Price estimation system

Abstract

To provide a price estimation system for efficiently and accurately calculating the price of an artwork.SOLUTION: A price estimation system comprises: a transaction result storage unit 22 which stores auction information including information related to a work identifier, an artist name, an attribute, a market price, a transaction date, and a transaction currency; and a control unit 21 which calculates the price of an evaluation object. Then, the control unit 21: estimates work size from the attribute of the auction information; identifies an original or a copy from the attribute of the auction information; identifies a work genre from the attribute of the auction information; estimates a parameter of a logarithm price linear regression model by a least squares method; and calculates a price theoretical value calculation formula.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a price estimation system for calculating the price of a work of art. [Background technology]

[0002] The price of an artwork can vary depending on where it is bought and sold. For example, in art galleries and department stores, prices are determined taking into account not only the cost of materials but also labor costs and the artist's career and activities. At auctions, the price is determined by the highest bidder for a work displayed by a seller (see, for example, Patent Document 1). The technology described in this document calculates future royalties and other income that the copyright holder of a work will be able to receive. Specifically, the average income and income variance of past annual income are calculated based on information on actual copyright royalties incurred for the work. The appraised price is then calculated using a formula that uses the average income and variance as variables.

[0003] Also under consideration is an art purchase and sale intermediation support system for reducing the financial risk of intermediaries due to the amount paid to sellers (see, for example, Patent Document 2). The art purchase and sale intermediation support device described in this document obtains an appraisal price calculated based on the appraisal results of the art piece that the seller wishes to sell. Then, based on the appraisal price, it calculates an amount obtained by deducting a safety deposit amount appropriate to the art piece from the appraisal price of the art piece as a first payment amount to be paid to the seller before the auction. Furthermore, based on the input appraisal price and the successful bid price of the art piece sold at auction, it calculates a second payment amount to be paid to the seller after the auction. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2022-106684 [Patent Document 2] Patent No. 7347878 Summary of the Invention [Problem to be solved by the invention]

[0005] It is generally impossible to calculate the value of an artwork from the cost of materials, functionality, etc. In other words, experts arbitrarily estimate the price based primarily on the visual elements of the artwork. For this reason, it is difficult to say that the price of an artwork is a reproducible figure that accurately reflects its value. Furthermore, as mentioned above, the price of the same work may vary depending on where it is sold. This lack of transparency in prices poses a challenge in terms of not increasing the number of participants in the art trade.

[0006] Figure 6 shows the price distribution of art auctions. Art auction prices tend to be overly concentrated in certain price ranges, and exhibit a distribution with a fat tail that shows high prices that cannot be represented by a normal distribution. This results in prices that are different from those of ordinary goods and services.

[0007] Figure 7 shows the annual fluctuations in winning bid prices. The high prices that make up the fat tail not only occur frequently, but also occur every year. Figure 8 shows the results of a simple multivariate analysis (multiple regression analysis) based on a normal distribution. The horizontal axis represents the actual value, and the vertical axis represents the price estimated by multivariate analysis. As can be seen from the distribution above, the coefficient of determination adjusted for the degrees of freedom for price estimation is 0.36, which means that good results are not obtained. [Means for solving the problem]

[0008] A price estimation system that solves the above problem includes a transaction history storage unit that stores auction information including information on the work identifier, artist name, attributes, transaction price, transaction date, and transaction currency, and a control unit that calculates the price of the item being evaluated. The control unit estimates the size of the work from the attributes of the auction information, identifies whether it is an original or a copy from the attributes of the auction information, identifies the genre of the work from the attributes of the auction information, estimates the parameters of a logarithmic price linear regression model A using the least squares method, and calculates a theoretical price value calculation formula. [Effects of the Invention]

