Model training method and device, electronic equipment, medium and product

Abstract

This application provides a model training method, apparatus, electronic device, medium, and product, relating to the field of artificial intelligence. The method includes: acquiring bond fund data, the bond fund data including the total return of the bond fund; inputting the bond fund data into an attribution model, and using the attribution model to decompose the total return of the bond fund into three sub-items: income effect, price effect, and trading return, as the performance attribution result corresponding to the bond fund data; wherein the trading return is used to characterize the contribution of intraday bond trading to the total return of the bond fund; and training a fund management model based on the performance attribution result, the fund management model being used to generate investment decisions for the bond fund. The method of this application can accurately and reliably train a fund management model, providing support for investment decisions.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and more particularly to a model training method, apparatus, electronic device, medium, and product. Background Technology

[0002] With the rapid development of financial markets, a large number of financial products have emerged. Bond funds are one of the important financial products, referring to a type of fund that invests more than 80% of its assets in bonds. These funds achieve portfolio investment through professional team management, mainly investing in fixed-income assets such as government bonds, financial bonds, and corporate bonds, and are characterized by stable returns and low risk.

[0003] Currently, the management model is used to process transaction information and news reports, automatically generating investment recommendations for bond funds and automating the transaction process. However, the accuracy of the investment recommendations from the management model still falls short of requirements, resulting in returns not meeting expectations.

[0004] Therefore, there is an urgent need for a reliable model training method for bond fund returns. Summary of the Invention

[0005] This application provides a model training method, apparatus, electronic device, medium, and product for accurately and reliably training fund management models to support investment decisions.

[0006] Firstly, this application provides a model training method, including:

[0007] Obtain bond fund data, including the total returns of bond funds;

[0008] The bond fund data is input into an attribution model, which decomposes the total return of the bond fund into three sub-items: income effect, price effect, and trading return. These sub-items serve as the performance attribution results corresponding to the bond fund data. Among them, the trading return is used to characterize the contribution of intraday bond trading to the total return of the bond fund.

[0009] Based on the performance attribution results, a fund management model is trained, which is used to generate investment decisions for bond funds.

[0010] Secondly, this application provides a model training apparatus, comprising:

[0011] The acquisition module is used to acquire bond fund data, which includes the total return of the bond fund.

[0012] The attribution module is used to input the bond fund data into the attribution model, and the attribution model decomposes the total return of the bond fund into three sub-items: income effect, price effect, and trading return, which serve as the performance attribution results corresponding to the bond fund data; wherein, the trading return is used to characterize the contribution of intraday bond trading to the total return of the bond fund;

[0013] The training module is used to train a fund management model based on the performance attribution results, and the fund management model is used to generate investment decisions for bond funds.

[0014] Thirdly, this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0015] The memory stores computer-executed instructions;

[0016] The processor executes computer execution instructions stored in the memory to implement the method described above.

[0017] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the method described above.

[0018] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the method described above.

[0019] The model training method, apparatus, electronic equipment, medium, and product provided in this application use an attribution model to decompose the total return of bond funds into three sub-items: income effect, price effect, and trading return. This enables performance attribution of bond fund returns under the decision-making of the fund management model, effectively uncovering the intraday trading returns of bonds that users are concerned about, making the attribution model more comprehensive, effectively evaluating the trading capabilities of the fund management model, and improving the accuracy and reliability of fund management model training. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0021] Figure 1 A schematic flowchart illustrating the model training method provided in this application embodiment;

[0022] Figure 2 This is a schematic diagram of the structure of the attribution model provided in the embodiments of this application;

[0023] Figure 3This is a schematic diagram of the structure of the model training device provided in the embodiments of this application;

[0024] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0025] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0026] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0027] It should be noted that the model training methods, devices, electronic devices, media and products provided in this application can be used in the field of artificial intelligence, or in any field other than artificial intelligence. The application fields of the model training methods, devices, electronic devices, media and products in this application are not limited.

[0028] First, the terms used in the embodiments of this application will be explained.

[0029] Yield to maturity (YTM) of a bond: refers to the annualized return assuming an investor buys a bond and holds it until maturity.

