Fund similarity calculation method, platform, device and readable storage medium

Abstract

The present invention discloses a fund similarity calculation method, including: obtaining classification information and the latest holding information corresponding to a target fund, determining a preset number of target candidate funds; determining a first relevance of the target fund, and a second relevance of each target candidate fund; obtaining a first minimum spanning tree corresponding to the first relevance, and a second minimum spanning tree corresponding to the second relevance; determining the similarity between the first minimum spanning tree and each second minimum spanning tree, determining the target minimum spanning tree in each second minimum spanning tree, and using the target candidate fund corresponding to the target minimum spanning tree as a similar fund to the target fund. The present invention also discloses a fund similarity calculation platform, equipment and readable storage medium. The present invention determines the similarity through the minimum spanning tree corresponding to the relevance between the holding products, improves the accuracy of similarity calculation between funds and the processing efficiency of the computer, and improves the accuracy of screening similar funds.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a fund similarity calculation method, platform, device and readable storage medium. Background Art

[0002] Investment funds were born with the development of the securities market. As individual investors' investment concepts continue to shift towards value-oriented investment and professional investment, the fund market has developed rapidly in recent years. Although the development history is not long, the number of funds has grown rapidly. At present, the continuous emergence of popular funds often causes individual investors to be unable to subscribe or the actual amount available for purchase is insufficient. Therefore, it is often necessary to find alternative funds similar to popular funds to meet the needs of a large number of individual investors.

[0003] At present, similarity calculations are often performed on securities or funds based on the changes in the net asset value of the fund over a period of time in order to screen out alternative funds that are similar to the target fund. However, due to the influence of the correlation of the fund's trends over a period of time, there are multiple alternative funds with high similarities, resulting in low differentiation between funds and difficulty in screening alternative funds. At the same time, since only the changes in the net asset value of the fund are considered, the screened alternative funds have large differences from the target fund in terms of holdings and other dimensions, resulting in inaccurate screened alternative funds.

[0004] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of the present invention is to provide a fund similarity calculation method, platform, device and readable storage medium, aiming to solve the technical problem that the existing funds screened by similarity are inaccurate.

[0006] To achieve the above object, the present invention provides a fund similarity calculation method, which comprises the following steps:

[0007] Obtaining classification information and latest holding information corresponding to the target fund, determining the maximum weighted industry corresponding to the target fund based on the latest holding information, and determining a preset number of target candidate funds based on the classification information and the maximum weighted industry;

[0008] Determine a first correlation between first holdings corresponding to the target fund, and a second correlation between second holdings corresponding to each target candidate fund, wherein the first correlation is a correlation representing the annual returns of the holdings of the target fund, and the second correlation is a correlation representing the annual returns of the holdings of the target candidate funds;

[0009] Obtaining a first minimum spanning tree corresponding to the first correlation and a second minimum spanning tree corresponding to the second correlation;

[0010] The first minimum spanning tree is feature-compared with each second minimum spanning tree to determine the similarity between the first minimum spanning tree and each second minimum spanning tree, and the target minimum spanning tree corresponding to the maximum similarity among each similarity is determined in each second minimum spanning tree, and the target alternative fund corresponding to the target minimum spanning tree is used as a similar fund of the target fund.

[0011] In one embodiment, the step of obtaining a first minimum spanning tree corresponding to the first correlation and a second minimum spanning tree corresponding to the second correlation includes:

[0012] Determine a first correlation matrix corresponding to the target fund based on the first correlation, and determine a second correlation matrix corresponding to each target candidate fund based on the second correlation;

[0013] Based on the Kruskal algorithm of the minimum spanning tree, the first correlation matrix is ​​converted into a first minimum spanning tree, and each second correlation matrix is ​​converted into a second minimum spanning tree.

[0014] In one embodiment, the step of comparing the first minimum spanning tree with each second minimum spanning tree to determine the similarity between the first minimum spanning tree and each second minimum spanning tree includes:

[0015] Determine a first similarity between a root node of the first minimum spanning tree and a root node of each second minimum spanning tree, determine a second similarity between the number of layers of the first minimum spanning tree and the number of layers of each second minimum spanning tree, and determine a third similarity between the number of nodes in each layer of the first minimum spanning tree and the number of nodes in each layer of each second minimum spanning tree;

[0016] The similarities corresponding to the second minimum spanning trees are calculated based on the first similarity, the second similarity, and the third similarity.

[0017] In one embodiment, the step of determining the first correlation between the first holding products corresponding to the target fund and the second correlation between the second holding products corresponding to each target candidate fund includes:

[0018] Obtaining the first-year return vector corresponding to each first-positioned product, and determining the first correlation between each first-positioned product based on each first-year return vector;

[0019] The second annual return vector corresponding to the second holding product of each target candidate fund is obtained, and the second correlation between the second holding products of each target candidate fund is determined based on each second annual return vector.

[0020] In one embodiment, the step of determining the first correlation between the first holding products based on the first annual return vectors includes:

[0021] Determine the cosine values ​​of the angles between each pair of first-year return vectors, and determine the first correlation between each first-position product based on the cosine values ​​of the angles.

[0022] In one embodiment, the steps of obtaining classification information and latest holding information corresponding to the target fund, determining the maximum weighted industry corresponding to the target fund based on the latest holding information, and determining a preset number of target candidate funds based on the classification information and the maximum weighted industry include:

[0023] Obtaining the holding ratio of the first holding product according to the latest holding information;

[0024] Determine the industry holding ratio of each industry corresponding to the target fund based on the holding ratio of the first holding product, and take the industry corresponding to the maximum industry holding ratio among the industry holding ratios as the maximum weighted industry;

[0025] Acquire alternative funds that match the classification information, and determine a preset number of target alternative funds based on the alternative funds and the maximum weighted industry.