[0009] The present invention allows for the efficient and accurate calculation of art prices. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is an explanatory diagram of a price estimation system according to an embodiment of the present invention. [Figure 2] FIG. 2 is an explanatory diagram of a hardware configuration of the present embodiment. [Figure 3] FIG. 2 is an explanatory diagram of a processing procedure according to the present embodiment. [Figure 4] FIG. 2 is an explanatory diagram of a processing procedure according to the present embodiment. [Figure 5] 1A to 1D are explanatory diagrams of the analysis results of the multivariate analysis of this embodiment, where (a) is an explanatory diagram of Pablo Picasso, (b) is an explanatory diagram of Marc Chagall, (c) is an explanatory diagram of Yayoi Kusama, and (d) is an explanatory diagram of Gerhard Richter. [Figure 6] FIG. 1 is an explanatory diagram of price characteristics of a conventional art auction. [Figure 7] FIG. 10 is an explanatory diagram of a conventional analysis result. [Figure 8] FIG. 1 is an explanatory diagram of the analysis results of conventional multivariate analysis. DETAILED DESCRIPTION OF THE INVENTION

[0011] An embodiment of a price estimation system will be described below with reference to Figures 1 to 5. In this embodiment, it is assumed that the price of a painting is to be calculated as an art piece. As shown in FIG. 1, in this embodiment, an auction site 10, a user terminal 15, and a management server 20 are used.

[0012] (Hardware configuration) 2, the hardware configuration of the information processing device H10 that constitutes the auction site 10, user terminal 15, and management server 20 will be described. The information processing device H10 includes a communication device H11, an input device H12, a display device H13, a storage device H14, and a processor H15. Note that this hardware configuration is an example, and it can also be realized by other hardware.

[0013] The communication device H11 is an interface that establishes a communication path with other devices and executes data transmission and reception, and is, for example, a network interface card or a wireless interface.

[0014] The input device H12 is a device that accepts input from a user, etc., and is, for example, a mouse, a keyboard, etc. The display device H13 is, for example, a display that displays various information. The storage device H14 is a storage device that stores data and various programs for executing various functions of the auction site 10, the user terminal 15, and the management server 20. Examples of the storage device H14 include a ROM, a RAM, and a hard disk.

[0015] The processor H15 uses programs and data stored in the storage device H14 to control each process in the auction site 10, the user terminal 15, and the management server 20. Examples of the processor H15 include a CPU and an MPU. The processor H15 loads programs stored in a ROM or the like into a RAM and executes various processes for each service.

[0016] (System Functions) Next, the functions of the auction site 10, the user terminal 15, and the management server 20 will be described with reference to FIGS.

[0017] The auction site 10 is a computer system that mediates auction transactions between sellers and buyers of artworks (art). The auction site 10 manages attribute information of artworks put up for auction and information regarding successful bids. The user terminal 15 is a computer terminal used by a user to calculate the price of an artwork.

[0018] The management server 20 is a computer system that calculates the prices of artworks. The management server 20 includes a control unit 21, a transaction record storage unit 22, and a model information storage unit 23. The control unit 21 performs the processes described below (each process of the acquisition stage, preprocessing stage, creation stage, evaluation stage, etc.) By executing a price calculation program for this purpose, the control unit 21 functions as an acquisition unit 211, a preprocessing unit 212, a creation unit 213, an evaluation unit 214, etc.

[0019] The acquisition unit 211 executes a process of acquiring various information from the auction site 10 and the user terminal 15 . The pre-processing unit 212 performs the following processes: size standardization of artworks, originality determination, and artist name matching. In the size standardization process, the area is estimated using the size in the attribute information of the artwork. In the original determination process, the material is identified in the attribute information of the artwork and it is determined whether it is an original work (one-of-a-kind) or a copy (multiple copies of the same work, such as prints). In the artist name matching process, the artist name in the attribute information of the artwork is used to estimate whether the artwork belongs to the same artist. For this reason, the pre-processing unit 212 stores matching pattern information for grouping together different artist names for the same artist.

[0020] The creation unit 213 executes a process of creating a model for calculating prices from the trading history of artworks. The evaluation unit 214 executes a process for calculating the price of the artwork designated by the user.

[0021] The transaction history storage unit 22 stores transaction history management data related to the prices of artworks that have been sold at art auctions. The transaction history management data is recorded when successful bid information is acquired from the auction site 10. This transaction history management data stores information related to the successful bid date, successful bid price, and attributes.

