JP7734231B1 - Operational model management device, operational model management method, and operational model management program - Google Patents

Abstract

[Problem] To provide a technology for managing investment funds using a portfolio that combines multiple investment strategies. [Solution] One aspect of the present disclosure relates to an investment model management device that includes a child investment model management unit that manages multiple child investment models, each of which is managed according to a predetermined investment strategy, a parent investment model management unit that manages a parent investment model that manages investment funds, and a portfolio determination unit that determines a portfolio for allocating the investment funds of the parent investment model to the multiple child investment models based on investment performance information of each child investment model. [Selected Figure] Figure 1

Description

This disclosure relates to an operational model management device, an operational model management method, and an operational model management program.

With the recent advances in information and communications technology, the financial industry is also actively conducting research and development into financial technologies that utilize information and communications technology, often referred to as "Fintech."

For example, in the field of asset management, investment funds such as investment trusts and private placement funds are now being managed using investment support information provided based on computer-programmed investment strategies. Each investment strategy is formulated in accordance with the investment policy of each investment fund. For example, such investment policies are determined depending on whether the investment targets are domestic securities or foreign securities, stocks or bonds, and whether the target investors are retail investors or institutional investors.

Japanese Patent Application Laid-Open No. 2022-017571

A variety of investment strategies can be formulated to correspond to each investment policy, and the investment models realized through these various investment strategies will also be diverse. The investment performance of the investment models realized through individual investment strategies may vary depending on factors such as economic conditions and market environments. If economic conditions or market environments become unexpected, investment behavior relying on a specific investment strategy may result in unexpected investment performance.

One objective of the present disclosure is to provide technology for managing investment funds through a portfolio that combines multiple investment strategies.

One aspect of the present disclosure relates to an operation model management device having a child operation model management unit that manages multiple child operation models, each of which is managed according to a predetermined investment strategy; a parent operation model management unit that manages a parent operation model that manages investment funds; and a portfolio determination unit that determines a portfolio for allocating the investment funds of the parent operation model to the multiple child operation models based on investment performance information of each child operation model.

This disclosure provides technology for managing investment funds through a portfolio that combines multiple investment strategies.

FIG. 1 is a schematic diagram illustrating an operation scheme according to an embodiment of the present disclosure. FIG. 1 is a conceptual diagram illustrating an operation model management device according to an embodiment of the present disclosure. FIG. 2 is a block diagram showing a hardware configuration of an operation model management device according to an embodiment of the present disclosure. 1 is a block diagram showing a functional configuration of an operation model management device according to an embodiment of the present disclosure. 10 is a flowchart illustrating an operation model management process according to an embodiment of the present disclosure.

Embodiments of the present disclosure will be described below with reference to the drawings.

In the following embodiment, an operational model management device that provides investment support information is disclosed.

[Operation scheme]
In an investment scheme according to one embodiment of the present disclosure, when managing investment funds such as investment trusts and private placement funds, as shown in Figure 1, a parent investment model that manages the entire investment funds allocates the investment funds to multiple child investment models #1, #2, ..., #m, and manages the investment funds by receiving investment returns from each child investment model.

Each child investment model #i (i = 1, 2, ..., m) manages the allocated investment funds according to an individual investment strategy. For example, child investment model #1 manages investment funds according to a stock investment strategy with a crisis avoidance trigger, child investment model #2 manages investment funds according to an economic environment judgment bond investment strategy, and child investment model #3 manages investment funds according to a yield curve spread trend following strategy. Note that these investment strategies are merely examples, and any other investment strategy may be used in child investment model #i.

An operation model management device 100 according to one embodiment of the present disclosure manages a parent operation model and, as shown in FIG. 2, acquires investment performance information (e.g., daily return data) for each child operation model from a child operation model database (DB) 20 that stores child operation model information for each child operation model. Based on the acquired investment performance information, the operation model management device 100 determines how to optimally allocate the investment funds of the parent operation model to the child operation model, i.e., determines the optimal portfolio for the child operation model. The operation model management device 100 then provides the determined portfolio as investment support information to a user terminal 30, such as the manager of the parent operation model.

In this way, an investment manager managing investment funds using a parent investment model can obtain a portfolio suitable for allocating investment funds to each child investment model as investment support information from the investment model management device 100, enabling them to manage investment funds with optimal fund allocation.

