JP6584713B1 - Financial transaction style feature mapping apparatus and transaction style feature map generation method - Google Patents

Abstract

The present invention provides a method for generating a transaction style feature map that can map a transaction style feature of an individual user together with a transaction style to be aimed at. A transaction style feature map 20 in which transaction feature points specified by a transaction style feature value that can be specified by a person are held in a map space in which each transaction feature point can be referred to by the transaction style feature value; A simulated transaction artificial intelligence unit 15 connected to each transaction feature point of the transaction style feature map 20 and executing a trading action on specified market price information using the transaction style feature value as a constraint condition and outputting the trading result value; The transaction style mapping apparatus is characterized in that by designating an arbitrary transaction feature point of the transaction style feature map 20, the performance score at the transaction feature point is output. [Selection] Figure 2

Description

  The present invention relates to a method for generating a transaction style feature map capable of mapping features of a financial transaction style, an apparatus for matching a transaction style feature of a specific user on the transaction style feature map, and an investment using the transaction style feature map It is about the support system.

  With increasing performance and portability of computers and broadbanding of connected networks, so-called online trade, in which transactions such as stocks, exchange rates and commodity futures are performed using the Internet, has become widespread. According to online trade, since there is no cost for operating an actual store, there is an advantage that the fee is generally lower than buying and selling at a window or dealing by telephone.

  By the way, online trade users mainly use general publicly available knowledge such as information sources and reference books on the network to learn investment methods.

  However, most of the information published in the above reference books is theoretical, and even if it teaches so-called How to, it is only intended for an exemplary user in an ideal market. Absent.

  On the other hand, the actual transaction styles of users vary widely according to their experiences, personality, investment preferences, etc., and the investment goals that individual users should aim for should be different accordingly.

  The present invention has been made in view of such circumstances, and a method for generating a transaction style feature map capable of mapping individual user's transaction style characteristics together with a transaction style to be aimed at, and a specification on the transaction style feature map. It is an object of the present invention to provide a device that matches the transaction style characteristics of users.

  To achieve the above object, according to a first main aspect of the present invention, a transaction feature point specified by a transaction style feature value that can be specified by a person is referred to by each transaction feature point in the transaction style feature value. A trading style feature map held in a possible map space, and each trading feature point of the trading style feature map is connected to each trading feature point, and the trading style feature value is used as a constraint to execute trading behavior for specified market price information. It has an artificial intelligence trading mechanism that outputs trading result values, and by specifying any trading feature point of the trading style feature map, the performance value at the trading feature point is output. A transaction style mapping device is provided.

  According to one embodiment of the present invention, in the transaction style mapping apparatus, means for inputting a user's current transaction style feature point, means for inputting a user's target transaction style feature point, and the transaction style Means for outputting, as an optimum route, the shortest distance from the current transaction style feature point of the user to the target transaction style feature point on the feature map with the smallest drop in the score between the transaction feature points. A transaction style mapping apparatus characterized by comprising: Here, it is preferable that the transaction style mapping apparatus further includes means for outputting a guidance for guiding the user from the current transaction style feature point to the target transaction style feature point based on the optimum route. .

  Further, the guidance guidance means performs guidance guidance by displaying a message prepared in advance based on a difference in transaction style feature value between a current transaction style feature and a next trading style feature point. It is desirable that

  Further, the transaction style mapping device, based on the matching rate, means for calculating the matching rate by comparing the trading style feature value of the user and the trading style feature value of a specific trading feature point, Preferably, the user has means for determining a transaction feature point at which the user is currently located.

  Furthermore, according to another embodiment of the present invention, in the transaction style mapping apparatus, the means for inputting transaction information and market price information for all users and the input transaction information and market price information for all users. There is provided a transaction style mapping apparatus characterized by comprising a transaction style determining means for performing clustering analysis and defining each cluster classified based on the clustering analysis as a transaction style.

  Here, it is preferable that the transaction style mapping apparatus further includes a means for displaying the determined range of the transaction style on the transaction style feature map. The transaction style mapping apparatus preferably further includes means for determining the user's current transaction style based on the determined transaction style and the user's current transaction style characteristic value.

  Further, the transaction style mapping device includes a means for inputting a user's target transaction style, and a transaction style feature near the center of gravity of the target transaction style from the current transaction style feature point of the user on the transaction style feature map. Preferably, the shortest distance to the point has a means for outputting, as the optimum route, the one having the smallest drop in the score between the transaction characteristic points. In this case, it is preferable to further include means for outputting a guidance for guiding the user from the current transaction style feature point to the target transaction style feature point based on the optimum route. Further, in this case, the guidance guide means may display a message prepared in advance based on a difference in transaction style feature value between the current transaction style feature and the next transaction style feature point to be guided. desirable.