[0022] The successful bid date information is information relating to the date on which the bid was made at the art auction (transaction date). The successful bid price information is information about the transaction price at which the bid was made, including information about the currency in which the bid was made (such as USD for the United States or GBP for the United Kingdom).

[0023] The attribute information is information about the artwork disclosed at the art auction, and includes data such as the name of the artwork, the artist's name, the year of completion, the materials used, the genre of the artwork, and the size.

[0024] The work name is the name of each work. The work name, artist name, and year of completion can be used as a work identifier to identify the work. The author name is an author identifier for identifying the author of this work. The year of completion is information that indicates the year this work was completed.

[0025] The compositional material is information about the medium, such as oil painting, watercolor, etc. This compositional material classification is used in the originality determination to determine whether the work is an original or a copy. The work genre is information indicating the categorization of the work, such as "contemporary art" or "post-war art." The size is information that indicates the size of the work.

[0026] A prediction model for calculating the price of an art piece is recorded in the model information storage unit 23. This model management data is recorded when a prediction model is generated in the learning stage.

[0027] The prediction model is information for calculating the price of an artwork. In this embodiment, the following logarithmic price linear regression model (A), theoretical price calculation formula (B), and price index formula (C) are used. [Formula 1] shows the log-price linear regression model (A). The log-price linear regression model (A) is a model for estimating prices for each parameter and is calculated using the least squares method. [Formula 1] assumes that the sample size of works obtained from art auction information is "n," the number of artists included in the representative group is "J," and the maximum number of years in which works were sold is "K." Artists are classified as representative and non-representative artists based on the frequency of auction transactions. For example, a representative artist is an artist whose works have been traded at auction more than a certain number of times in a given period (e.g., 40 times in the past 20 years). Representative artists are then evaluated individually as one artist, while non-representative artists are collectively evaluated as one artist.

[0028]

number

[0029] P ijk : The winning bid price of work i by artist j in year k, A[j] i : A dummy variable that is "1" if work i is by artist j, and "0" otherwise. T[k] i : A dummy variable that is "1" if the transaction year of work i is k, and "0" otherwise. O i : "Original flag" is used as a dummy variable that is set to "1" if work i is an original work and "0" otherwise. X i :Another multidimensional control variable matrix for work i, which includes the following variables as dimensions:

[0030] Artwork size [number] Variable obtained by squaring the work size [number] Dummy variable for the currency unit used in the auction Dummy variables for the currency unit used in the auction and the transaction year dummy variable T[k] i Cross-term variables of - Quarter dummy variable at the time of transaction (1st to 4th quarter) Dummy variables for the genre of artwork (variables indicating the category of artwork, such as "contemporary art" or "post-war art") Dummy variable for work genre α j ,β k ,γ,δ jk ,ζ j ,η k ,θ jk , b: unknown parameters to be estimated u ijk :Error term [Formula 2] is the theoretical price calculation formula (B) that calculates the theoretical price. The theoretical price calculation formula (B) calculates the estimated price of the work under the prerequisite that all necessary variable data, such as the artist's name, size, and year of transaction, are available.

[0031]

number

[0032] [Equation 3] is the price index formula (C) for a representative artist. The price index indicates the price growth rate at a certain point in time over time, with the price in the reference base year (for example, 2000) set at "1." When you want to estimate the price of a work by a certain artist, if there is no transaction record for that transaction year, the price index formula (C) is used to supplement the data for that transaction year.

[0033]

number

[0034] [Equation 4] is the price index equation (D) for non-representative artists.

[0035]

number

[0036] The missing values ​​are filled in using the price index values ​​derived from price index formulas (C) and (D) as proxy variables. Specifically, by using price index formulas (C) and (D), it is possible to calculate price trends for each representative artist and non-representative artist that are independent of the work.