[0026] In one embodiment, the step of determining a preset number of target candidate funds based on the candidate funds and the industry with the largest weight includes:

[0027] Obtain the industries with the largest proportions corresponding to each candidate fund, and determine the funds to be screened whose industries with the largest proportions and the industries with the largest weights are consistent among the candidate funds;

[0028] According to the first holding product corresponding to the fund to be screened and the second holding product corresponding to the target fund, the overlap degree of each concept exposure list between each fund to be screened and the target fund is determined, and a preset number of funds with the largest overlap degree of concept exposure list among the funds to be screened are taken as target candidate funds.

[0029] In addition, to achieve the above-mentioned purpose, the present invention further provides a fund similarity calculation platform, the fund similarity calculation platform comprising:

[0030] A first acquisition module is used to acquire classification information and latest holding information corresponding to the target fund, determine the maximum weighted industry corresponding to the target fund based on the latest holding information, and determine a preset number of target candidate funds based on the classification information and the maximum weighted industry;

[0031] A first determination module is used to determine a first correlation between first holdings corresponding to the target fund, and a second correlation between second holdings corresponding to each target candidate fund, wherein the first correlation is a correlation representing the annual returns of the holdings of the target fund, and the second correlation is a correlation representing the annual returns of the holdings of the target candidate funds;

[0032] A second acquisition module, used to acquire a first minimum spanning tree corresponding to the first correlation and a second minimum spanning tree corresponding to the second correlation;

[0033] The second determination module is used to compare the features of the first minimum spanning tree with each second minimum spanning tree to determine the similarity between the first minimum spanning tree and each second minimum spanning tree, determine the target minimum spanning tree corresponding to the maximum similarity among each similarity, and use the target alternative fund corresponding to the target minimum spanning tree as a similar fund of the target fund.

[0034] In addition, to achieve the above-mentioned purpose, the present invention also provides a fund similarity calculation device, which includes: a memory, a processor, and a fund similarity calculation program stored in the memory and executable on the processor, and the fund similarity calculation program implements the steps of the aforementioned fund similarity calculation method when executed by the processor.

[0035] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which a fund similarity calculation program is stored, and when the fund similarity calculation program is executed by a processor, the steps of the aforementioned fund similarity calculation method are implemented.

[0036] The present invention obtains classification information and the latest holding information corresponding to the target fund, determines the maximum weight industry corresponding to the target fund based on the latest holding information, and determines a preset number of target candidate funds based on the classification information and the maximum weight industry, then determines the first correlation between the first holding products corresponding to the target fund, and the second correlation between the second holding products corresponding to each target candidate fund, and then obtains the first minimum spanning tree corresponding to the first correlation, and the second minimum spanning tree corresponding to the second correlation, and then performs feature comparison between the first minimum spanning tree and each second minimum spanning tree to determine the similarity between the first minimum spanning tree and each second minimum spanning tree, and determines each similarity. The target minimum spanning tree corresponding to the maximum similarity in is obtained, and the target candidate fund corresponding to the target minimum spanning tree is used as the similar fund of the target fund. The similar funds are determined by the target fund's holdings, the maximum weighted industry, and the target candidate fund's holdings, so that the similar funds and the target fund have a high degree of industry correlation. The industry classification of the funds is used to improve the distinction between the funds to be screened, so that the similar funds are closer to the target fund, and the future trends of the similar funds are closer to the future trends of the target fund. The similarity is determined by the minimum spanning tree corresponding to the correlation between the holdings, which improves the accuracy of the similarity calculation between the funds and the computer's processing efficiency, thereby improving the accuracy of screening similar funds. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of the structure of a fund similarity calculation device in a hardware operating environment involved in an embodiment of the present invention;

[0038] Figure 2 It is a flow chart of the first embodiment of the fund similarity calculation method of the present invention;

[0039] Figure 3 A detailed flow chart of the steps of obtaining a first minimum spanning tree corresponding to a first correlation and a second minimum spanning tree corresponding to a second correlation in the second embodiment of the fund similarity calculation method of the present invention;

[0040] Figure 4 A detailed flow chart of the step of comparing the features of the first minimum spanning tree with each second minimum spanning tree to determine the similarity between the first minimum spanning tree and each second minimum spanning tree in the third embodiment of the fund similarity calculation method of the present invention;

[0041] Figure 5 This is a schematic diagram of the functional modules of an embodiment of a fund similarity calculation platform of the present invention.

[0042] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0043] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0044] like Figure 1 As shown, Figure 1 It is a schematic diagram of the structure of a fund similarity calculation device in a hardware operating environment involved in an embodiment of the present invention.

[0045] The fund similarity calculation device of the embodiment of the present invention may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0046] Those skilled in the art will understand that Figure 1 The fund similarity calculation device structure shown in the figure does not constitute a limitation on the fund similarity calculation device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0047] like Figure 1 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a fund similarity calculation program.

[0048] exist Figure 1 In the fund similarity calculation device shown, the network interface 1004 is mainly used to connect to the background server and communicate data with the background server; the user interface 1003 is mainly used to connect to the client (user end) and communicate data with the client; and the processor 1001 can be used to call the fund similarity calculation program stored in the memory 1005, and perform the operations in the various embodiments of the following fund similarity calculation method.

[0049] The present invention also provides a fund similarity calculation method, referring to Figure 2 , Figure 2 It is a flowchart of the first embodiment of the fund similarity calculation method of the present invention.

[0050] In this embodiment, the fund similarity calculation method includes:

[0051] Step S100, obtaining classification information and latest holding information corresponding to the target fund, determining the maximum weighted industry corresponding to the target fund based on the latest holding information, and determining a preset number of target candidate funds based on the classification information and the maximum weighted industry;

[0052] In this embodiment, the target fund can be a fund whose annual yield reaches the first preset yield, so as to facilitate users to screen out funds similar to the popular fund. Alternatively, the target fund can also be some funds with poor performance, such as funds with an annual yield less than the second preset yield, so that users can avoid screened out similar funds. The first preset yield can be 30%, 50%, etc., and the second preset yield can be 5%.