  According to another embodiment of the present invention, in the transaction style mapping apparatus of the present invention, the current transaction style feature value of the user is compared with the transaction style to which the current transaction style feature point of the user belongs. Thus, there is provided a transaction style mapping device comprising means for determining a fixing rate of the user to the transaction style and means for displaying the fixing rate to the user.

According to a second main aspect of the present invention, there is a method for generating a transaction style feature map comprising:
a. A computer mapping transaction feature points to a map space, each transaction feature point being identified in the map space by referring to a transaction style feature value that can be set by a person in the map space. The process,
b. A computer constructs a counterfeit artificial intelligence mechanism that imitates a person's buying and selling behavior for arbitrary market price information with the transaction style feature value as a constraint for each of the transaction feature points, and connecting to each of the transaction feature points;
c. A computer calculating a performance value at the transaction feature point by causing the counterfeit artificial intelligence mechanism connected to the transaction feature point to execute a pseudo-transaction; and
d. A computer generating a transaction style feature map by associating and storing each transaction feature point mapped in the predetermined map space with the result value in the transaction feature point;
The transaction style feature map is for outputting the score by designating an arbitrary transaction feature point in the map space. A transaction style feature map generation method is provided.

FIG. 1 is a schematic configuration diagram showing an embodiment of the present invention.

FIG. 2 is also a schematic system configuration diagram.

FIG. 3 is a schematic diagram showing a transaction style feature map.

FIG. 4 is a schematic diagram showing a system interface, similarly.

FIG. 5 is a schematic diagram showing the optimum route on the transaction style feature map.

FIG. 6 is a schematic diagram showing a system interface, similarly. FIG. 7 is a schematic diagram showing another product simulation result display interface, similarly.

  Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings.

1. Summary of the Invention In the present invention, the inventors analyzed the trading patterns of individual users using machine learning, and found that there was a certain transaction style for each user. Based on this knowledge, the current transaction style of the user is specified, and an investment support system that can lead to the transaction style that each user should aim for is completed.

  In other words, when individual users improve their investment skills, it is important how to set a goal (goal) for improvement. According to the conventional idea, as described above, the user tries to follow an ideal role model as described in a reference book or the like.

  On the other hand, the actual user's experience, personality, investment orientation, etc. vary widely, and the role model that each user should model should be different accordingly. Moreover, in an actual voluntary market, there are not always individuals suitable for the role model.

  Therefore, the present inventors define different transaction styles for each user by clustering the user's transaction orientation based on the past transaction information and market price information of all users, and define each transaction style that is a user's target. I was able to identify.

  In order to guide the user to the target transaction style, a human role model that executes different transaction styles is constructed with AI, so that the transaction style can be changed in a more reasonable route. A feature map was generated, and the optimum route was displayed on it.

  Also, in this system, the user's current and target transaction style adaptation rate and retention rate are regarded as one of the evaluation criteria, and appropriate feedback is provided to the user so that the user can steadily lead to the target transaction style. It is a thing.

  Details will be described below.

2. 1 and 2 are schematic configuration diagrams illustrating an embodiment of the present invention.

  In FIG. 1, 1 indicates an investment support system which is an embodiment of the present invention, and 2 indicates a user terminal device. The user terminal device 2 can be connected to the investment support system 1 via the Internet 3.

  FIG. 2 is a schematic configuration diagram showing the investment support system 1 in detail.

  The investment support system 1 includes a program storage unit 11 and a data storage unit 12 connected to a bus 10 to which a CPU 5, a RAM 6, an output unit / display unit 7 and an input unit 8 are connected.

  The program storage unit 11 includes a transaction style feature value identification unit 13, a transaction style feature map generation unit 14, a simulated transaction artificial intelligence unit 15, a transaction style definition unit 16, a user transaction style identification unit 17, an optimum route search unit 18, a market price The difficulty level specifying unit 19 is stored. The data storage unit 12 stores a transaction style feature map 20, transaction style feature value definition information 21, a simulated transaction artificial intelligence model 22, and a transaction information accumulation unit 23. These are actually computer software programs or data, which are read out by the CPU 5 and expanded on the RAM 6 to be appropriately executed or processed, thereby functioning as the respective components of the present invention. .

  Hereinafter, the detailed configuration and operation of each of the components 13 to 23 will be described through actual processing.

3. Definition of Transaction Style Feature Value In this system 1, first, the transaction style feature value specifying unit 13 defines a transaction style feature value for specifying a user's transaction style. In this embodiment, the following four items are used as items of transaction style feature values.
(1) Transaction frequency: Number of times of buying / selling within a unit period (2) Open interest holding period: Period from when a new order is placed to the corresponding counter-buying (settlement) (3) Ordering rate: Number of times / total number of trades (4) Number of positions: median number of positions (+1) just before new order The transaction style feature value specifying unit 13 sets the actual value for the above four items. It is determined by analyzing past transaction information 24 and market price information 25 of all users stored in the information accumulation unit 23.

