KR20210082879A - Method and system for determining investment weight using similarity of time-series data - Google Patents

Abstract

The present invention relates to a method and a system for investment proportion determination using the similarity of time-series data. The method includes: (a) a step in which investment strategies are selected: (b) a step of collecting investment performance data regarding each of the investment strategies; (c) a step of measuring investment strategy pair-specific time-series similarity using the investment performance data regarding each of the investment strategies; (d) a step of combining a pair of investment strategies closest in terms of the measured time-series similarity in an investment strategy cluster; (e) a step of calculating the investment performance data of the investment strategy cluster; and (f) a step of generating an investment strategy cluster hierarchical structure of sequential combination of the investment strategies by repeating (c) to (e) with respect to the investment strategy cluster and the investment strategies not combined in the investment strategy cluster. The method may further include a step of determining the investment proportion of each investment strategy based on the generated investment strategy cluster hierarchical structure.

Description

BACKGROUND ART Method and system for determining investment weight using similarity of time-series data}

The present invention relates to a method and system for determining the proportion of investment using the similarity of time series data, and more particularly, to determine a hierarchical structure in the form of an investment strategy cluster tree based on the similarity of investment performance by investment strategy, and an investment strategy according to the determined hierarchy It relates to a method and system for determining the percentage of investment by star.

Investment is a set of the following five decisions (① what, ② when ③ how much to buy, and ④ when ⑤ how much to sell). For example, buying and holding a specific stock is an investment strategy and is an investment target.

Investment efficiency is greatly affected by the logic of determining the proportion of investment between investment strategies, that is, the logic of distributing funds.

Among the fund allocation logic between investment strategies, there are the following types of allocation logic. After obtaining time-series data of investment returns for each investment strategy, measuring the linearity between each of the time-series data of investment returns, based on the linearity between the measured returns, the proportion of investment between investment strategies is determined according to a predetermined algorithm.

However, one of the limitations of this conventional method is that the proportion of investment between investment strategies is determined based only on “linearity between investment returns” without considering the characteristics of time series data. Therefore, there was a limit to its application to investment activities carried out according to the passage of time.

Japanese Laid-Open Patent Publication No. 2008-250975 (published date: October 16, 2008)

The technical task of the present invention is to provide a method and system for determining a hierarchical structure in the form of an investment strategy cluster tree based on the similarity of investment performance for each investment strategy, and determining the investment weight for each investment strategy according to the determined hierarchical structure.

The investment weight determination method using the similarity of time series data according to an embodiment of the present invention for solving the above technical problem includes the steps of (a) receiving a plurality of investment strategies, (b) each of the plurality of investment strategies Collecting investment performance data, (c) measuring the time-series similarity for each investment strategy pair using the investment performance data of each of the plurality of investment strategies, (d) the measured time-series similarity of the closest pair bundling the investment strategy into an investment strategy cluster, (e) calculating the investment performance data of the investment strategy cluster, and (f) adding the remaining investment strategies and investment strategy clusters among the plurality of investment strategies that are not grouped into the investment strategy cluster and repeating steps (c) to (e) for generating an investment strategy cluster hierarchical structure in which the plurality of investment strategies are sequentially bundled.

The method may further include (g) determining an investment weight of each of the plurality of investment strategies based on the generated investment strategy cluster hierarchical structure.

The investment performance data of each of the plurality of investment strategies may be yield time series data of each of the plurality of investment strategies in a predetermined period.

The return time series data may be a return based on the daily closing price of the predetermined period or a return based on the closing price at the end of each week, month, quarter, or year.

In step (c), the similarity of each investment strategy pair can be measured by at least one method of Dynamic Time Wrapping (DTW), Correlation Based Dynamic Time Wrapping (CBDTW), and Global Alignment Kernels (GAK).

The proportion of investment between a pair of investment strategies grouped in an investment strategy cluster may be set equally, or may be determined in a way that equalizes the amount of risk between the pair of investment strategies.

The investment performance data of the investment strategy cluster may be calculated in consideration of the investment weight of the investment strategy bundled into the investment strategy cluster.