[0053] In this embodiment, for each target fund, first obtain the current classification information and the latest holding information of the target fund, the classification information includes stock funds, bond funds, money market funds, futures funds, or open-end funds and closed-end funds, etc.; for example, obtain the classification information corresponding to the target fund through the basic information of the fund, and obtain the latest holding information of the target fund (such as the latest quarterly holding) from the latest disclosure information of the target fund. According to the latest holding information, obtain all the first holding products of the target fund, the first holding products include stocks, bonds, funds, etc., and determine the industry corresponding to each first holding product and the holding ratio of the first holding product in the target fund. Then, according to the latest holding information of the target fund, the industry in which the first holding product held by the target fund is located is mapped to the industry in which the target fund is located, that is, all industries corresponding to the target fund (i.e., industries corresponding to each holding product) are determined according to each first holding product, and the industry holding ratio corresponding to each industry is determined according to the holding ratio of the first holding product. For each industry, the industry holding ratio is the sum of the holding ratios of the holding products belonging to the industry. Then, determine the largest industry holding ratio among all industry holding ratios, and take the industry corresponding to the largest industry holding ratio as the industry with the largest weight.

[0054] Then, based on the classification information and the industry with the largest weight, a preset number of target alternative funds are determined. For example, among all the funds (except the target fund), funds whose classification information is consistent with that of the target fund are selected as candidate funds, and a preset number of funds including the industry with the maximum weight are selected as target candidate funds among the candidate funds; or, the candidate funds including the industry with the maximum weight are sorted from large to small according to the holding ratio corresponding to the industry with the maximum weight, and a preset number of funds with the largest holding ratio corresponding to the industry with the maximum weight selected in the sorting are selected as target candidate funds; or, the maximum proportion industry corresponding to each candidate fund is obtained, and the funds to be screened whose maximum proportion industry is consistent with the maximum weight industry among the candidate funds are determined, and the funds to be screened are sorted from large to small according to the holding ratio corresponding to the industry with the maximum weight, and a preset number of funds with the largest holding ratio corresponding to the industry with the maximum weight selected in the sorting are selected as target candidate funds; or, the maximum proportion industry corresponding to each candidate fund is obtained, and the funds to be screened whose maximum proportion industry is consistent with the maximum weight industry among the candidate funds are determined, and the absolute value of the difference between the industry holding ratio corresponding to the maximum proportion industry of each candidate fund and the industry holding ratio corresponding to the maximum weight industry of the target fund is calculated, and a preset number of funds with the smallest absolute value among the funds to be screened are selected as target candidate funds.

[0055] Step S200, determining a first correlation between first holdings corresponding to the target fund, and a second correlation between second holdings corresponding to each target candidate fund, wherein the first correlation is a correlation representing the annual returns of the holdings of the target fund, and the second correlation is a correlation representing the annual returns of the holdings of the target candidate funds;

[0056] In this embodiment, after obtaining the target candidate fund, the first correlation between the first holding products corresponding to the target fund is obtained, and the first annual return vector corresponding to the first holding product of the target fund is obtained, wherein the first annual return vector is an annual return time series, and the annual return time series is a time series of yields for one year, that is, a sequence of the yields of each trading day in one year sorted in chronological order, and the vector corresponding to the annual return time series is the first yield time series vector corresponding to the first holding product. Then, based on each first annual return vector, the first correlation between each first holding product is determined; specifically, the first angle cosine value between each first annual return vector is calculated, and the first angle cosine value is normalized to map each first angle cosine value to (0, 1), for example, the first angle cosine value is normalized based on the range of (0, 1) to obtain the first mapping value of each first angle cosine value, and each first mapping value is used as the first correlation between the corresponding first holding products.

[0057] Similarly, for each target alternative fund, the second holding product of each target alternative fund is obtained, and the second annual return vector corresponding to each second holding product is obtained, and then, the second correlation between each second holding product is determined based on each second annual return vector; specifically, the second angle cosine value between each second annual return vector is calculated, and the second angle cosine value is normalized to map each second angle cosine value to (0, 1), for example, the second angle cosine value is normalized based on the range of (0, 1) to obtain the second mapping value of each second angle cosine value, and each second mapping value is used as the second correlation between the corresponding second holding products, thereby obtaining the second correlation of the target alternative funds.

[0058] Step S300, obtaining a first minimum spanning tree corresponding to the first correlation and a second minimum spanning tree corresponding to the second correlation;

[0059] In this embodiment, a minimum spanning tree algorithm can be used to generate a corresponding first minimum spanning tree according to the first correlation, and a corresponding second minimum spanning tree can be generated according to each second correlation, wherein the minimum spanning tree algorithm can adopt Prim's algorithm or Kruskal's algorithm.

[0060] Step S400, compare the features of the first minimum spanning tree with each second minimum spanning tree to determine the similarity between the first minimum spanning tree and each second minimum spanning tree, determine the target minimum spanning tree corresponding to the maximum similarity among each similarity in each second minimum spanning tree, and use the target candidate fund corresponding to the target minimum spanning tree as a similar fund of the target fund.

[0061] In this embodiment, when the first minimum spanning tree and each second minimum spanning tree are obtained, a feature comparison is performed on the first minimum spanning tree and each second minimum spanning tree, wherein the features for feature comparison include the root node, the number of layers, the number of nodes in each layer, etc. The feature similarity corresponding to each feature is determined based on the comparison result, and the similarity between the first minimum spanning tree and each second minimum spanning tree is determined based on the feature similarity.