[0037] <Outline of processing procedure> An outline of the processing procedure will be explained using Figures 3 to 5. Here, the learning process (Figure 3) and the prediction process (Figure 4) will be explained in that order. (Learning process) The learning process will be described with reference to FIG. First, the control unit 21 of the management server 20 executes a process for acquiring auction information (step S11). Specifically, the acquisition unit 211 of the control unit 21 collects the successful bid price data and associated attribute data (artist name, constituent materials, estimated lower limit price, estimated upper limit price, size, etc.) posted on the auction site's website after the art auction is held. The acquisition unit 211 then generates transaction history management data that records the acquired successful bid information and records it in the transaction history storage unit 22.

[0038] Next, the control unit 21 of the management server 20 repeats the following process for each work. Here, the control unit 21 of the management server 20 executes a size determination process (step S12). Specifically, the preprocessing unit 212 of the control unit 21 identifies the size of the artwork using the attribute data. Here, the area of ​​the artwork size model is calculated from the size information in the attribute data.

[0039] Next, the control unit 21 of the management server 20 executes an original determination process (step S13). Specifically, the preprocessing unit 212 of the control unit 21 uses the attribute data to determine whether the work is an original or a copy. Here, material information in the attribute data is determined. The classification of mediums (component materials) such as "oil painting," "watercolor painting," and "sculpture" is used to determine whether the work is an original or a copy.

[0040] Next, the control unit 21 of the management server 20 executes a name collation process (step S14). Specifically, the pre-processing unit 212 of the control unit 21 performs name collation of author names using the attribute data. Then, the above processing (steps S12 to S14) is repeatedly executed for each work.

[0041] Next, the control unit 21 of the management server 20 executes a model creation process (step S15). Specifically, first, the creation unit 213 of the control unit 21 estimates each parameter. Here, each parameter is calculated using the least squares method in the logarithmic price linear regression model (A) using the preprocessed data recorded in the transaction history storage unit 22. Furthermore, the theoretical value of the logarithmic price is calculated using the price theoretical value calculation formula (B). In calculation formula (A), the reference standard for the variable (author dummy variable) that is "1" if work i is by author j and "0" otherwise is a non-representative group of authors. The reference standard for the variable (year dummy variable) that is "1" if work i is traded in year k and "0" otherwise is the data for the year 2000. The reference standard for the variable (original dummy variable) that is set to "1" if work i is an original work and "0" otherwise using the "original flag" is the data for copies. The reference standard for the variable for the currency unit used in the auction (currency dummy variable) is data traded in USD. The reference standard for the quarter dummy variable for the time of transaction is data traded in the first quarter. The reference standard for the dummy variable for the genre of artwork is data for artworks classified as from the 19th century.

[0042] Then, for each artist, the control unit 21 of the management server 20 executes a price index calculation process (step S16). Specifically, the creation unit 213 of the control unit 21 calculates a price index from the obtained regression parameters. Here, a price index for the original works and a price index for the copies of each representative artist, as well as a price index for the original works and a price index for the copies of the other aggregated non-representative artists, are calculated.

[0043] The price index of the original works of each representative artist in the price index formula (C) includes the following: Individual effect of representative writers (α j ) estimate - Annual individual effect (β k ) estimate Estimated original work effect (γ) · Interaction effect between representative author and year (δ jk ) estimate · Interaction effect between representative artists and original works (ζ j ) estimate Interaction effect between year and original work (η k ) estimate · The interaction effect of representative author, year, and original work (θ jk ) estimate The price index of each representative artist's copy work is calculated by taking into account the individual effect of the representative artist (α j ), individual effects of year (β k ), the interaction effect between representative author and year (δ jk ) is calculated based on the estimated value of

[0044] The price index of original works by non-representative artists is calculated by the individual effects of the year (β k ), original work effect (γ), interaction effect between year and original work (η k ) is calculated based on the estimated value of The price index of copies of non-representative artists in the price index formula (D) is calculated by taking into account the individual effects of the year (β k ) is calculated based on the estimated value of

[0045] (Prediction stage) The prediction process will be described with reference to FIG.