Here, the operational model management device 100 is realized by a computing device such as a server or personal computer, and may have a hardware configuration such as that shown in FIG. 3. That is, the operational model management device 100 has a storage device 101, a processor 102, an interface device 103, and a communication device 104, which are interconnected via a bus B.

The programs or instructions that implement the various functions and processes described below in the operation model management device 100 may be downloaded from an external device via a network, or may be provided from a removable storage medium such as a CD-ROM (Compact Disk-Read Only Memory) or flash memory.

Storage device 101 may be implemented using random access memory, flash memory, a hard disk drive, or the like, and stores installed programs or instructions as well as files, data, and the like used to execute the programs or instructions. Storage device 101 may also include non-transitory storage media.

The processor 102 may be realized by one or more CPUs (Central Processing Units), GPUs (Graphics Processing Units), processing circuitry, etc., which may be composed of one or more processor cores, and executes various functions and processes of the operation model management device 100 (described below) in accordance with programs, instructions, and data such as parameters required to execute the programs or instructions stored in the storage device 101.

The interface device 103 provides an interface between the database 20 and user terminal 30 and the operational model management device 100. For example, a user operates a keyboard, mouse, etc. on a GUI (Graphical User Interface) displayed on the display or touch panel of the user terminal 30 to send and receive various information, data, instructions, etc. between the operational model management device 100 and the interface device 103.

The communication device 104 is realized by various communication circuits that perform communication processing with communication networks such as external devices, the Internet, and LANs (Local Area Networks).

However, the above-described hardware configuration is merely an example, and the operation model management device 100 according to the present disclosure may be realized using any other appropriate hardware configuration.

[Operational Model Management Device]
Next, an operation model management device 100 according to an embodiment of the present disclosure will be described. Fig. 4 is a block diagram showing the functional configuration of the operation model management device 100 according to an embodiment of the present disclosure.

As shown in FIG. 4, the operation model management device 100 has a child operation model management unit 110, a parent operation model management unit 120, and a portfolio determination unit 130.

The child investment model management unit 110 manages multiple child investment models, each of which is managed according to a predetermined investment strategy. Each child investment model manages investment funds according to its own individual investment strategy. The investment strategy of each child investment model is set, for example, according to the investment policy of the child investment model, and may be defined using a mathematical formula, computer program, or the like. For example, the investment funds of a parent investment model may be allocated to m child investment models #1, #2, ..., #m.

The child investment model management unit 110 grasps the investment status of each child investment model based on, for example, the investment performance information of each child investment model extracted from the child investment model DB20. The investment performance information of the child investment model may be, for example, daily return data. The child investment model management unit 110 may calculate statistics such as the expected return μ and the variance-covariance matrix Σ based on the daily return data r of each child investment model, or it may be any data for deriving the expected return μ, the variance-covariance matrix Σ, etc.

The parent investment model management unit 120 manages the parent investment model that manages investment funds. Here, the parent investment model is managed by an investment manager such as a fund manager, and specifically, the investment funds of the parent investment model are allocated to each child investment model according to the weight w = ( _w1 , ..., _wm ) for each child investment model #1, ..., #m instructed by the investment manager, where 0 ≤ wi _≤ 1 and _w1 + ... + _wm = 1.

The parent investment model management unit 120 appropriately checks the investment status of each child investment model to which investment funds have been allocated, and understands the investment status of the parent investment model based on the investment status of each child investment model. For example, the parent investment model management unit 120 calculates the expected return of the parent investment model's portfolio based on the expected return of each child investment model.

The portfolio determination unit 130 determines a portfolio for allocating the investment funds of the parent investment model to multiple child investment models based on the investment performance information of each child investment model. That is, the portfolio determination unit 130 determines the optimal weights w = ( _w1 , ..., _wm ) for allocating the investment funds of the parent investment model to each child investment model #1, ..., #m as the portfolio.