[0062] Then, after obtaining the similarity between the first minimum spanning tree and each second minimum spanning tree, each similarity is compared to obtain the maximum similarity among each similarity, and the second minimum spanning tree corresponding to the maximum similarity is obtained in each second minimum spanning tree as the target minimum spanning tree, and then the target alternative fund corresponding to the target minimum spanning tree is used as a similar fund of the target fund.

[0063] The fund similarity calculation method proposed in this embodiment obtains the classification information and the latest holding information corresponding to the target fund, determines the maximum weight industry corresponding to the target fund based on the latest holding information, and determines a preset number of target candidate funds based on the classification information and the maximum weight industry, then determines the first correlation between the first holding products corresponding to the target fund, and the second correlation between the second holding products corresponding to each target candidate fund, and then obtains the first minimum spanning tree corresponding to the first correlation and the second minimum spanning tree corresponding to the second correlation, and then performs feature comparison between the first minimum spanning tree and each second minimum spanning tree to determine the similarity between the first minimum spanning tree and each second minimum spanning tree. Determine the target minimum spanning tree corresponding to the maximum similarity among various similarities, and use the target candidate fund corresponding to the target minimum spanning tree as a similar fund to the target fund. Determine similar funds through the target fund's holdings, the industry with the largest weight, and the target candidate fund's holdings, so that similar funds and the target fund have a high degree of industry correlation. Improve the distinction between funds to be screened through the industry classification of funds, make similar funds closer to the target fund, and make the future trends of similar funds closer to the future trends of the target fund. Determine similarity through the minimum spanning tree corresponding to the correlation between holdings, improve the accuracy of similarity calculation between funds and computer processing efficiency, and improve the accuracy of screening similar funds.

[0064] Based on the first embodiment, a second embodiment of the fund similarity calculation method of the present invention is proposed, referring to Figure 3 In this embodiment, step S300 includes:

[0065] Step S310, determining a first correlation matrix corresponding to the target fund based on the first correlation, and determining a second correlation matrix corresponding to each target candidate fund based on the second correlation;

[0066] Step S320: Based on the minimum spanning tree Kruskal algorithm, the first correlation matrix is ​​converted into a first minimum spanning tree, and each second correlation matrix is ​​converted into a second minimum spanning tree.

[0067] In this embodiment, after obtaining the first correlation and the second correlation, the first correlation matrix and the second correlation matrix corresponding to each target candidate fund are determined. Specifically, the first correlation matrix is ​​obtained by using a matrix to store the first correlation, that is, each first correlation is stored in a preset matrix to obtain the first correlation matrix. For example, each row of the first correlation matrix is ​​the correlation corresponding to product 1, the correlation corresponding to product 2...the correlation corresponding to product N, and each column is the correlation corresponding to product 1, the correlation corresponding to product 2...the correlation corresponding to product N, then the elements of the first row of the first correlation matrix are the correlation between product 1 and product 1, the correlation between product 1 and product 2, the correlation between product 1 and product 3...the correlation between product 1 and product n; at the same time, the second correlation corresponding to each target candidate fund is respectively stored in a matrix, and the second correlation corresponding to each target candidate fund is respectively stored in the preset matrix to obtain the second correlation matrix corresponding to each target candidate fund, wherein the element structure of the second correlation matrix is ​​similar to the element structure of the first correlation matrix.

[0068] Then, based on the Kruskal algorithm of the minimum spanning tree, the first correlation matrix is ​​converted into the first minimum spanning tree, and each second correlation matrix is ​​converted into the second minimum spanning tree, so as to replace the holding information of each fund with the minimum spanning tree. Specifically, the first correlation matrix is ​​input into the Kruskal algorithm, and the output of the Kruskal algorithm is used as the first minimum spanning tree, and the second correlation matrix is ​​input into the Kruskal algorithm, and the output of the Kruskal algorithm is used as the second minimum spanning tree. For example, the first correlation matrix is ​​input into the conversion script corresponding to the existing Kruskal algorithm, and the output of the conversion script is the first minimum spanning tree.

[0069] The fund similarity calculation method proposed in this embodiment is as follows: Based on the first correlation, a first correlation matrix corresponding to the target fund is determined, and based on the second correlation, a second correlation matrix corresponding to each target candidate fund is determined; then, based on the Kruskal algorithm of the minimum spanning tree, the first correlation matrix is ​​converted into a first minimum spanning tree, and each second correlation matrix is ​​converted into a second minimum spanning tree respectively, by converting the correlation into a correlation matrix and converting the correlation matrix into a minimum spanning tree through the Kruskal algorithm, so as to replace the holding information of each fund with the minimum spanning tree, so as to facilitate the calculation of the similarity between the target fund and the target candidate fund, and improve the efficiency and accuracy of the fund similarity calculation.

[0070] Based on the first embodiment, a third embodiment of the fund similarity calculation method of the present invention is proposed, referring to Figure 4 In this embodiment, step S400 includes:

[0071] Step S410, determining a first similarity between a root node of the first minimum spanning tree and a root node of each second minimum spanning tree, determining a second similarity between the number of layers of the first minimum spanning tree and the number of layers of each second minimum spanning tree, and determining a third similarity between the number of nodes in each layer of the first minimum spanning tree and the number of nodes in each layer of each second minimum spanning tree;

[0072] Step S420: Calculate the similarities corresponding to the second minimum spanning trees based on the first similarity, the second similarity, and the third similarity.

[0073] In this embodiment, when determining the first similarity, the first similarity corresponding to the spanning tree whose root node in the second minimum spanning tree is the same as the root node of the first minimum spanning tree can be set to a first preset value, such as 1, and the first similarity corresponding to the spanning tree with different root nodes can be set to a second preset value, such as 0. Alternatively, the root node difference between the root node in each second minimum spanning tree and the root node of the first minimum spanning tree is calculated, and the root node difference is divided by the root node of the first minimum spanning tree to obtain the root node ratio corresponding to each second minimum spanning tree, and each root node ratio is normalized to obtain each first similarity.