After adding the investment performance invested in the plurality of investment strategies for a predetermined period according to the investment weight determined based on the generated hierarchical structure to the investment performance data of each of the plurality of investment strategies, the above (c) to (g) ) can be repeated.

It may include a computer-readable recording medium in which a program for executing the method in a computer according to an embodiment of the present invention for solving the above technical problem is recorded.

The investment weight determination system using the similarity of time series data according to an embodiment of the present invention for solving the above technical problem is a data collection unit for collecting investment performance data of each of a plurality of investment strategies, each of the plurality of investment strategies A similarity measuring unit that measures the time series similarity of each investment strategy pair using investment performance data, a hierarchical clustering unit that groups the pair of investment strategies having the closest measured time series similarity into an investment strategy cluster, and the investment strategy cluster and a data calculation unit for calculating investment performance data.

The hierarchical clustering unit is time-series again for the remaining investment strategies and investment strategy clusters that are not grouped into the investment strategy cluster among the plurality of investment strategies. By repeating the process of bundling a pair with the closest similarity into an investment strategy cluster, an investment strategy cluster hierarchical structure is created in which the plurality of investment strategies are sequentially bundled.

The system may further include an investment proportion determining unit that determines an investment proportion of each of the plurality of investment strategies based on the generated investment strategy cluster hierarchical structure.

According to the present invention, it is possible to more accurately grasp the similarity between investment strategies and the hierarchical structure, and to reclassify into a desired number of clusters. In addition, it is possible to improve investment efficiency by accurately classifying various investment targets in the order of similarity, and has the advantage of improving the return on investment in asset management and enabling precise investment risk management and control.

1 is a configuration diagram of a system for determining an investment weight using a degree of similarity of time series data according to an embodiment of the present invention.
FIG. 2 shows a dendrogram, which is a tree-shaped structure, indicating an order in which a plurality of individual entities are sequentially and hierarchically combined through hierarchical clustering.
3 is a cluster generated according to an embodiment of the present invention - illustrates an investment strategy hierarchical structure in the form of a tree.
4 is a flowchart illustrating a method of determining an investment weight using a similarity of time series data according to an embodiment of the present invention.

Hereinafter, with reference to the accompanying drawings, embodiments of the present invention will be described in detail so that those of ordinary skill in the art to which the present invention pertains can easily implement them. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part "includes" a certain component, it means that other components may be further included, rather than excluding other components, unless otherwise stated.

In addition, the terms “…unit” and “…module” described in the specification mean a unit that processes at least one function or operation, which may be implemented as hardware or software or a combination of hardware and software.

1 is a configuration diagram of a system for determining an investment weight using a degree of similarity of time series data according to an embodiment of the present invention.

Referring to FIG. 1 , the investment weight determination system 100 according to the present invention includes an investment strategy input unit 110 , a data collection unit 120 , a similarity measurement unit 130 , a hierarchical clustering unit 140 , and a data calculation unit. It may include a 150 and an investment weight determining unit 160 .

The investment strategy input unit 110 may receive a plurality of investment strategies selected from the user. Here, the investment strategy may be the simplest investment target stock. In other words, buying and holding the relevant investment target stock itself can be an investment strategy. Here, the investment target stock may be a stock traded on a stock exchange, a fund, etc., or various investment product stocks such as a futures trading stock and an option trading stock.

On the other hand, purchasing and holding a plurality of investment target stocks at a certain weight may also be an investment strategy. For example, buying and holding stock A by 60% and stock B by 40% may be an investment strategy.

Also, when to buy and when to sell an investment target may be an investment strategy. For example, one investment strategy may be to buy and hold stock A, then sell stock A and buy stock B when certain conditions are met.

The data collection unit 120 may collect investment performance data of each of a plurality of investment strategies. Here, the investment performance data for each investment strategy may be time series data consisting of a rate of return obtained according to the corresponding investment strategy for a predetermined period. For example, it may be time series data consisting of a return based on the daily closing price of the relevant stock for a certain period, or time series data consisting of a return based on the monthly closing price of the relevant stock. Of course, in the case of two or more stocks, it may be time series data consisting of daily or monthly closing price-based returns reflecting the investment weight. Of course, it may be time series data consisting of returns determined by criteria other than daily and monthly closing prices.