[0046] First, the control unit 21 of the management server 20 executes a process for acquiring evaluation target information (step S21). Specifically, the evaluation unit 214 of the control unit 21 acquires evaluation target information about the evaluation target from the user terminal 15. This evaluation target information includes information about the work, the artist, and the year of transaction.

[0047] Next, the control unit 21 of the management server 20 executes a price estimation process (step S22). Specifically, the preprocessing unit 212 of the control unit 21 executes an author name identification process based on the evaluation target information. Then, the evaluation unit 214 identifies representative authors and non-representative authors.

[0048] Next, the evaluation unit 214 calculates the parameters (α j ,β k ,γ,δ jk ,ζ j ,η k ,θ jk , b) are substituted into the theoretical price calculation formula (B) to estimate the price. Here, artist j uses the identified representative artist and non-representative artist. Furthermore, the evaluation unit 214 complements the artist's price trend by calculating a price index derived from the price index formulas (C) and (D). The evaluation unit 214 then outputs the estimated price and the artist's price trend to the display device H13 of the user terminal 15.

[0049] Figure 5 is an explanatory diagram of the results of multivariate analysis. It compares the actual auction winning bid prices with the theoretical values ​​calculated using the theoretical price calculation formula (B). Figure 5(a) is an explanatory diagram for Pablo Picasso, Figure 5(b) for Marc Chagall, Figure 5(c) for Yayoi Kusama, and Figure 5(d) for Gerhard Richter. The coefficients of determination adjusted for the degrees of freedom for price estimation were (a: Pablo Picasso) 0.8872, (b: Marc Chagall) 0.9568, (c: Yayoi Kusama) 0.8564, and (d: Gerhard Richter) 0.9608, demonstrating favorable results for each.

[0050] As described above, according to this embodiment, the following effects can be obtained. (1) In this embodiment, the control unit 21 of the management server 20 executes a size determination process (step S12), which allows art pieces of various sizes that affect prices to be standardized by area.

[0051] (2) In this embodiment, the control unit 21 of the management server 20 executes an original determination process (step S13). Original works (one-of-a-kind items) and copies (prints and other items with multiple copies of the same work) tend to have different influences on the price of artworks, so they can be distinguished and a prediction model can be created.

[0052] (3) In this embodiment, the control unit 21 of the management server 20 executes a name integration process (step S14), which allows the same author who has been registered under different names due to variations in spelling, etc., to be unified.

[0053] (4) In this embodiment, authors are divided into representative authors and non-representative authors according to the frequency of auction transactions. This allows authors with few transactions to be collectively evaluated as non-representative authors.

[0054] (5) In this embodiment, the control unit 21 of the management server 20 executes a model creation process (step S15), which allows creation of a prediction model that estimates prices according to the artwork, artist, and transaction year.

[0055] Multidimensional control variable matrix X i uses a variable that is the square of the work size [number]. This allows us to take into account nonlinear changes in price. Multidimensional control variable matrix X i uses a dummy variable for the currency unit used in the auction, which allows us to assess the purchasing power of the trading country in terms of currency units.

[0056] Multidimensional control variable matrix X i is a dummy variable for the currency unit used in the auction and a dummy variable for the transaction year T[k] i We use a cross-term variable. This allows us to adjust the purchasing power of each country on an annual basis by using a cross-term. In other words, we can treat the difference between the price at the place of transaction (country) as the reference price (2000) and the price at the time of transaction (year of transaction) as the purchasing power of each country on an annual basis.

[0057] Multidimensional control variable matrix X i uses a quarterly dummy variable for the time of transaction. This allows us to take into account that auctions are held four times a year, and prices change depending on the time of sale. Specifically, this allows us to account for the tendency for the same work to be cheapest in the first quarter and more expensive in the third quarter. Multidimensional control variable matrix X i uses a dummy variable for the genre of the work, which allows us to respond to prices according to the genre of the work.

[0058] (6) In this embodiment, the control unit 21 of the management server 20 executes a price index calculation process (step S16). This allows the price index to be used to supplement data for a transaction year when there is no transaction record for that transaction year.