[0074] Next, determine the second similarity between the number of layers of the first minimum spanning tree and the number of layers of each second minimum spanning tree. Specifically, the difference in the number of layers between the first minimum spanning tree and the number of layers of each second minimum spanning tree can be calculated, and the difference in the number of layers can be divided by the number of layers of the first minimum spanning tree to obtain the number of layers ratio, and 1-the number of layers ratio is used as the second similarity, and the number of layers difference is an absolute value.

[0075] Then, determine the third similarity between the number of nodes in each layer of the first minimum spanning tree and the number of nodes in each layer of each second minimum spanning tree. Specifically, for each second minimum spanning tree, first calculate the node deviation rate between each layer of the second minimum spanning tree and the corresponding each layer of the first minimum spanning tree, and take the average of the node deviation rates corresponding to each layer as the third similarity. For example, for each layer, first calculate the node difference (absolute value) between the number of nodes of the target layer in the first minimum spanning tree and the number of nodes of the target layer in the second minimum spanning tree, divide the node difference by the number of nodes in the first minimum spanning tree of the layer to obtain the node difference ratio, and take 1-the node difference ratio as the node deviation rate of the target layer.

[0076] Finally, the similarity corresponding to each second minimum spanning tree is calculated based on the first similarity, the second similarity and the third similarity. Specifically, the similarity corresponding to each second minimum spanning tree is calculated by the weights corresponding to the root node, the number of layers and the number of nodes in each layer, and the first similarity, the second similarity and the third similarity. For example, similarity = first similarity * corresponding weight + second similarity * corresponding weight + third similarity * corresponding weight.

[0077] The fund similarity calculation method proposed in this embodiment determines the first similarity between the root node of the first minimum spanning tree and the root nodes of each second minimum spanning tree, determines the second similarity between the number of layers of the first minimum spanning tree and the number of layers of each second minimum spanning tree, and determines the third similarity between the number of nodes in each layer of the first minimum spanning tree and the number of nodes in each layer of each second minimum spanning tree; then calculates the similarity corresponding to each second minimum spanning tree based on the first similarity, the second similarity and the third similarity, and can accurately obtain the similarity between the target fund and the target alternative fund according to the minimum spanning tree, thereby improving the accuracy of the similarity calculation, and then improving the accuracy of the screening of similar funds, so that the future trend of similar funds is closer to the future trend of the target fund.

[0078] Based on the first embodiment, a fourth embodiment of the fund similarity calculation method of the present invention is proposed. In this embodiment, step S200 includes:

[0079] Step S210, obtaining the first-year return vector corresponding to each first-positioned product, and determining the first correlation between each first-positioned product based on each first-year return vector;

[0080] Step S220, obtaining the second annual return vector corresponding to the second holding product of each target candidate fund, and determining the second correlation between the second holding products of each target candidate fund based on each second annual return vector.

[0081] In this embodiment, after obtaining the target candidate fund, the first annual return vector corresponding to each first holding product is obtained, and the first correlation between each first holding product is determined based on each first annual return vector, wherein the first annual return vector is an annual return time series, and the annual return time series is a one-year yield time series, that is, a sequence of the yields of each trading day in a year sorted in chronological order, and the vector corresponding to the annual return time series is the first yield time series vector corresponding to the first holding product. Then, the first angle cosine value between each first annual return vector is calculated, and the first angle cosine value is normalized to map each first angle cosine value to (0, 1), for example, the first angle cosine value is normalized based on the range of (0, 1) to obtain the first mapping value of each first angle cosine value, and each first mapping value is used as the first correlation between the corresponding first holding products.

[0082] Similarly, for each target alternative fund, the second annual return vector corresponding to the second holding product of each target alternative fund is obtained, and the second correlation between the second holding products of each target alternative fund is determined based on each second annual return vector; specifically, the second angle cosine value between each second annual return vector is calculated, and the second angle cosine value is normalized to map each second angle cosine value to (0, 1). For example, the second angle cosine value is normalized based on the range of (0, 1) to obtain the second mapping value of each second angle cosine value, and each second mapping value is used as the second correlation between the corresponding second holding products, thereby obtaining the second correlation of the target alternative funds.

[0083] Furthermore, in one embodiment, step S220 includes: determining the cosine value of the angle between each pair of first-year return vectors, and determining the first correlation between each first holding product based on the cosine value of the angle.

[0084] The fund similarity calculation method proposed in this embodiment obtains the first-year return vector corresponding to each first holding product, and determines the first correlation between each first holding product based on each first-year return vector; then obtains the second-year return vector corresponding to the second holding product of each target candidate fund, and determines the second correlation between the second holding products of each target candidate fund based on each second-year return vector. The first correlation and the second correlation can be accurately obtained according to the holding products of the target fund and the holding products of the target candidate fund. Through the holding products, the future trend of the obtained similar funds is closer to the future trend of the target fund, thereby improving the accuracy of screening similar funds.

[0085] Based on the above embodiments, a fifth embodiment of the fund similarity calculation method of the present invention is proposed. In this embodiment, step S100 includes:

[0086] Step S110, obtaining the holding ratio of the first holding product according to the latest holding information;

[0087] Step S120, determining the industry holding ratios of various industries corresponding to the target fund based on the holding ratios of the first holding product, and taking the industry corresponding to the maximum industry holding ratio among the industry holding ratios as the maximum weighted industry;

[0088] Step S130, obtaining candidate funds that match the classification information, and determining a preset number of target candidate funds based on the candidate funds and the maximum weighted industry.