The data collection unit 120 may collect from a database server (not shown) that is constructing a database for the rate of return of each investment target item.

The similarity measuring unit 130 may measure the similarity of each investment strategy pair by using the investment performance data of each investment strategy. As described above, the investment performance data of each investment strategy may be time series data consisting of returns. In this case, by obtaining the similarity between two different time series data pairs, the time series similarity for each investment strategy pair can be measured.

The similarity measuring unit 130 measures the time-series similarity to a pair of investment performance data, which is time-series data consisting of returns, using methods such as DTW (Dynamic Time Wrapping), CBDTW (Correlation Based Dynamic Time Wrapping) and GAK (Global Alignment Kernels) can be measured

Since DTW, CBDTW, and GAK methods are widely known as methods for obtaining time-series similarity between different time-series data, a detailed description will be omitted here. In particular, the DTW method can obtain the similarity between two time series data of different lengths.

The hierarchical clustering unit 140 may generate an investment strategy cluster hierarchical structure in which the plurality of investment strategies are sequentially bundled by performing hierarchical clustering on the plurality of investment strategies.

FIG. 2 illustrates a dendrogram, which is a tree-shaped structure, indicating an order in which a plurality of individual entities are sequentially and hierarchically combined through hierarchical clustering.

The hierarchical clustering unit 140 according to the present invention may group a plurality of investment strategies into an investment strategy cluster hierarchical structure having a tree-like structure as illustrated in FIG. 2 .

3 is a cluster generated according to an embodiment of the present invention - illustrates an investment strategy hierarchical structure in the form of a tree.

The hierarchical clustering unit 140 repeats the process of tying a pair with the closest time series similarity to the investment strategy cluster for the remaining investment strategies and investment strategy clusters that are not grouped into the investment strategy cluster among the plurality of investment strategies. An investment strategy cluster hierarchical structure can be created by sequentially grouping investment strategies.

For example, as a result of measuring the time-series similarity of each investment strategy pair for investment strategies A, B, C, D and E, if the similarity between A and B is the highest, as illustrated in FIG. Create a cluster AB of And by considering cluster AB as one investment strategy, the similarity of each investment strategy pair can be measured for cluster AB and investment strategies C, D, and E again. And if cluster AB and investment strategy C have the highest similarity, cluster AB and investment strategy C are combined to form cluster ABC. The hierarchical clustering unit 140 may generate an investment strategy cluster hierarchical structure while repeating the above method.

The data calculation unit 150 may calculate investment performance data of the corresponding cluster whenever an investment strategy is grouped into a cluster in the hierarchical clustering unit 140 . The data calculation unit 150 may calculate the investment performance data of the corresponding cluster in consideration of the investment weight of the investment strategy grouped into the investment strategy cluster.

The investment weight determining unit 160 may provide the data calculation unit 150 with a result of determining the investment weight of the investment strategy bundled into the corresponding cluster whenever the investment strategy is grouped into a cluster.

When calculating cluster investment performance data, the ratio of investment between investment strategies can be applied equally or differently. If applied equally, investment performance data can be calculated as time series data consisting of the average value of the returns of both investment strategies.

Of course, it is possible to determine the investment weight by a method that equalizes the amount of risk between a pair of investment strategies grouped in an investment strategy cluster, for example, a risk parity method, and calculates investment performance data accordingly. For example, if the proportion of investment strategy A determined by the risk parity method is 20% and the proportion of investment strategy B is 80%, a weight of 0.2 is given to the rate of return of investment strategy A and a weight of 0.8 is given to the rate of return of investment strategy B. It is possible to calculate the return data of the cluster A and B are tied together.

The risk parity method, which configures the degree of contribution of each asset's return to the overall risk of the portfolio to be the same, is already well known, so a detailed description will be omitted.

The investment proportion determining unit 160 may determine the investment proportion of each of the plurality of investment strategies based on the investment strategy cluster hierarchical structure generated by the hierarchical clustering unit 140 .