[0030] The coupon rate of a bond is the interest rate stated on the bond certificate, which is used to calculate and pay interest on the bond throughout its life until maturity.

[0031] The par value of a bond refers to its face value, which represents the amount the issuer promises to pay to bondholders on a specific future date.

[0032] Macaulay duration: The term "duration" commonly refers to Macaulay duration. It is calculated by discounting future cash flows to their present value using yield, multiplying each present value by the time elapsed since the cash flow occurred, summing the results, and then dividing this sum by the bond price. It can also be understood as a weighted average of the time elapsed since each cash flow occurred.

[0033] Modified duration: measures how many percentage points the bond price will change when the bond yield changes by 1%. Modified duration = duration / (1 + YTM).

[0034] With the rapid development of financial markets, a large number of financial products have emerged. Bond funds are one of the important financial products, referring to a type of fund that invests more than 80% of its assets in bonds. These funds achieve portfolio investment through professional team management, mainly investing in fixed-income assets such as government bonds, financial bonds, and corporate bonds, and are characterized by stable returns and low risk.

[0035] Currently, the management model is used to process transaction information and news reports, automatically generating investment recommendations for bond funds and automating the transaction process. However, the accuracy of the investment recommendations from the management model still falls short of requirements, resulting in returns not meeting expectations.

[0036] Therefore, there is an urgent need for a reliable model training method for bond fund returns.

[0037] The model training methods, apparatus, electronic devices, media, and products provided in this application are intended to solve the above-mentioned technical problems of the prior art.

[0038] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0039] Figure 1 This is a schematic flowchart illustrating the model training method provided in an embodiment of this application. Figure 1 As shown, the method includes:

[0040] S101. Obtain bond fund data, which includes the total return of bond funds.

[0041] S102. Input the bond fund data into the attribution model, and decompose the total return of the bond fund into three sub-items: income effect, price effect, and trading return, which serve as the performance attribution results corresponding to the bond fund data; among them, trading return is used to characterize the contribution of intraday bond trading to the total return of the bond fund.

[0042] S103. Based on the performance attribution results, train the fund management model, which is used to generate investment decisions for bond funds.

[0043] Figure 2 This is a schematic diagram of the attribution model provided in an embodiment of this application. Figure 2 As shown in the embodiments of this application, the attribution model is used to decompose the total return of a bond fund into multiple explainable return segments to identify the contribution of different factors to the return, so that users can better understand the source of the combined return of the bond fund and the driving factors behind it.

[0044] The attribution model breaks down total returns into three sub-components: income effect, price effect, and trading return, thus achieving performance attribution. Trading return represents the price difference gains resulting from buying and selling bonds during intraday trading.

[0045] For example, the transaction return is the difference between the closing price of the day and the average purchase price, or the difference between the average selling price and the closing price of the previous day.

[0046] If intraday trading is a net purchase, then the trading return = closing price of the day - average purchase price; if intraday trading is a net sale, then the trading return = average sale price - previous day's closing price.

[0047] Furthermore, attribution models can be used to evaluate the total returns generated by the fund management model's decisions. Price effects can be used to examine the fund management model's ability to predict macroeconomic and interest rate trends, while trading returns can be used to examine its market timing capabilities. Adjustments to the fund management model's parameters and rules can then be made to train the model.

[0048] In this embodiment, an attribution model is used to decompose the total return of a bond fund into three sub-items: income effect, price effect, and trading return. This enables performance attribution of bond fund returns under the fund management model's decision-making, effectively uncovering the intraday trading returns of bonds that users are interested in. This makes the attribution model more comprehensive, effectively assesses the trading capabilities of the fund management model, and improves the accuracy and reliability of the fund management model training.

[0049] In one implementation, the income effect is used to characterize the coupon income earned during the holding period of the bond.

[0050] Optionally, the income effect refers to the coupon income from the bonds held, i.e., the fixed interest income earned during the holding period. This is the most stable part of a bond fund's total return, unaffected by market price fluctuations. The income effect mainly stems from the coupon rate and holding period of the bonds held.