[0089] In this embodiment, all first holding products of the target fund are obtained according to the latest holding information, and the first holding products include stocks, bonds, funds, etc., and the industries corresponding to each first holding product and the holding ratio of the first holding product in the target fund are determined. Then, according to the latest holding information of the target fund, the industries in which the first holding products held by the target fund are located are mapped to the industries in which the target fund is located, that is, all industries corresponding to the target fund (that is, the industries corresponding to each holding product) are determined according to each first holding product, and the industry holding ratio corresponding to each industry is determined according to the holding ratio of the first holding product. For each industry, the industry holding ratio is the sum of the holding ratios of the holding products belonging to the industry. Then, the maximum industry holding ratio among the industry holding ratios is determined, and the industry corresponding to the maximum industry holding ratio is used as the industry with the maximum weight.

[0090] Then, alternative funds matching the classification information are obtained, and a preset number of target alternative funds are determined based on the alternative funds and the industries with the largest weights.

[0091] The fund similarity calculation method proposed in this embodiment obtains the holding ratio of the first holding product according to the latest holding information; then determines the industry holding ratio of each industry corresponding to the target fund based on the holding ratio of the first holding product, and takes the industry corresponding to the maximum industry holding ratio among the industry holding ratios as the maximum weighted industry; then obtains the alternative funds matching the classification information, and determines a preset number of target alternative funds based on the alternative funds and the maximum weighted industry, and accurately screens the target alternative funds according to the holding products of the target funds, thereby improving the correlation between the target alternative funds and the target funds, thereby improving the accuracy of similar fund screening.

[0092] Based on the fifth embodiment, a sixth embodiment of the fund similarity calculation method of the present invention is proposed. In this embodiment, step S130 includes:

[0093] Step S131, obtaining the maximum proportion industry corresponding to each candidate fund, and determining the fund to be screened whose maximum proportion industry and maximum weight industry are consistent among the candidate funds;

[0094] Step S132, based on the first holding product corresponding to the fund to be screened and the second holding product corresponding to the target fund, determine the overlap of each concept exposure list between each fund to be screened and the target fund, and select a preset number of funds with the largest overlap of concept exposure lists among the funds to be screened as target candidate funds.

[0095] In this embodiment, the maximum proportion of industries corresponding to each candidate fund is obtained. Specifically, according to the latest holding information of each candidate fund, the industry of the second holding product held by the candidate fund is mapped to the industry of the candidate fund, that is, all industries corresponding to the candidate fund (that is, the industries corresponding to each second holding product) are determined according to each second holding product of the candidate fund, and the industry holding ratio corresponding to each industry of the candidate fund is determined according to the holding ratio of the second holding product. The maximum industry holding ratio among the industry holding ratios corresponding to each industry of the candidate fund is determined, and the industry corresponding to the maximum industry holding ratio of the candidate fund is used as the industry with the maximum proportion. Then, the fund to be screened whose maximum proportion industry is consistent with the maximum weighted industry in the candidate fund is determined.

[0096] Then, based on the first holding product corresponding to the fund to be screened and the second holding product corresponding to the target fund, the overlap degree of each concept exposure list between each fund to be screened and the target fund is determined. For example, the number of holding products that are consistent between the fund to be screened and the target fund can be divided by the number of first holding products corresponding to the target fund to obtain the overlap degree of the concept exposure list, and a preset number of funds with the largest overlap degree of concept exposure lists among the funds to be screened are used as target candidate funds.

[0097] The fund similarity calculation method proposed in this embodiment obtains the maximum proportion industry corresponding to each candidate fund, and determines the funds to be screened whose maximum proportion industry and maximum weight industry are consistent among the candidate funds; then, based on the first holding product corresponding to the fund to be screened and the second holding product corresponding to the target fund, the overlap of each concept exposure list between each fund to be screened and the target fund is determined, and a preset number of funds with the largest overlap in the concept exposure list among the funds to be screened are used as target candidate funds. The target candidate funds are screened through the maximum proportion industry and the overlap of the concept exposure list, thereby improving the correlation between the target candidate funds and the target fund, and thereby improving the accuracy of similar fund screening.

[0098] Reference Figure 5 , Figure 5 This is a schematic diagram of the functional modules of an embodiment of a fund similarity calculation platform of the present invention.

[0099] In this embodiment, the fund similarity calculation platform includes:

[0100] The first acquisition module 10 is used to obtain the classification information and the latest holding information corresponding to the target fund, determine the maximum weight industry corresponding to the target fund based on the latest holding information, and determine a preset number of target candidate funds based on the classification information and the maximum weight industry;

[0101] The first determination module 20 is used to determine a first correlation between first holding products corresponding to the target fund, and a second correlation between second holding products corresponding to each target candidate fund, wherein the first correlation is a correlation representing the annual returns of the holding products of the target fund, and the second correlation is a correlation representing the annual returns of the holding products of the target candidate funds;

[0102] A second acquisition module 30, used to acquire a first minimum spanning tree corresponding to the first correlation and a second minimum spanning tree corresponding to the second correlation;

[0103] The second determination module 40 is used to compare the features of the first minimum spanning tree with each second minimum spanning tree to determine the similarity between the first minimum spanning tree and each second minimum spanning tree, determine the target minimum spanning tree corresponding to the maximum similarity among each similarity, and use the target alternative fund corresponding to the target minimum spanning tree as a similar fund of the target fund.

[0104] Furthermore, the second acquisition module 30 is further configured to:

[0105] Determine a first correlation matrix corresponding to the target fund based on the first correlation, and determine a second correlation matrix corresponding to each target candidate fund based on the second correlation;

[0106] Based on the Kruskal algorithm of the minimum spanning tree, the first correlation matrix is ​​converted into a first minimum spanning tree, and each second correlation matrix is ​​converted into a second minimum spanning tree.