For example, the portfolio determination unit 130
The portfolio w may be determined according to the following equation: where w is a column vector representing the weights of the child investment models #1, ..., #m, μ is a column vector representing the expected returns of the child investment models #1, ..., #m, _λ1 represents the degree of risk aversion with respect to variance, Σ(∈R ^m,m ) represents the variance-covariance matrix of the expected returns of the child investment models #1, ..., #m, and T represents the transpose of the matrix. In other words, (Equation 1) is an objective function that aims to balance the expected return of the portfolio and the predicted risk by multiplying them by a coefficient. Attempting to achieve a large return tends to increase the risk accordingly, so optimization of (Equation 1) aims for as large a return as possible while adjusting to avoid taking too large a risk. In other words, the portfolio determination unit 130 determines the portfolio according to mean-variance optimization.

Furthermore, the portfolio determination unit 130
The portfolio w may be determined according to the following equation: where t represents a parameter of the moment generating function, λ ₂ represents the degree of risk aversion with respect to EVaR (Entropic Value at Risk), p _j represents the occurrence probability of r ^j , r ^j represents the jth row of r (where r∈R ^N,m ), and α represents the risk level of EVaR. In other words, the portfolio determination unit 130 determines a portfolio based on the portfolio's volatility and the degree of risk aversion with respect to downside risk. Note that when λ ₂ =0, (Equation 1) and (Equation 2) are the same. The risk in (Equation 1) only considers diversification risk, but it is known that relatively large declines occur in the market more frequently than expected in a normal distribution. Therefore, by incorporating a risk index that also considers losses in the tail of the return distribution (losses not expected in a normal distribution) into the objective function, it is possible to aim to build a portfolio that has a certain degree of resistance to occasional large losses. The third term in (Equation 2) is a rewrite of EVaR in (Equation 3) into a form that can be calculated from data. EVaR is a risk indicator that focuses on the tail of the distribution and functions to reduce the occasional large losses mentioned above.

Here, EVaR is defined as follows:
Here, X represents a random variable representing portfolio losses, and M _X represents the moment generating function of the random variable, M _X = E[e ^tX ]. Furthermore, the third term in (Equation 2) corresponds to EVaR in (Equation 3), and if z in (Equation 3) is replaced with t in (Equation 2) and the argument of the logarithm in (Equation 3) is separated, it becomes close to (Equation 2). The return data r for each strategy is input data (given values) to the optimization system, and once the portfolio weights are determined, the value of (Equation 2) can be specifically calculated. Therefore, various values of w are tried to find the optimal value of w. z is the same as t in (Equation 2) and represents the parameter of the moment generating function, α is the confidence level, and EVaR evaluates losses that occur with a probability of α% or less.

EVaR is known as a coherent risk measure that satisfies the following properties:
- Translation invariance
・Subadditivity
・Monotonicity
・Positive homogeneity

For example, the portfolio determination unit 130 considers allocating 10 billion yen of investment funds from a parent investment model to three child investment models #1 to #3. Here, child investment model #1 may manage the investment funds according to a stock investment strategy with a crisis avoidance trigger. Specifically, the stock investment strategy with a crisis avoidance trigger uses two types of crisis avoidance triggers defined by the value of the VIX index. Each trigger takes two values, risk-on and risk-off, and the strategy determines E-mini S&P 500 futures positions based on their combination.

The first trigger is risk-on if both of the following two conditions are met, otherwise it is risk-off:
Condition 1: The closing price of the VIX index is less than 30. Condition 2: The closing price of the VIX index is less than the upper limit of the VIX Bollinger band (defined as 14 days and 0.7 standard deviations).

The second trigger is defined by the following two conditions:
-Condition 1: The closing price of the VIX index is 24 or higher -Condition 2: The VIX trend (120-day difference between the closing prices of the VIX index) is positive If both conditions 1 and 2 are met, the trigger will be risk-off. If there was risk-off on the previous day, and either condition 1 or 2 is not met, the trigger will be risk-on. The E-mini S&P 500 futures position is defined as 100% long when both types of triggers are risk-on, 50% long when either is risk-off, and neutral when both are risk-off.