[0051] For example, the income effect is the product of the face value of the bond held, the current coupon rate, and the holding period, divided by the initial price.

[0052] The income effect is calculated as: coupon income of the held bonds × initial price.

[0053] Coupon income = face value of bonds held × current coupon rate × holding period.

[0054] Therefore, the income effect = face value of bonds held × current coupon rate × holding period × initial price.

[0055] In one implementation, the price effect is used to characterize the gains or losses resulting from changes in the price of the bonds held.

[0056] like Figure 2 As shown, the price effect can be further subdivided. In one implementation, the price effect includes the government bond effect, the interest rate spread effect, and the bond selection effect. Bond fund data is input into an attribution model, which decomposes the total return of bond funds into three sub-items: income effect, price effect, and trading return.

[0057] Input bond fund data into the attribution model, and then break down the total return of bond funds into five sub-items: income effect, government bond effect, interest rate spread effect, bond selection effect, and trading return.

[0058] In practice, the price effect can be further subdivided into three components: the government bond effect, the interest rate spread effect, and the bond selection effect. Therefore, the attribution model can decompose the total return of bond funds into five sub-components: income effect, government bond effect, interest rate spread effect, bond selection effect, and trading return. This makes the attribution model's decomposition of bond returns more comprehensive and complete, and improves the interpretability of performance attribution.

[0059] Optionally, the Treasury bond effect is used to characterize the changes in the price of held bonds caused by changes in Treasury bond interest rates.

[0060] The government bond effect reflects the impact of changes in the risk-free interest rate (such as the government bond yield) on the price of bonds held. When the market risk-free interest rate falls, bond prices rise; when the market risk-free interest rate rises, bond prices fall.

[0061] For example, the government bond effect is the inverse product of the modified duration of the bond holdings and the change in government bond yield.

[0062] In practice, the Treasury bond effect equals the adjusted duration of the bond holdings multiplied by the change in Treasury bond yield. Adjusted duration quantifies the sensitivity of bond prices to changes in interest rates.

[0063] Optionally, the spread effect is used to characterize the price changes of held bonds caused by changes in the benchmark spread;

[0064] The credit spread effect refers to the impact of changes in credit spreads on the prices of bonds held. Changes in credit spreads can be derived by comparing changes in the yield of held bonds with changes in the risk-free rate over the same period, reflecting the credit risk of the bonds. When credit spreads narrow, bond prices rise; when credit spreads widen, bond prices fall.

[0065] For example, the spread effect is the inverse product of the modified duration of the bond holdings and the change in the spread.

[0066] In practice, the interest rate spread is the difference between the yield of the bonds held and the risk-free rate (for government bonds of the same maturity). The interest rate spread effect = - the modified duration of the bonds held × the change in the interest rate spread.

[0067] In this process, instead of using simple interpolation, the yield of government bonds with the same maturity is taken from the yield curve, thereby improving the accuracy of the attribution model.

[0068] Optionally, the bond selection effect is used to characterize the excess returns or losses resulting from bond selection.

[0069] The bond selection effect can reflect the excess returns brought about by the fund manager's active bond selection ability, that is, the residual returns that cannot be explained by the government bond effect and the interest rate spread effect.

[0070] The bond selection effect is the total return of bond funds minus the income effect, the government bond effect, the interest rate spread effect, and the transaction return.

[0071] In practice, after breaking down the total return of a bond fund into the income effect, government bond effect, interest rate spread effect, and trading return, the remaining term is the bond selection effect. Bond selection effect = Total return of bond fund - Income effect - Government bond effect - Interest rate spread effect - Trading return.

[0072] In this embodiment, an attribution model is used to decompose the total return of a bond fund into three sub-items: income effect, price effect, and trading return. This enables performance attribution of the total return of the bond fund, effectively uncovering the intraday trading returns of bonds that users are interested in. This makes the attribution model more comprehensive, effectively assesses the fund manager's timing and trading abilities, improves the interpretability of performance attribution, and provides users with more detailed and comprehensive references for subsequent investment decisions and asset allocation.