[0107] Furthermore, the second determining module 40 is further configured to:

[0108] Determine a first similarity between a root node of the first minimum spanning tree and a root node of each second minimum spanning tree, determine a second similarity between the number of layers of the first minimum spanning tree and the number of layers of each second minimum spanning tree, and determine a third similarity between the number of nodes in each layer of the first minimum spanning tree and the number of nodes in each layer of each second minimum spanning tree;

[0109] The similarities corresponding to the second minimum spanning trees are calculated based on the first similarity, the second similarity, and the third similarity.

[0110] Furthermore, the first determining module 20 is further configured to:

[0111] Obtaining the first-year return vector corresponding to each first-positioned product, and determining the first correlation between each first-positioned product based on each first-year return vector;

[0112] The second annual return vector corresponding to the second holding product of each target candidate fund is obtained, and the second correlation between the second holding products of each target candidate fund is determined based on each second annual return vector.

[0113] Furthermore, the first determining module 20 is further configured to:

[0114] Determine the cosine values ​​of the angles between each pair of first-year return vectors, and determine the first correlation between each first-position product based on the cosine values ​​of the angles.

[0115] Furthermore, the first acquisition module 10 is further configured to:

[0116] Obtaining the holding ratio of the first holding product according to the latest holding information;

[0117] Determine the industry holding ratio of each industry corresponding to the target fund based on the holding ratio of the first holding product, and take the industry corresponding to the maximum industry holding ratio among the industry holding ratios as the maximum weighted industry;

[0118] Acquire alternative funds that match the classification information, and determine a preset number of target alternative funds based on the alternative funds and the maximum weighted industry.

[0119] Furthermore, the first acquisition module 10 is further configured to:

[0120] Obtain the industries with the largest proportions corresponding to each candidate fund, and determine the funds to be screened whose industries with the largest proportions and the industries with the largest weights are consistent among the candidate funds;

[0121] According to the first holding product corresponding to the fund to be screened and the second holding product corresponding to the target fund, the overlap degree of each concept exposure list between each fund to be screened and the target fund is determined, and a preset number of funds with the largest overlap degree of concept exposure list among the funds to be screened are taken as target candidate funds.

[0122] It should be noted that the various embodiments of the fund similarity calculation platform are basically the same as the various embodiments of the above-mentioned fund similarity calculation method, and will not be described in detail here.

[0123] In addition, an embodiment of the present invention further proposes a computer-readable storage medium, on which a fund similarity calculation program is stored. When the fund similarity calculation program is executed by a processor, the steps of the fund similarity calculation method as described in any one of the above are implemented.

[0124] The specific embodiments of the computer-readable storage medium of the present invention are basically the same as the embodiments of the above-mentioned fund similarity calculation method, and will not be described in detail here.

[0125] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0126] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0127] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0128] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A fund similarity calculation method, characterized in that: The fund similarity calculation method comprises the following steps: Obtaining classification information and latest holding information corresponding to the target fund, determining the maximum weighted industry corresponding to the target fund based on the latest holding information, and determining a preset number of target candidate funds based on the classification information and the maximum weighted industry; Determine a first correlation between the first holdings corresponding to the target fund, and a second correlation between the second holdings corresponding to each target candidate fund, wherein the first correlation is a correlation representing the annual returns of the holdings of the target fund, and the second correlation is a correlation representing the annual returns of the holdings of the target candidate funds; Obtaining a first minimum spanning tree corresponding to the first correlation and a second minimum spanning tree corresponding to the second correlation; The first minimum spanning tree is compared with each of the second minimum spanning trees to determine the similarity between the first minimum spanning tree and each of the second minimum spanning trees, a target minimum spanning tree corresponding to the maximum similarity among the similarities is determined in each of the second minimum spanning trees, and the target candidate fund corresponding to the target minimum spanning tree is used as a similar fund to the target fund; wherein, Determining the maximum weighted industry corresponding to the target fund based on the latest holding information includes: acquiring all first holding products of the target fund according to the latest holding information, the first holding products including stocks, bonds, and funds, and determining the industry corresponding to each first holding product and the holding ratio of the first holding product in the target fund; then, according to the latest holding information of the target fund, mapping the industry of the first holding products held by the target fund to the industry in which the target fund is located, that is, determining all industries corresponding to the target fund according to each first holding product, and determining the industry holding ratio corresponding to each industry according to the holding ratio of the first holding product, for each industry, the industry holding ratio is the sum of the holding ratios of the holding products belonging to the industry; then, determining the maximum industry holding ratio among the industry holding ratios, and taking the industry corresponding to the maximum industry holding ratio as the maximum weighted industry; Determining the first correlation between the first holdings corresponding to the target fund includes: obtaining the first annual return vector of the first holding product corresponding to the target fund, the first annual return vector is an annual return time series, the annual return time series is a one-year yield time series, that is, a sequence of the yields of each trading day in a year sorted in chronological order, and the vector corresponding to the annual return time series is the first yield time series vector corresponding to the first holding product; then, determining the first correlation between each first holding product based on each first annual return vector: calculating the first angle cosine value between each first annual return vector, normalizing each first angle cosine value to map each first angle cosine value to (0, 1), obtaining the first mapping value of each first angle cosine value, and using each first mapping value as the first correlation between the corresponding first holding products; Determining the second correlation between the second holdings corresponding to each target alternative fund includes: for each target alternative fund, obtaining the second holdings of each target alternative fund, and obtaining the second annual return vector corresponding to each second holding, and then determining the second correlation between each second holding based on each second annual return vector: calculating the second angle cosine value between each second annual return vector, normalizing each second angle cosine value to map each second angle cosine value to (0, 1), obtaining the second mapping value of each second angle cosine value, using each second mapping value as the second correlation between the corresponding second holdings, and then obtaining the second correlation between the second holdings of the target alternative funds.