Referring to FIG. 3 , for example, 50% is distributed to cluster ABC and 50% is distributed to cluster DC. Then, it goes down one tier and distributes 50% to cluster ABC, 25% to cluster AB, and 25% to investment strategy C. And 50% is distributed to cluster DC, and 25% is distributed to each of investment strategies D and E. Finally, as 25% is distributed to cluster AB, 12.5% can be distributed to clusters A and B. Of course, it is also possible to determine the investment weight for each investment strategy while equally matching the risk for each cluster using the risk parity method.

4 is a flowchart illustrating a method of determining an investment weight using a similarity of time series data according to an embodiment of the present invention.

1 and 4 , first, the investment strategy input unit 110 may receive a plurality of investment strategies selected from the user ( S410 ).

Then, the data collection unit 120 may collect investment performance data of each of the plurality of investment strategies input in step S410 (S420). Here, it may be time series data consisting of a rate of return obtained according to a corresponding investment strategy for a predetermined period. For example, it may be time-series data such as a return based on the daily closing price and a return based on the closing price on the last day of each month obtained according to the corresponding investment strategy.

Next, the similarity measuring unit 130 may measure the time-series similarity of each investment strategy pair by using the investment performance data of each of the plurality of investment strategies collected in step S420 (S430). The time series similarity of each investment strategy pair can be measured by methods such as DTW (Dynamic Time Wrapping), CBDTW (Correlation Based Dynamic Time Wrapping), and GAK (Global Alignment Kernels) for the return time series data of each investment strategy.

Thereafter, the hierarchical clustering unit 140 may group a pair of investment strategies having the closest time-series similarity measured in step S420 into an investment strategy cluster (S440).

And the data calculation unit 150 may calculate the investment performance data of the investment strategy cluster bundled in step S440 (S450). In step S450, the investment performance data may calculate the investment performance data of the corresponding cluster in consideration of the investment weight of the investment strategy bundled into one investment strategy cluster.

Next, the hierarchical clustering unit 140 repeats steps (S430) to (S450) for the remaining investment strategies and investment strategy clusters that are not bundled into the investment strategy cluster among the plurality of investment strategies, It is possible to create an investment strategy cluster hierarchical structure in which the investment strategies are sequentially bundled (S460).

Thereafter, the investment proportion determining unit 160 may determine the investment proportion of each of the plurality of investment strategies based on the investment strategy cluster hierarchical structure generated in step S460 (S470).

Next, it may be invested in a plurality of investment strategies for a predetermined period according to the investment weight determined in step (S470) (S480).

In addition, the data collection unit 120 may collect the investment performance in step S480 and add it to the investment performance data of each of a plurality of investment strategies (S490). And the investment weight determination system 100 may repeat steps (S430) to (S470).

The embodiments described above may be implemented by a hardware component, a software component, and/or a combination of a hardware component and a software component. For example, the apparatus, methods and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate (FPGA) array), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For convenience of understanding, although one processing device is sometimes described as being used, one of ordinary skill in the art will recognize that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that can include For example, the processing device may include a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as parallel processors.

The software may comprise a computer program, code, instructions, or a combination of one or more thereof, which configures a processing device to operate as desired or is independently or collectively processed You can command the device. The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or device, to be interpreted by or to provide instructions or data to the processing device. , or may be permanently or temporarily embody in a transmitted signal wave. The software may be distributed over networked computer systems, and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic such as floppy disks. - includes magneto-optical media, and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with reference to the limited drawings, those skilled in the art may apply various technical modifications and variations based on the above. For example, the described techniques are performed in a different order than the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result.