Furthermore, the child investment model #m may manage investment funds according to an economic environment-sensitive bond investment strategy. Specifically, the economic environment-sensitive bond investment strategy is a strategy that determines positions based on a composite indicator that adds the values of employment, price, and economic indicators created from three economic indicators (U.S. Non-Farm Payrolls, U.S. PCE Price Index, and U.S. ISM Manufacturing Index). Initially, the employment indicator takes a value between 0 and 3. Let X be the average value of the NFP TCH Index (ACTUAL RELEASE) for the three months including the most recent month available on date T, and then, if "X<100", it is 0; if "100≦X<200", it is 1; if "200≦X<300", it is 2; and if "300<X", it is 3. Next, the price indicator takes a value between 0 and 3. Let X be the average of the PCE CMOM Index (ACTUAL RELEASE) for the three months including the most recent month available on date T, and let it be 0 if X < 0.2, 1 if 0.2 < X < 0.3, 2 if 0.3 < X < 0.4, and 3 if 0.4 < X. The economic indicator takes a value between 0 and 3. Let X be the average of the napmpmi Index (ACTUAL RELEASE) for the three months including the most recent month available on date T, and let it be 0 if X < 50, 0 if 50 < X < 55, 1 if 55 < X < 60, 2 if 60 < X. The composite indicator takes a value between 0 and 9, calculated by adding the three indicators together. If the value is greater than 4, the 10-year U.S. Treasury futures are short, and if it is less than 4, the 10-year U.S. Treasury futures are long.

Furthermore, child investment model #n may manage investment funds according to a yield curve spread trend-following strategy. Specifically, the yield curve spread trend-following strategy creates two types of moving averages, one long and one short, from the spread between the 10-year U.S. Treasury bond yield and the 2-year U.S. Treasury bond yield, and determines positions in 10-year U.S. Treasury bond futures and 2-year U.S. Treasury bond futures based on their golden cross or death cross. First, the difference between the 10-year U.S. Treasury bond yield and the 2-year U.S. Treasury bond yield is calculated daily to obtain a daily series of the 2-year/10-year spread, after which a 50-day moving average and a 100-day moving average are created. If the 50-day moving average crosses the 100-day moving average from bottom to top, go 100% short on 10-year Treasury futures and 335% long on 2-year Treasury futures. Conversely, if the short-term moving average crosses the long-term moving average from top to bottom, go 100% long on 10-year Treasury futures and 335% short on 2-year Treasury futures.

The child investment model management unit 110 first obtains daily return data r for each child investment model #1 to #3 for a certain period. Then, the child investment model management unit 110 calculates the statistics required for portfolio optimization calculations. For example, the expected return μ for each child investment model #1 to #3 may use the average return of each child investment model #1 to #3 in the past. However, the expected return μ according to the present disclosure is not limited to this, and other expected return values may also be used.

In addition, the child investment model management unit 110 can calculate the variance-covariance matrix Σ of each child investment model #1 to #3 from the daily return data r of each child investment model #1 to #3. However, the variance-covariance matrix Σ according to the present disclosure is not limited to this, and other variance-covariance matrices Σ may also be used.

The parent investment model management unit 120 then sets the risk aversion λ ₁ and/or λ _2. The risk aversion may be specified by the investment manager of the parent investment model, and may be a value that enables the portfolio to achieve the required return and risk. For example, using a known backtesting method, the investment manager may select an appropriate risk aversion based on values such as performance, return, standard deviation of return (volatility), and drawdown in past actual investments.

After the operations manager determines the parameters required for the optimization calculations of (Equation 1) and (Equation 2), such as μ, Σ, _λ1 , and _λ2 , the optimization calculations are performed. Since both the objective function and constraints of (Equation 1) and (Equation 2) are convex functions, these problems are convex programming problems. The interior point method is used as the optimization calculation algorithm.

For example, if an investment manager selects risk aversion levels λ ₁ = 70 and λ ₂ = 0.01, and performs optimization calculations using market data from August 2005 to August 2023, the portfolio determination unit 130 calculates the optimal portfolio as (w ₁ , w ₂ , w ₃ ) = (0.178, 0.340, 0.482). In this case, the parent investment model management unit 120 will allocate 1.78 billion yen to child investment model #1, 3.40 billion yen to child investment model #2, and 4.82 billion yen to child investment model #3 out of an investment fund of 10 billion yen.