2. The fund similarity calculation method according to claim 1, characterized in that: The step of obtaining a first minimum spanning tree corresponding to the first correlation and a second minimum spanning tree corresponding to the second correlation comprises: Determine a first correlation matrix corresponding to the target fund based on the first correlation, and determine a second correlation matrix corresponding to each of the target candidate funds based on the second correlation; Based on the Kruskal algorithm of the minimum spanning tree, the first correlation matrix is ​​converted into a first minimum spanning tree, and each of the second correlation matrices is converted into a second minimum spanning tree.

3. The fund similarity calculation method according to claim 1, characterized in that: The step of comparing the first minimum spanning tree with each of the second minimum spanning trees to determine the similarity between the first minimum spanning tree and each of the second minimum spanning trees comprises: Determine a first similarity between a root node of the first minimum spanning tree and a root node of each of the second minimum spanning trees, determine a second similarity between the number of layers of the first minimum spanning tree and the number of layers of each of the second minimum spanning trees, and determine a third similarity between the number of nodes in each layer of the first minimum spanning tree and the number of nodes in each layer of each of the second minimum spanning trees; Similarities corresponding to each second minimum spanning tree are calculated based on the first similarity, the second similarity, and the third similarity.

4. The fund similarity calculation method according to any one of claims 1 to 3, characterized in that: The steps of obtaining classification information and latest holding information corresponding to the target fund, determining the maximum weighted industry corresponding to the target fund based on the latest holding information, and determining a preset number of target candidate funds based on the classification information and the maximum weighted industry include: Acquire the holding ratio of the first holding product according to the latest holding information; Determine the industry holding ratio of each industry corresponding to the target fund based on the holding ratio of the first holding product, and take the industry corresponding to the maximum industry holding ratio among the industry holding ratios as the maximum weighted industry; Acquire alternative funds that match the classification information, and determine a preset number of target alternative funds based on the alternative funds and the maximum weighted industry.

5. The fund similarity calculation method according to claim 4, characterized in that: The step of determining a preset number of target candidate funds based on the candidate funds and the maximum weighted industry comprises: Obtain the industry with the largest proportion corresponding to each of the candidate funds, and determine the fund to be screened whose industry with the largest proportion among the candidate funds is consistent with the industry with the largest weight; Based on the first holding product corresponding to the fund to be screened and the second holding product corresponding to the target fund, the overlap degree of each concept exposure list between each fund to be screened and the target fund is determined, and a preset number of funds with the largest overlap degree of concept exposure list among the funds to be screened are taken as the target candidate funds.

6. A fund similarity calculation platform, characterized in that: The fund similarity calculation platform includes: A first acquisition module is used to acquire classification information and latest holding information corresponding to the target fund, determine the maximum weighted industry corresponding to the target fund based on the latest holding information, and determine a preset number of target candidate funds based on the classification information and the maximum weighted industry; A first determination module is used to determine a first correlation between first holdings corresponding to the target fund, and a second correlation between second holdings corresponding to each target candidate fund, wherein the first correlation is a correlation representing the annual returns of the holdings of the target fund, and the second correlation is a correlation representing the annual returns of the holdings of the target candidate funds; A second acquisition module, used to acquire a first minimum spanning tree corresponding to the first correlation and a second minimum spanning tree corresponding to the second correlation; The second determination module is used to compare the first minimum spanning tree with each of the second minimum spanning trees to determine the similarity between the first minimum spanning tree and each of the second minimum spanning trees, determine the target minimum spanning tree corresponding to the maximum similarity among each of the similarities in each of the second minimum spanning trees, and use the target candidate fund corresponding to the target minimum spanning tree as a similar fund to the target fund; wherein, Determining the maximum weighted industry corresponding to the target fund based on the latest holding information includes: acquiring all first holding products of the target fund according to the latest holding information, the first holding products including stocks, bonds, and funds, and determining the industry corresponding to each first holding product and the holding ratio of the first holding product in the target fund; then, according to the latest holding information of the target fund, mapping the industry of the first holding products held by the target fund to the industry in which the target fund is located, that is, determining all industries corresponding to the target fund according to each first holding product, and determining the industry holding ratio corresponding to each industry according to the holding ratio of the first holding product, for each industry, the industry holding ratio is the sum of the holding ratios of the holding products belonging to the industry; then, determining the maximum industry holding ratio among the industry holding ratios, and taking the industry corresponding to the maximum industry holding ratio as the maximum weighted industry; Determining the first correlation between the first holdings corresponding to the target fund includes: obtaining the first annual return vector of the first holding product corresponding to the target fund, the first annual return vector is an annual return time series, the annual return time series is a one-year yield time series, that is, a sequence of the yields of each trading day in a year sorted in chronological order, and the vector corresponding to the annual return time series is the first yield time series vector corresponding to the first holding product; then, determining the first correlation between each first holding product based on each first annual return vector: calculating the first angle cosine value between each first annual return vector, normalizing each first angle cosine value to map each first angle cosine value to (0, 1), obtaining the first mapping value of each first angle cosine value, and using each first mapping value as the first correlation between the corresponding first holding products; Determining the second correlation between the second holdings corresponding to each target alternative fund includes: for each target alternative fund, obtaining the second holdings of each target alternative fund, and obtaining the second annual return vector corresponding to each second holding, and then determining the second correlation between each second holding based on each second annual return vector: calculating the second angle cosine value between each second annual return vector, normalizing each second angle cosine value to map each second angle cosine value to (0, 1), obtaining the second mapping value of each second angle cosine value, using each second mapping value as the second correlation between the corresponding second holdings, and then obtaining the second correlation between the second holdings of the target alternative funds.

7. A fund similarity calculation device, characterized in that: The fund similarity calculation device comprises: a memory, a processor and a fund similarity calculation program stored in the memory and executable on the processor. When the fund similarity calculation program is executed by the processor, the steps of the fund similarity calculation method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a fund similarity calculation program, and when the fund similarity calculation program is executed by a processor, the steps of the fund similarity calculation method according to any one of claims 1 to 5 are implemented.

Images