  Specifically, in this embodiment, the transaction style feature value specifying unit 13 determines the median value of the distribution of the transaction style feature values of all users, and one or more transactions at regular intervals in a grid pattern around it. A list of feature value pairs is generated. Do not use each user's transaction style feature values directly, ie resampling, otherwise the number of underlying transaction feature values would be too many, requiring considerable computing resources, This is because the distribution of the target users can be affected.

  Next, the transaction style feature map generation unit 14 generates a transaction style feature map 20 as shown in FIG. Specifically, the transaction feature points of all users (points specified on the map by the feature value pairs) can be plotted on one map by two or more pairs of the above four feature values. The transaction style feature map 20 is generated so that the feature points on the map are uniquely specified as grid grid points (intersection points).

  In the example of the transaction style feature map 20 shown in FIG. 3, the transaction style feature value is a two-dimensional grid with the transaction frequency on the vertical axis and the number of positions on the horizontal axis. Grid points indicated by black circles on the map 20 are transaction feature points (= lattice points) specified by a set of transaction style feature values.

  Information on the transaction style feature map 20 generated in this way is stored in the data storage unit 12.

  The example of the transaction style feature map 20 shown in FIG. 3 is a two-dimensional one that specifies a transaction feature point by the transaction frequency and the number of open positions, but a four-dimensional one that uses all the transaction style feature values described above. Alternatively, a three-dimensional one using three may be used. Further, the transaction style feature map 20 itself may be a two-dimensional display or may actually identify each transaction feature point with four transaction style feature values.

Further, as a method for determining the above-mentioned order, in this embodiment, the last fixed number of candlesticks at the time of buying and selling passes through the maximum value of High and one point after, and is equal to or higher than all the remaining High values. The absolute value of the slopes of two straight lines that pass through one point after the minimum of the slope of the straight line and the lowest value of the slope of the straight line, and is equal to or less than all the remaining values. When both of the slopes of which are smaller than the threshold are defined as the rising market, the time when both are smaller than the threshold are defined as the falling market.
If it is bought at the rising market price or sold at the falling market price, it is judged as a forward order.

4). Construction of Pseudo Transaction Artificial Intelligence and Calculation of Results Score Next, the simulated transaction artificial intelligence unit 15 creates a simulated artificial intelligence model (AI model) 22 for each transaction feature point (grid point) of the transaction style feature map 20. The score is calculated at each point using the AI model 22.

Specifically, as shown in FIG. 2, the simulated transaction artificial intelligence unit 15 includes a transaction information statistical analysis unit 30, an artificial intelligence model generation unit 31, and a result calculation unit 32, and executes the following steps.
(1) The transaction information statistical analysis unit 30 takes a histogram for each transaction style feature value of each transaction feature point based on the accumulated transaction information 24 of all users,
(2) the transaction information statistical analysis unit 30 taking a histogram for feature values other than the transaction style feature value based on the accumulated transaction information of all users;
(3) Based on a histogram of transaction style feature values, the artificial intelligence model generation unit 31 sets a transaction style feature value at an arbitrary transaction feature point in the AI model 22 associated with each transaction feature point;
(4) the artificial intelligence model generating unit 31 initializing internal parameters of the AI model 22;
(5) The artificial intelligence unit 15 makes a judgment on buying and selling from internal parameters using the AI model 22 based on the past market information 25, and performs a pseudo-transaction,
(6) Based on the difference from the trading characteristic value set in (3) above and the trading profit / loss resulting from the trading result of the pseudo-transaction conducted in (5) above, Calculating the process,
(7) The artificial intelligence model generation unit 31 updates the internal parameters of the AI model 22 based on the reward value of the pseudo transaction,
(8) The artificial intelligence unit 15 calculates the pseudo value of the AI model 22 of (5) for the feature values of the above (1) and (2), and initializes internal parameters when the values are out of the histogram The process of
(9) The artificial intelligence model generation unit 31 performs a pseudo-transaction by the AI model 22 determining trading based on internal parameters based on the past market information 25,
(10) For the specific transaction style feature value set in the AI model 22, the score calculation unit 32 calculates the reward value obtained based on the pseudo-transaction performed in (9) as a score. , And storing the score in the data storage unit 12 in association with each transaction feature point of the transaction style feature map 20.

  According to such a process, the AI model 22 that conducts a transaction that cannot be performed by a human can be excluded in the process (8). As a result, it is possible to prevent the transaction feature point associated with the AI model from being used.

  In addition, although the reward score of (6) is used as it is for the score of (10), it is also possible to use only the trading profit / loss as the score. It is also possible to use only the reward value excluding trading profit / loss as the score.

5. Calculation of Market Difficulty In the system 1, in addition to the score, the market difficulty is stored in association with each transaction feature point of the transaction style feature map 20.

