be determined according to actual needs, as the dotted line in Fig. 2 moves from right to left
Dynamic, classification number gradually increases, and index system complexity increases.
In conclusion the screening criteria of investor sentiment measurement index is established, so that index screening more normative and reasonable；It opens
Formula system is put, screening range is improved, so that index is more representative；By a series of unrelated analyses, correlation analysis, cluster
Analysis, the system quantifies analysis of significance analysis, so that screening is more scientific, meanwhile, process visualizes completely, mentions for artificial decision
For foundation；Solid foundation has been established for the research of investor sentiment Synthetic Measurement, while being also financial regulation agencies in reality
Supervision and investor decision provide reference frame.
Claims (8)
1. a kind of stock market's investor sentiment index back-and-forth method based on cluster with maximum entropy delta ratio, which is characterized in that described
Method includes:
Mass-election module, any index related with stock market or investor can bring alternative index storehouse into；
Primary election module rejects data non-availability or the discontinuous invalid index of data；
Screening module rejects unrelated index by unrelated analysis, calculates each index and stock market's deep bid index and dividend yield rate
Related coefficient, available unrelated degree and significance test value reject the poor index of conspicuousness；
Select module, the index that amount of redundant information is greater than threshold value deleted by correlation analysis and clustering, calculate each index it
Between related coefficient, by related coefficient be greater than threshold value index sort out, delete information overlap index；
Preferred module, by calculating sensitive factor, i.e. comentropy is divided than mean entropy and maximum entropy than the ratio between mean entropy in conjunction with cluster
The tree-shaped classification map and index quantity complexity of analysis, finally stay and select the highest index of conspicuousness.
2. stock market's investor sentiment index back-and-forth method according to claim 1 based on cluster with maximum entropy delta ratio,
It is characterized in that, any index of the mass-election module can be current index, it is also possible to the derivative index of p phase in advance, i.e.,
For { An,An-1,An-2...An-p, wherein p is the positive integer greater than 1.
3. stock market's investor sentiment index back-and-forth method according to claim 1 based on cluster with maximum entropy delta ratio,
It is characterized in that, the unrelated degree relative coefficient of the screening module is 0.3 and following.
4. stock market's investor sentiment index back-and-forth method according to claim 1 based on cluster with maximum entropy delta ratio,
It is characterized in that, the screening module significance value is t statistic, which obeys the t distribution of n-2 freedom degree, probability
Boundary may be selected 1% to 5%.
5. stock market's investor sentiment index back-and-forth method according to claim 1 based on cluster with maximum entropy delta ratio,
It is characterized in that, the threshold value for selecting module is 0.9 or more.
6. stock market's investor sentiment index back-and-forth method according to claim 1 based on cluster with maximum entropy delta ratio,
It is characterized in that, the distance of the clustering is related coefficient distance.
7. stock market's investor sentiment index back-and-forth method according to claim 1 based on cluster with maximum entropy delta ratio,
It is characterized in that, sensitive factor can be used in the conspicuousness of the preferred module, i.e., comentropy is than mean entropy and maximum entropy than flat
The ratio between homoentropic differentiates
Z=ln (m), (3)
In formula, piFor the probability of discrete type system mode, i is system mode number, and m is system mode maximum value, and probability is worked as in definition
When being 0, entropy zero；H is the information entropy of index；For mean entropy；Z is maximum entropy；D is sensitive factor, i.e., comentropy is than flat
Homoentropic and maximum entropy are than the ratio between mean entropy.
8. stock market's investor sentiment index back-and-forth method according to claim 1 based on cluster with maximum entropy delta ratio,
It is characterized in that, the preferred module complexity can be stayed according to artificial experience and demand selects index, it should be in clustering map
In, classified by major class to group, in each category equilibrium selection index, the quantity of classification can be taking human as decision.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810701748.2A CN109118041A (en) | 2018-06-30 | 2018-06-30 | Stock market's investor sentiment index back-and-forth method based on cluster with maximum entropy delta ratio |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810701748.2A CN109118041A (en) | 2018-06-30 | 2018-06-30 | Stock market's investor sentiment index back-and-forth method based on cluster with maximum entropy delta ratio |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109118041A true CN109118041A (en) | 2019-01-01 |
Family
ID=64822373
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810701748.2A Pending CN109118041A (en) | 2018-06-30 | 2018-06-30 | Stock market's investor sentiment index back-and-forth method based on cluster with maximum entropy delta ratio |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109118041A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117172611A (en) * | 2023-09-27 | 2023-12-05 | 北京瑞风协同科技股份有限公司 | Method, system and equipment for evaluating all-machine fastener in design and manufacturing process |
-
2018
- 2018-06-30 CN CN201810701748.2A patent/CN109118041A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117172611A (en) * | 2023-09-27 | 2023-12-05 | 北京瑞风协同科技股份有限公司 | Method, system and equipment for evaluating all-machine fastener in design and manufacturing process |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Smith | The changing and relative efficiency of European emerging stock markets | |
| CN109034469A (en) | A kind of tourist flow prediction technique based on machine learning | |
| Kaurova et al. | Cross-country comparison of statistical indicators | |
| CN104850868A (en) | Customer segmentation method based on k-means and neural network cluster | |
| Egorov et al. | The role of universities in economic development of Russian regions | |
| CN112884013A (en) | Energy consumption partitioning method based on data mining technology | |
| Matenga | Assessment of energy market’s progress towards achieving Sustainable Development Goal 7: A clustering approach | |
| CN109146232A (en) | Stock market's investor sentiment of principal component and the coefficient of variation acts on behalf of index selection method | |
| CN109118041A (en) | Stock market's investor sentiment index back-and-forth method based on cluster with maximum entropy delta ratio | |
| CN109034545A (en) | Stock market's investor sentiment index selection method based on cluster with mean entropy difference group | |
| CN108961056A (en) | Stock market's investor sentiment based on entropy dimension and maximum entropy ratio acts on behalf of index selection method | |
| Tang et al. | The average environmental efficiency technique and its application to Chinese provincial panel data | |
| CN109190867A (en) | Index selection method is acted on behalf of based on cluster and stock market's investor sentiment of maximum entropy ratio | |
| Zongkeng et al. | Evaluation and analysis of socio-economic development level and management in Guangxi province of China | |
| CN109146231A (en) | Stock market's investor sentiment based on cluster and maximum entropy difference acts on behalf of index selection method | |
| CN109087202A (en) | Index selection method is acted on behalf of based on cluster and stock market's investor sentiment of comentropy group | |
| Alireza et al. | Evaluation of the Financial Ratio Capability to Predict the Financial Crisis of Companies. | |
| CN109118367A (en) | Stock market's investor sentiment index selection method based on cluster with maximum entropy production | |
| Bire | Efficiency analysis of hotels listed in the Indonesian stock exchange market | |
| CN109102148A (en) | Stock market's investor sentiment index selection method based on principal component and maximum entropy production | |
| CN109190868A (en) | Stock market's investor sentiment based on principal component and comentropy acts on behalf of index selection method | |
| CN108932666A (en) | Stock market's investor sentiment index selection method of entropy dimension and maximum entropy production group | |
| CN109102147A (en) | Principal component and stock market's investor sentiment of maximum entropy ratio act on behalf of index selection method | |
| CN109165805A (en) | Stock market's investor sentiment index selection method based on principal component and average entropy production | |
| CN108985592A (en) | Stock market's investor sentiment index selection method of principal component and maximum entropy production group |
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: 20190101 |