100: investment weight determination system
110: investment strategy input unit
120: data collection unit
130: similarity measurement unit
140: hierarchical clustering unit
150: data calculator
160: investment proportion determining unit

Claims (17)

(a) step of selecting a plurality of investment strategies;
(b) collecting investment performance data for each of the plurality of investment strategies;
(c) measuring the time-series similarity of each investment strategy pair using the investment performance data of each of the plurality of investment strategies;
(d) bundling a pair of investment strategies with the closest measured time-series similarity into an investment strategy cluster;
(e) calculating the investment performance data of the investment strategy cluster, and
(f) an investment strategy cluster in which the plurality of investment strategies are sequentially bundled by repeating steps (c) to (e) for the remaining investment strategies and investment strategy clusters among the plurality of investment strategies Steps to create a hierarchy
A method of determining the proportion of investment, including In claim 1,
(g) determining the investment weight of each of the plurality of investment strategies based on the generated investment strategy cluster hierarchy
A method of determining the proportion of the investment further comprising a. In claim 2,
The investment performance data of each of the plurality of investment strategies,
An investment weighting method that is return time series data of each of the plurality of investment strategies in a predetermined period. In claim 3,
The yield time series data is,
An investment weighting method comprising a return based on the daily closing price of the predetermined period or a return based on the closing price at the end of each week, month, quarter or year. In claim 3,
In step (c),
An investment weight determination method for measuring the time-series similarity of each investment strategy pair by at least one of DTW (Dynamic Time Wrapping), CBDTW (Correlation Based Dynamic Time Wrapping), and GAK (Global Alignment Kernels). In claim 3,
A method of determining the proportion of investment in which the proportion of investment between a pair of investment strategies grouped into an investment strategy cluster is set equally, or in a way that equalizes the amount of risk between a pair of investment strategies. In claim 6,
The investment performance data of the investment strategy cluster,
A method of determining the investment proportion calculated by considering the investment proportion of the investment strategy grouped into the investment strategy cluster. In claim 2,
After adding the investment performance invested in the plurality of investment strategies for a predetermined period according to the investment weight determined based on the generated hierarchical structure to the investment performance data of each of the plurality of investment strategies, the above (c) to (g) ) method to determine the investment weighting by repeating steps. A computer-readable recording medium in which a program for executing any one of the methods of any one of claims 1 to 8 is recorded on a computer. A data collection unit that collects investment performance data for each of a plurality of investment strategies;
A similarity measuring unit for measuring the time-series similarity of each investment strategy pair using the investment performance data of each of the plurality of investment strategies;
A hierarchical clustering unit tying a pair of investment strategies having the closest measured time-series similarity into an investment strategy cluster, and
A data calculation unit for calculating investment performance data of the investment strategy cluster
including,
The hierarchical clustering unit,
For the remaining investment strategies and investment strategy clusters that are not grouped into the investment strategy cluster among the plurality of investment strategies, the process of tying the pair with the closest time-series similarity to the investment strategy cluster is repeated, and the plurality of investment strategies are sequentially bundled. An investment weighting system that creates an investment strategy cluster hierarchy. 11. In claim 10,
An investment proportion determining unit that determines an investment proportion of each of the plurality of investment strategies based on the generated investment strategy cluster hierarchical structure
Investment weight determination system further comprising a. In claim 11,
The investment performance data of each of the plurality of investment strategies,
An investment weighting system that is return time series data of each of the plurality of investment strategies in a predetermined period. In claim 12,
The yield time series data is,
An investment weighting system comprising a return based on the daily closing price of the predetermined period or a return based on the closing price at the end of each week, month, quarter or year. In claim 12,
The similarity measurement unit,
The investment weight determination system for measuring the time-series similarity of each investment strategy pair by at least one method of DTW (Dynamic Time Wrapping), CBDTW (Correlation Based Dynamic Time Wrapping), and GAK (Global Alignment Kernels). In claim 12,
An investment weighting system in which the percentage of investment between a pair of investment strategies grouped in an investment strategy cluster is set equally or in a way that equalizes the amount of risk between a pair of investment strategies. 16. In claim 15,
The investment performance data of the investment strategy cluster,
An investment proportion determination system calculated by considering the investment proportion of the investment strategies grouped into the investment strategy cluster. In claim 11,
After adding the investment performance invested in the plurality of investment strategies for a predetermined period to the investment performance data of each of the plurality of investment strategies according to the investment weight determined based on the generated investment strategy cluster hierarchical structure, the plurality of investment strategies An investment weighting system that regenerates a sequentially grouped investment strategy cluster hierarchy.

Images