[0073] Before using attribution models for performance attribution, data collection is required. In some embodiments, the method further includes:

[0074] Obtain the face value, coupon rate, daily closing price, adjusted duration, and initial price of the bonds held from the data terminal;

[0075] Additionally, data on bond funds is obtained from the custody system, including the total return of the bond funds, the holding period of the bonds held, the average purchase price, and the average selling price.

[0076] Optionally, you can obtain the face value, coupon rate, daily closing price, modified duration, and initial price of the bonds you hold from official and authoritative channels such as the China Bond Information Network, the official websites of the Shanghai Stock Exchange / Shenzhen Stock Exchange, and the Shanghai Clearing House (SHCH).

[0077] Optionally, you can obtain the face value, coupon rate, daily closing price, modified duration, and initial price of the bonds you hold from data terminals such as Wind, Choice, iFinD, or other free financial data platforms.

[0078] Optionally, the face value, coupon rate, daily closing price, modified duration, and initial price of the bonds held can be obtained from the bank over-the-counter market.

[0079] Optionally, valuation data and certificate data of bond funds can be obtained from the custody system, and the total return of bond funds, holding period of bonds, average purchase price and average selling price can be extracted from them.

[0080] In some embodiments, the method further includes:

[0081] The performance attribution results are input into the trained fund management model so that the fund management model can adjust investment decisions based on the performance attribution results.

[0082] In practical implementation, the fund management model can be adjusted not only during training based on performance attribution results, but also in its actual application, effectively improving its reliability and accuracy. For example, if the market timing contribution is positive, it indicates that the fund management model's judgment of major macroeconomic turning points is relatively accurate. In this case, the current macroeconomic analysis framework should be maintained, and the threshold for position adjustment should even be appropriately increased to amplify the excess returns brought by market timing. If the market timing contribution is negative, it indicates that frequent market timing actually drags down returns. In this case, the fund management model can make a decision to "weaken market timing and anchor to the central point."

[0083] The above describes the model training method provided in the embodiments of this application. Figure 3 This is a schematic diagram of the structure of the model training device provided in an embodiment of this application. Figure 3 As shown, the method includes:

[0084] Module 31 is used to acquire bond fund data, which includes the total return of bond funds.

[0085] Attribution module 32 is used to input bond fund data into the attribution model. The attribution model decomposes the total return of the bond fund into three sub-items: income effect, price effect, and trading return, which serve as the performance attribution results corresponding to the bond fund data. Among them, trading return is used to characterize the contribution of intraday bond trading to the total return of the bond fund.

[0086] Training module 33 is used to train the fund management model based on the performance attribution results. The fund management model is used to generate investment decisions for bond funds.

[0087] In practical applications, model training devices can be implemented through computer programs, such as application software; or they can be implemented as media storing relevant computer programs, such as USB flash drives or cloud drives; or they can be implemented through physical devices that integrate or install relevant computer programs, such as chips or servers.

[0088] It should be noted that the model training device is used to execute the model training method as described above. For the specific implementation method, please refer to the method embodiment provided in the embodiments of this application, which will not be repeated here.

[0089] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, such as... Figure 4 As shown, the electronic device includes:

[0090] The electronic device includes a processor 291 and a memory 292; it may also include a communication interface 293 and a bus 294. The processor 291, memory 292, and communication interface 293 can communicate with each other via the bus 294. The communication interface 293 can be used for information transmission. The processor 291 can invoke logical instructions stored in the memory 292 to execute the methods of the above embodiments.

[0091] Furthermore, the logic instructions in the aforementioned memory 292 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0092] The memory 292, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as program instructions / modules corresponding to the methods in the embodiments of this application. The processor 291 executes functional applications and data processing by running the software programs, instructions, and modules stored in the memory 292, thereby implementing the methods in the above-described method embodiments.

[0093] The memory 292 may include a program storage area and a data storage area. The program storage area may store the operating system and application programs required for at least one function; the data storage area may store data created based on the use of the terminal device. Furthermore, the memory 292 may include high-speed random access memory and may also include non-volatile memory.

[0094] This application provides a non-transitory computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the methods described in the foregoing embodiments.

[0095] This application provides a computer program product, including a computer program that, when executed by a processor, implements the methods provided in any of the embodiments described above.