  This market difficulty level is a ratio in which the market difficulty level calculation unit 18 does not exceed a predetermined threshold value for the score generated by the score calculation unit 32 based on the market price information 25 of various financial transaction data ( Alternatively, the ratio excluded in the step 4 (8)) is calculated as the market difficulty level associated with the type of financial transaction data. Then, the market difficulty level for each type of financial transaction data calculated in this way is stored in association with each feature point of the transaction style feature map 20 in the same manner as the score.

  Through the above steps, the transaction style feature map 20 and the artificial intelligence model 22 connected to each transaction feature point of this map are completed, and are stored in the data storage unit 12 and can be used.

  Hereinafter, various processes using the transaction style feature map 20 and the artificial intelligence model 22 will be described.

6). Identification of transaction style of target point First, the transaction style definition unit 16 executes clustering based on the transaction information 24 and the market price information 25 of all users, and based on this, defines the “transaction style” that each user should aim for. To do.

  Here, the transaction style defines an investment tendency that characterizes a user's transaction. Specifically, the transaction style definition unit 16 accumulates past transaction information 24 of each user (all users). It is determined statistically and determined by performing cluster classification on the group structure that appears when they are aggregated for all customers.

Specifically, the following steps are executed.
(1) A process of calculating a transaction style feature value related to a transaction together with the market price information 25 for the accumulated transaction information 24 of all users (transaction style feature values include: Transaction rate, etc.).
(2) Clustering users based on the transaction style feature values extracted in (1) above (clustering using k-Means on the eigenspace of the eigenvectors of the variance-covariance matrix by principal component analysis) ).

The clusters thus determined are shown on the transaction style feature map 20 as clusters 1 and 2 indicated by ellipses in FIG.
(3) A step of calculating, for each cluster obtained in (2) above, the centroid (indicated by x in FIG. 3) of transaction feature points of all users belonging to the cluster as a representative point of the cluster.
(4) A step of attaching a label (for example, transaction style 1, transaction style 2, etc.) specifying a transaction style to each cluster obtained in (2) above.

  The above steps are executed at regular time intervals, updated as the transaction style feature map information 20 and stored in the data storage unit 12.

  In the above step (3), the center of gravity of the cluster (transaction style) is specified as a representative point. However, a specific transaction feature point is specified directly on the map, or an arbitrary transaction style feature value is specified. By doing so, the representative point may be specified.

7. Next, the user transaction style identification unit 17 identifies the current transaction style of the user based on the current transaction information of the user.

  The specification of this user transaction style starts from the starting point of the optimum route described later and is continuously executed until the user reaches the target point.

  Specifically, the user transaction style specifying unit 17 obtains the user's transaction style feature value based on the user's current and / or past transaction information 24 and the market price information 25, and to which cluster the user's transaction feature point is located. The transaction style labeled in the cluster is set as the user's current transaction style 43 and displayed on the user interface described later.

  For example, in FIG. 4, if the position indicated by ★ is the current transaction feature value of the user, the user's current transaction style is “transaction style 2”.

  Note that the transaction style of the user at the start point is not only determined based on the current transaction feature point of the user, but also a method in which the user directly specifies the transaction style of the start point or a method in which the transaction feature point is directly specified. It may be. Moreover, you may determine by designating arbitrary transaction characteristic values manually by a user.

8). User Interface FIG. 4 shows an example of the user interface of this system.

  In this system 1, as will be described later with reference to FIG. 5, based on the current transaction style of the user and the transaction style desired by the user, the optimum progress route there is displayed on the transaction style feature map. It is presented to the user as a road map.

  Therefore, as shown in FIG. 4, this interface includes a target style input unit 40, a target arrival rate display unit 42, a current transaction style detail display unit 43, a target transaction style detail display unit 44, and a target. The road map display unit 45 and a guidance message display unit 46 for the user.

  In the target style input unit 40, a transaction style that can be the target point specified in the above-described process is displayed so as to be selectable, for example, in a drop-down manner.

  Here, the details of the transaction style determined based on the current transaction feature point of the user described above are displayed on the current transaction style detail display unit 43. As described above, the user can also specify an arbitrary transaction feature point as the current transaction feature point. In this case, the user can also specify the feature using an interface as shown in FIG. In the interface of FIG. 6, four types of transaction style feature values can be set independently by moving the slider left and right.

  Note that the transaction style of the target point can also be set manually, and in this case, it can also be specified by the interface shown in FIG.

  These interfaces are actually generated as HTML documents by the investment support system 1, displayed on the user terminal device 2 in response to a request from the user terminal device 2 through the Internet 3, and input from the user is transmitted to the Internet. 3 is received by the system 1.

  The optimum route guidance through this interface will be described below.

9. Calculation of Optimal Route In this step, the optimal route search unit 18 is based on the current transaction style of the user specified above and the target transaction style input by the user in the target transaction style input unit 40 of the interface shown in FIG. Then, the optimum route for reaching it is calculated and displayed as a road map 45 on the user interface together with the map 20.