[0059] (7) In this embodiment, the control unit 21 of the management server 20 executes a price estimation process (step S22). This allows for a quantitative method to estimate the reproducible logarithmic price of an artwork. As a result, reliable price data can be used as a basis for making decisions when trading artworks. Furthermore, the present invention makes it possible to estimate the current price of an artwork from its past price. In other words, since the future price of an artwork can also be estimated, it becomes possible to consider paintings as appreciating assets (investment targets). For example, when holding artworks as assets alongside other financial products such as stocks and bonds, it is possible to compare the returns of artworks with those of other financial products. Furthermore, since artworks are not traded on financial markets and their prices are determined and fluctuate according to different principles, they may provide a complementary effect in the event of a sudden drop in the financial market.

[0060] This embodiment can be modified as follows: This embodiment and the following modifications can be combined and implemented within the scope of technical compatibility. In the above embodiment, the control unit 21 of the management server 20 performs a learning process. This process may be performed when the evaluation target information is acquired.

[0061] In the above embodiment, it is assumed that the price of a painting is calculated as an art object. However, the object of evaluation is not limited to a painting, and may be other art objects such as sculptures. In the above embodiment, a representative artist is defined as an artist who has had 40 or more auction transactions in the past 20 years. However, the artists identified as representative artists are not limited to this. [Explanation of symbols]

[0062] 10...auction site, 15...user terminal, 20...management server, 21...control unit, 211...acquisition unit, 212...preprocessing unit, 213...creation unit, 214...evaluation unit, 22...transaction record storage unit, 23...model information storage unit.

Claims

1. A price estimation system comprising: a transaction history storage unit that stores auction information including information on a work identifier, artist name, attributes, transaction price, transaction date, and transaction currency; and a control unit that calculates the price of an item to be evaluated, The control unit Estimating the size of the work from the attributes of the auction information; Identifying an original or a copy from attributes of the auction information; Identifying the genre of the work from the attributes of the auction information; Alpha of log-price linear regression model A j , β k , γ, δ jk , ζ j , η k , θ jk , b are estimated by the least squares method to calculate the theoretical price value calculation formula B. [Equation 1] P ijk : The winning bid price of work i by artist j in year k A[j] i : A dummy variable that is "1" if work i is by artist j, and "0" otherwise. T[k] i : A dummy variable that is "1" if the transaction year of work i is k, and "0" otherwise. O i : A dummy variable that uses the "original flag" to set "1" if work i is an original work, and "0" if not. X i : Other multidimensional control variable matrix for work i, which includes the following variables as dimensions: ・Artwork size [number] ・Variable obtained by squaring the work size [number] - Dummy variable for the currency unit used in the auction - Dummy variables for the currency unit used in the auction and the transaction year dummy variable T[k] i Cross-term variables of - Quarter dummy variable at the time of transaction (1st to 4th quarter) - Dummy variables for the genre of artwork (variables indicating the category of artwork, such as "contemporary art" or "post-war art") - Dummy variable for work genre α j , β k , γ, δ jk , ζ j , η k , θ jk , b: unknown parameters to be estimated u ijk : Error term. [Equation 2]

2. 2. The price estimation system according to claim 1, wherein the control unit performs name matching of artists using the auction information.

3. The control unit Using the auction information for each artist, identify the number of transactions during a predetermined period; 3. The price estimation system according to claim 1, wherein the logarithmic price linear regression model is calculated for each representative author whose number of transactions is equal to or greater than a reference number.

4. The price estimation system described in claim 3, characterized in that the control unit calculates the log-price linear regression model for a group of non-representative authors that includes non-representative authors whose number of transactions is less than a standard number.

5. 4. The price estimation system according to claim 3, wherein the control unit uses a calculation formula C to calculate a price index of original works and a price index of copy works for each representative artist. [Equation 3]

6. 5. The price estimation system according to claim 4, wherein the control unit uses a calculation formula D to calculate the price index of original works and the price index of copy works of the non-representative group of artists. [Equation 4]

Images