The above-described investment model management device 100 can include EVaR, a convex risk indicator, in the objective function, allowing for a portfolio optimization method that takes both volatility and downside risk into account as a convex optimization problem. The third term in (Equation 2) penalizes large portfolio losses. This penalty is intended to reduce the risk of sudden large portfolio losses and to penalize particularly large losses, rather than equally penalizing above- and below-average returns, as with risk measured by variance. Because market conditions are known to persist for a certain period of time, periods in which a portfolio suffers large losses are likely to be unevenly distributed over time. By reducing sudden large risks (downside risk), it is possible to reduce cumulative portfolio losses, such as drawdowns. For example, comparing (A) when λ ₁ = 10 and λ ₂ = 0 with (B) when λ ₂ = 0.1, case (A) is consistent with the mean-variance method of (Equation 1). When a backtest is performed using data from August 2005 to January 2024, the maximum two-week drawdown occurs at 4.67% in case (a), whereas in case (b) EVAR reduces the maximum two-week drawdown to 3.76%.

[Operational Model Management Processing]
Next, an operation model management process according to an embodiment of the present disclosure will be described. The operation model management process is executed by the above-described operation model management device 100, and more specifically, may be realized by one or more processors 102 of the operation model management device 100 executing one or more programs or instructions stored in one or more storage devices 101. Figure 5 is a flowchart showing the operation model management process according to an embodiment of the present disclosure.

As shown in FIG. 5, in step S101, the operation model management device 100 acquires investment performance information for multiple child operation models. Here, each child operation model is managed according to a predetermined investment strategy. For example, the investment performance information may be daily return data r for each child operation model. Based on the acquired daily return data r, the operation model management device 100 can calculate statistics such as the average return and variance-covariance matrix of each child operation model.

In step S102, the operation model management device 100 determines a portfolio for allocating the investment funds of the parent operation model to multiple child operation models based on the investment performance information of each child operation model. For example, the operation model management device 100 may determine the portfolio based on the degree of risk aversion to portfolio volatility. The operation model management device 100 may also determine the portfolio based on the degree of risk aversion to portfolio downside risk.

In step S103, the operation model management device 100 outputs the portfolio as investment support information. Specifically, the operation model management device 100 may provide the determined portfolio to the manager of the parent operation model as investment support information. Upon obtaining the investment support information, the manager may refer to the provided portfolio to determine how to allocate the investment funds of the parent operation model to each child operation model.

According to the above-described embodiment, an investment manager managing investment funds using a parent investment model can obtain a portfolio suitable for allocating investment funds to each child investment model as investment support information from the investment model management device 100, and manage investment funds through optimal fund allocation.

The above describes in detail examples of the present disclosure, but the present disclosure is not limited to the specific embodiments described above, and various modifications and variations are possible within the scope of the gist of the present disclosure as set forth in the claims.

20 Child operation model database 30 User terminal 100 Operation model management device 110 Child operation model management unit 120 Parent operation model management unit 130 Portfolio determination unit

Claims (4)

A memory that stores investment performance information of a plurality of child investment models, each of which is managed according to a predetermined investment strategy, and management status information that indicates the management status of a parent investment model that manages investment funds;
a processor that determines a portfolio for allocating investment funds of the parent investment model to the plurality of child investment models based on the expected return of each child investment model calculated based on the investment performance information acquired from the memory and the covariance matrix of the expected return, in accordance with the degree of risk aversion to volatility and downside risk;
An operation model management device having: The operation model management device of claim 1, wherein the processor calculates the expected return of the portfolio based on the expected return of each child operation model. Acquiring from memory investment performance information of a plurality of child investment models, each of which is managed according to a predetermined investment strategy, and management status information indicating the management status of a parent investment model that manages the investment funds;
determining a portfolio for allocating investment funds of the parent investment model to the plurality of child investment models based on the expected return of each child investment model calculated based on the investment performance information and the covariance matrix of the expected return, in accordance with the degree of risk aversion to volatility and downside risk;
outputting the portfolio as investment support information;
A computer-implemented operational model management method. Acquiring from memory investment performance information of a plurality of child investment models, each of which is managed according to a predetermined investment strategy, and management status information indicating the management status of a parent investment model that manages the investment funds;
determining a portfolio for allocating investment funds of the parent investment model to the plurality of child investment models based on the expected return of each child investment model calculated based on the investment performance information and the covariance matrix of the expected return, in accordance with the degree of risk aversion to volatility and downside risk;
outputting the portfolio as investment support information;
An operational model management program that causes a computer to execute the above.