Specifically, the optimum route searching unit 18 includes a search start point specifying unit 50, a search end point specifying unit 51, and a guidance information generating unit 52, and executes the following steps.
(1) The search start point identifying unit 50 sets a transaction feature point determined by the current user's transaction style feature value as a search start point.

For example, in FIG. 5, if ★ is a point indicating the current transaction style feature value of the user, the grid point closest to ★ is determined as the transaction feature point for the start point 54.
(2) The search end point specifying unit 51 acquires a target transaction style specified by the user through the interface, and sets a transaction feature point close to the center of gravity of the target transaction style as a search end point.

For example, if the target transaction style designated by the user is “cluster 1” shown in FIG. 4, the grid point closest to the center of gravity (indicated by x) of the cluster is determined as the search end point 55.
(3) The route search unit 52 searches for the optimum route using the search start point 54, the search end point 55, and the score of the lattice points between them.

  In this embodiment, the optimum route search is performed by approximating the minimum decline in the maximum result of the transaction characteristic point through the shortest route (route candidate) between the start point 54 and the end point 55 that pass through the grid points. Is determined as the optimum route 56.

  Here, the shortest path candidate may be all shortest paths obtained in a permutation combination, but in this embodiment, in order to save computer resources, the next adjacent path in the direction from the start point 54 to the end point 55 is used. It is obtained by randomly determining the transaction style feature value, which is the moving direction of the grid point. Then, the performance score is extracted from each consecutive transaction feature point constituting the shortest route candidate determined at random in this way, and the difference in the performance score between adjacent transaction feature points is sequentially calculated, and the maximum is obtained for each route candidate. Calculate the decline in grades. Then, the route candidate that minimizes the maximum grade drop is determined as the optimum route.

  Thus, by randomly generating the shortest path, the number of shortest paths can be reduced, and the calculation cost for the search can be reduced. Of course, when the number of grid points is small, all the shortest paths from the start point 54 to the end point 55 can be searched.

  In addition, it is also possible to reduce the number of search routes by limiting the number of direction changes with respect to the randomly generated shortest route, and determine the optimum route from among them.

  In addition, as an evaluation method for the optimum route, a method in which the total value of the score of transaction characteristic points on the route is the maximum, or the highest score is obtained from the start point 54 without specifying the end point 55. It is also possible to use a path that leads to an adjacent grid point until the grade score does not increase.

10. Route guidance (display of guidance message)
When the optimum route is determined as described above, the route guidance information generation unit 53 generates information for guiding the user to the next grid point to be passed to the guidance message display unit 46 of the interface. indicate.

  Specifically, the route guidance information generation unit 53 has a style fixing rate that is an index indicating the fluctuation width of the user's current transaction style, and a transaction style feature of a transaction feature point that is the next grid point to be aimed at. A style matching rate that is an index indicating the degree of coincidence with the value is calculated, a predetermined message based on the calculated style fixing rate and style matching rate is displayed on the message display unit 46, and the transaction characteristic value of the user is displayed. Based on this, the current position is displayed on the map.

(1) Style fixing rate First, the style fixing rate is determined by the route guidance information generating unit 53 to the artificial transaction artificial intelligence model 22 corresponding to the current transaction style feature point of the user at the timing of the latest financial transaction of the user. Then, a transaction judgment is performed, and a coincidence rate with the user's transaction judgment is calculated by AUC (Area under the Curve), and the value is output as a style fixing rate.

  This style fixing rate is displayed on the transaction style feature map as indicated by 48 in the figure.

(2) Style Matching Rate Next, the style matching rate is calculated based on whether the route guidance information generating unit 53 is a pseudo-trading artificial intelligence corresponding to the current trading style feature point of the user and a trading style feature point that is an arbitrary lattice point. Let the corresponding artificial artificial intelligence perform financial transactions for a specific period, calculate the agreement rate of each transaction judgment at multiple time points sampled over time series by AUC (Area under the Curve), The value is output as a style match rate.

  Thereby, the user can know how much his / her trading style is blurred (or how stable).

(3) Generation of Guidance Message The guidance message is generated by the route guidance information generation unit 53 using a specific wording of the template document based on the value of the style fixing rate and the value of the style matching rate from the guidance text as a template. Is changed according to a certain rule, and a guidance document is generated.

  For example, in this embodiment, the following text is prepared as a template and stored in the data storage unit.

  “In order to grow to the goal,“ A ”and“ B ”are the points. This time, “C” was “D” compared to the previous time, approaching the target. Next, let's proceed further toward the goal by making a transaction with an awareness of "F" for "E". The improvement difficulty is “G”. According to the data of the past two weeks, trading with “H” or “I” is recommended.