[0096] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0097] It should be further noted that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0098] It should be understood that the above-described device embodiments are merely illustrative, and the device of this application can also be implemented in other ways. For example, the division of units / modules in the above embodiments is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units, modules, or components may be combined, or integrated into another system, or some features may be ignored or not executed.

[0099] Furthermore, unless otherwise specified, the functional units / modules in the various embodiments of this application can be integrated into one unit / module, or each unit / module can exist physically separately, or two or more units / modules can be integrated together. The integrated units / modules described above can be implemented in hardware or as software program modules.

[0100] When integrated units / modules are implemented in hardware, the hardware can be digital circuits, analog circuits, etc. The physical implementation of the hardware structure includes, but is not limited to, transistors, memristors, etc. Unless otherwise specified, the processor can be any suitable hardware processor, such as a CPU, GPU, FPGA, DSP, and ASIC, etc. Unless otherwise specified, the storage unit can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc.

[0101] If the integrated unit / module is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0102] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0103] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0104] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A model training method, characterized in that, include: Obtain bond fund data, including the total returns of bond funds; The bond fund data is input into an attribution model, which decomposes the total return of the bond fund into three sub-items: income effect, price effect, and trading return. These sub-items serve as the performance attribution results corresponding to the bond fund data. Among them, the trading return is used to characterize the contribution of intraday bond trading to the total return of the bond fund. Based on the performance attribution results, a fund management model is trained, which is used to generate investment decisions for bond funds.

2. The method according to claim 1, characterized in that, The income effect is used to characterize the coupon income obtained during the holding period of the bond; The price effect is used to characterize the gains or losses resulting from changes in the price of the bonds held.

3. The method according to claim 1, characterized in that, The price effect includes the government bond effect, interest rate spread effect, and bond selection effect; the process involves inputting the bond fund data into an attribution model, which then decomposes the total return of the bond fund into three sub-items: income effect, price effect, and trading return. The data of the bond fund is input into the attribution model, which decomposes the total return of the bond fund into five sub-items: income effect, government bond effect, interest rate spread effect, bond selection effect, and trading return.

4. The method according to claim 3, characterized in that, The so-called government bond effect is used to characterize the changes in the price of held bonds caused by changes in government bond interest rates; The interest rate spread effect is used to characterize the price changes of held bonds caused by changes in the benchmark interest rate spread; The bond selection effect is used to characterize the excess returns or losses resulting from bond selection.

5. The method according to claim 4, characterized in that, The income effect is the product of the face value of the bonds held, the current coupon rate, and the holding period, divided by the initial price. The so-called government bond effect is the inverse product of the modified duration of the held bonds and the change in government bond yield; The aforementioned spread effect is the inverse product of the adjusted duration of the held bonds and the change in the spread; The transaction return is the difference between the closing price of the day and the average purchase price, or the difference between the average selling price and the closing price of the previous day. The bond selection effect is the total return of the bond fund minus the income effect, the government bond effect, the interest rate spread effect, and the transaction return.

6. The method according to claim 1, characterized in that, The method further includes: The performance attribution results are input into the trained fund management model so that the fund management model can adjust investment decisions based on the performance attribution results.

7. The method according to any one of claims 1-6, characterized in that, The method further includes: Obtain the face value, coupon rate, daily closing price, adjusted duration, and initial price of the bonds held from the data terminal; Additionally, the total return of the bond fund and the holding period, average purchase price, and average selling price of the bonds held are obtained from the custody system to obtain the bond fund data.

8. A model training device, characterized in that, include: The acquisition module is used to acquire bond fund data, which includes the total return of the bond fund. The attribution module is used to input the bond fund data into the attribution model, and the attribution model decomposes the total return of the bond fund into three sub-items: income effect, price effect, and trading return, which serve as the performance attribution results corresponding to the bond fund data; wherein, the trading return is used to characterize the contribution of intraday bond trading to the total return of the bond fund; The training module is used to train a fund management model based on the performance attribution results, and the fund management model is used to generate investment decisions for bond funds.

9. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 6.

11. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

Images