On the other hand, since the consistency of transactions is low, it is also important to establish a style by setting “J” to “K”. ]
And the said guidance information production | generation part 53 produces | generates the guidance document which changes said AK dynamically with the following conditions and algorithms, and displays to a user.

(1) Setting of Calculated Transaction Style Feature Values X1 to X4 First, assume that the user's current, target, and past transaction styles are as follows.

X1, X2, X3, X4 = transaction frequency, open interest holding period, forward rate, open interest
-User's target transaction style "X1: Large, X2: Medium, X3: Small, X4: Large"
User's current transaction style “X1: Large, X2: Small, X3. Large, X4: Large”
・ User's past transaction style “X1: Medium, X2: Small, X3. Large, X4. Large”
In the above transaction style feature map display, two types of transaction style feature values are used, that is, two-dimensional. However, in the generation of a guidance document, four types of transaction style features are described as follows. You may perform in four dimensions using a value.

(2) Optimal route It is assumed that a part of the optimum route to be guided, which is obtained by the optimum route search unit 18 based on the optimum route calculation, is as follows.

    “X3 from large to medium => X2 from small to medium => X3, medium => small”.

(3) Market difficulty The market difficulty calculated by the market difficulty calculator 18 based on the market difficulty calculation for the grid point (transaction feature point) closest to the current user spot transaction style is It is assumed that the second lowest type of financial transaction data is as follows.

・ Type of financial transaction data with the lowest market difficulty: “USD / JPY”
・ The second type of financial transaction data with the lowest market difficulty: “AUD / JPY”.

(4) Style fixing rate and style adaptation rate Assume that the style fixing rate and the style adaptation rate are as follows.

-Style conformity rate: 80%
-Style fixing rate: 30% (assuming lower than average of users).

(5) Template embedding algorithm Assume that the template embedding algorithm is determined as follows.

  Rule 1. “A” and “B” are input with the first and second items from the most recent in the optimum route.

  Rule 2. For “C” and “D”, the item having the largest absolute value is input as the difference from the past style to the target style.

  Rule 3. In “E” and “F”, the latest item in the optimum route is input.

  In Rule 4. “G”, the style matching rate up to the next grid point that should be aimed at is entered.

  In rule 5. “H” and “I”, the type of financial transaction data with the lowest market difficulty is input first and second.

  Rule 6. If consistency is lower than average, enter “J” and “K” for the most diversified transaction feature of the customer.

  As a result, the following guidance document is generated.

  “In order to grow to the target, the period of holding open interest and the method of new entry are the points. This time, the frequency of transactions per day increased from 3 to 5 times, and it was close to the target. Next, let's move forward toward the goal by making a deal with the aim of upside down at the time of entry. Improvement difficulty is 80%. According to the data of the past two weeks, trading in "USD / JPY" or "AUD / JPY" is recommended.

On the other hand, since the consistency of transactions is low, it is also important to establish a style with an awareness of keeping the open interest period constant. ]
The guidance document generated by the above steps is displayed on the guidance message display section 46 of the interface for each user.

12 FIG. 7 shows a user interface 71 that displays a simulation result of another product obtained by the other product simulation unit 26. FIG.

  In this example, according to the aforementioned guidance document, “USD / JPY” or “AUD / JPY” is recommended according to the data of the past two weeks. As shown in FIG. 7, the interface 21 shown in FIG. 7 is further detailed in terms of calculating and comparing the market difficulty level for currency pairs (different financial products) that are different from the current passing pair. Information is provided (however, in the example of FIG. 7, the currency pair currently traded is “USD / JPY” and recommends trading in “EUR / JP”).

  That is, the other product simulation unit 26 uses the simulated artificial intelligence model (AI model) 22 to transfer the other items at each point to the simulated transaction artificial intelligence unit 15 connected to each feature point belonging to the user's current transaction style. The score value when using a passing pair is calculated.

  Here, even when a simulation is performed for other passing pairs, the AI model 22 itself may be used in common with the one already created above. It means only arithmetic.

  Then, the other product simulation unit 26 applies the score value of each point to a predetermined algorithm and selects the one with the highest profit or the one with the least variance of the results as a recommended currency pair, and the comparison result is the interface 71. Display above. The comparison algorithm is not limited to the profit value and the profit distribution, and may be a combination of other factors. The products to be compared are not limited to FX products, and completely different investment products may be displayed.

  In this embodiment, the other product simulation unit 26 as a comparison result of the current passing pair “USD / JPY” and the recommended currency pair “EUR / JP”, in addition to the detailed display 72, the market difficulty-revenue rate map 73 , The revenue distribution graph 74 is displayed so as to be understood at a glance.

  By simulating with other products in this way, it is possible to recommend a product that allows the user to increase profits without changing the current transaction style.

13. Effect According to the configuration as described above, by clustering the user's transaction orientation based on the past transaction information and market price information of all users, different transaction styles are defined depending on the user, and each user's target The transaction style can be specified.

  In order to guide the user to the target transaction style, a human role model that executes different transaction styles is constructed with AI, so that the transaction style can be changed in a more reasonable route. A feature map was generated, and the optimum route was displayed on it.

  In this system, the user's current and target transaction style conformity rate and retention rate is one of the evaluation criteria, and it is fed back to the user as appropriate, thereby steadily leading to the target transaction style. I was able to do it.

  According to the configuration as described above, a method for generating a transaction style feature map that can map the characteristics of each user's transaction style together with the transaction style to be aimed at, and matching the transaction style features of a specific user on the transaction style feature map The apparatus which performs can be provided.

  Needless to say, the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the scope of the invention.

DESCRIPTION OF SYMBOLS 1 ... Investment support system 2 ... User terminal device 3 ... Internet 5 ... CPU
6 ... RAM
DESCRIPTION OF SYMBOLS 7 ... Display part 8 ... Input part 10 ... Bus 11 ... Program storage part 12 ... Data storage part 13 ... Transaction style characteristic value specific | specification part 13-23 ... Configuration part 14 ... Transaction style characteristic map generation part 15 ... Simulated transaction artificial intelligence part DESCRIPTION OF SYMBOLS 16 ... Transaction style definition part 17 ... User transaction style specific part 18 ... Optimal route search part 19 ... Market price difficulty specific part 20 ... Transaction style feature map 21 ... Definition information 22 ... Simulated transaction artificial intelligence model 23 ... Transaction information accumulation part 24 ... Transaction information 25 ... Market price information 30 ... Transaction information statistical analysis unit 31 ... Artificial intelligence model generation unit 32 ... Results calculation unit 40 ... Target transaction style input unit 42 ... Reach rate display unit 43 ... Current transaction style detail display unit 44 ... Target Transaction style detail display section 45 ... Road map display section 46 ... Guidance message display section 50 ... Search start point identification section 51 ... Search end point specifying unit 52 ... guide information generator 53 ... route guidance information generation unit 54 ... search starting point 55 ... search end point 56 ... optimum route

Claims (28)

A transaction style mapping device for mapping a user's transaction style in a financial transaction,
A transaction style feature map in which transaction feature points specified by a transaction style feature value that can be specified by a person are held in a map space in which each transaction feature point can be referred to by the transaction style feature value;
An artificial intelligence trading mechanism that is connected to each trading feature point of the trading style feature map, executes trading behavior for specified market price information with the trading style feature value as a constraint, and outputs the trading result value;
The transaction style mapping device characterized in that by specifying an arbitrary transaction feature point in the transaction style feature map, the trading result value at the transaction feature point is output. The transaction style mapping device according to claim 1,
further,
Means for entering the user's current transaction style features;
Means for inputting the target transaction style feature point of the user;
Among the shortest distances from the current transaction style feature point of the user to the target transaction style feature point on the transaction style feature map, an optimum route having the smallest decline in the trading performance value between the transaction feature points is selected. As a means to output as
Transaction style mapping apparatus characterized by having The transaction style mapping device according to claim 2,
further,
A transaction style mapping apparatus comprising: means for outputting a guidance for guiding the user from a current transaction style feature point to a target transaction style feature point based on the optimum route. The transaction style mapping device according to claim 3,
The means for outputting the guidance guidance performs guidance guidance by displaying a message prepared in advance based on a difference in transaction style feature value between a current trading style feature and a next trading style feature point. Transaction style mapping device characterized by being a thing. The transaction style mapping device according to claim 3,
further,
Means for calculating the relevance ratio by comparing the transaction style characteristic value of the user with a transaction style characteristic value of a specific transaction characteristic point; and determining the transaction characteristic point at which the user is currently located based on the relevance ratio And a transaction style mapping device. The transaction style mapping device according to claim 1,
A means to input transaction information and market price information for all users,
A transaction style determining means for performing clustering analysis based on the input transaction information and market price information of all users and defining each cluster classified based on the analysis as a transaction style Mapping device. The transaction style mapping device according to claim 6, wherein
further,
A transaction style mapping apparatus comprising: means for displaying the determined range of transaction styles on the transaction style feature map. The transaction style mapping device according to claim 6, wherein
further,
A transaction style mapping apparatus comprising means for determining a user's current transaction style based on the determined transaction style and the user's current transaction style characteristic value. The transaction style mapping device according to claim 7, wherein
A means of entering the user's target transaction style;
Of the shortest distance from the current transaction style feature point of the user to the transaction style feature point in the vicinity of the center of gravity of the target transaction style on the transaction style feature map, the decrease in the trading performance value between the transaction feature points Means for outputting the one having the smallest value as the optimum route,
Transaction style mapping apparatus characterized by having The transaction style mapping device according to claim 9, wherein
further,
A transaction style mapping apparatus comprising: means for outputting a guidance for guiding the user from a current transaction style feature point to a target transaction style feature point based on the optimum route. The transaction style mapping device according to claim 10,
The means for outputting the guidance guidance performs guidance guidance by displaying a message prepared in advance based on a difference in transaction style feature value between a current trading style feature and a next trading style feature point. Transaction style mapping device characterized by being a thing. The transaction style mapping device according to claim 6, wherein
further,
Means for determining a retention rate of the user in the transaction style by comparing the user's current transaction style feature value with the transaction style to which the current transaction style feature point of the user belongs;
A transaction style mapping apparatus comprising: means for displaying the fixing rate to a user. The transaction style mapping device according to claim 8,
For each feature point in a current transaction style user determined above, the user obtains the trading results values of different other financial instruments and financial instruments to identify the transaction style, the trading performance value of the other financial products A transaction style mapping device, comprising: another product simulation means for displaying. The transaction style mapping device according to claim 13,
The other product simulation means compares the sales performance values of a plurality of other financial products, determines other financial products recommended to the user based on the comparison result, and displays the sales performance values of the other financial products. A transaction style mapping device characterized by that. A method for generating a transaction style feature map, comprising:
a. A computer mapping transaction feature points to a map space, each transaction feature point being identified in the map space by referring to a transaction style feature value that can be set by a person in the map space. The process,
b. A computer constructs a counterfeit artificial intelligence mechanism that imitates a person's buying and selling behavior for arbitrary market price information with the transaction style feature value as a constraint for each of the transaction feature points, and connecting to each of the transaction feature points;
c. A computer calculating a trading performance value at the transaction feature point by causing the artificial artificial intelligence mechanism connected to the transaction feature point to execute a pseudo-transaction; and
d. Computer, before Kemah Tsu each transaction feature points are mapped to up space, which has a step of generating a transaction style feature maps by storing in association with the trading performance value at the transaction point feature,
The transaction style feature map is for outputting the trading result value by designating an arbitrary transaction feature point in the map space. The transaction style feature map generation method according to claim 15,
further,
Entering the user's current transaction style features;
Inputting the target transaction style feature point of the user;
Among the shortest distances from the current transaction style feature point of the user to the target transaction style feature point on the transaction style feature map, an optimum route having the smallest decline in the trading performance value between the transaction feature points is selected. And a process of outputting as
A transaction style feature map generation method characterized by comprising: The transaction style feature map generation method according to claim 16,
further,
A method for generating a transaction style feature map, comprising a step of outputting a guidance for guiding the user from a current transaction style feature point to a target transaction style feature point based on the optimum route. The transaction style feature map generation method according to claim 17,
Trading the induction guide, characterized in that it is carried out by displaying a message prepared in advance on the basis of the difference between the transaction style characteristic values between the current trading style features and trading style feature points that induce the following Style feature map generation method . The transaction style feature map generation method according to claim 17,
further,
A step of calculating the relevance ratio by comparing the transaction style characteristic value of the user with a transaction style characteristic value of a specific transaction characteristic point, and determining the transaction characteristic point where the user is currently located based on the relevance ratio And a transaction style feature map generation method . The transaction style mapping device according to claim 15,
The process of inputting transaction information and market price information for all users,
A transaction style determination step of performing a clustering analysis based on the input transaction information and market price information of all the users and defining each cluster classified based on the analysis as a transaction style. Feature map generation method . The transaction style feature map generation method according to claim 20,
further,
A method of generating a transaction style feature map , comprising: displaying the determined range of the transaction style on the transaction style feature map . The transaction style feature map generation method according to claim 20,
further,
A transaction style feature map generation method comprising: determining a user's current transaction style based on the determined transaction style and the user's current transaction style feature value. The transaction style feature map generation method according to claim 21,
Entering the user's target transaction style;
Of the shortest distance from the current transaction style feature point of the user to the transaction style feature point in the vicinity of the center of gravity of the target transaction style on the transaction style feature map, the decrease in the trading performance value between the transaction feature points Outputting the one having the smallest value as the optimum route;
A transaction style feature map generation method characterized by comprising: The transaction style feature map generation method according to claim 23,
further,
A method for generating a transaction style feature map, comprising a step of outputting a guidance for guiding the user from a current transaction style feature point to a target transaction style feature point based on the optimum route. The transaction style feature map generation method according to claim 24,
Trading the induction guide, characterized in that it is carried out by displaying a message prepared in advance on the basis of the difference between the transaction style characteristic values between the current trading style features and trading style feature points that induce the following Style feature map generation method . The transaction style feature map generation method according to claim 20,
further,
Determining the retention rate of the user in the transaction style by comparing the user's current transaction style feature value with the transaction style to which the current transaction style feature point of the user belongs;
A transaction style feature map generation method comprising: displaying the fixing rate to a user. The transaction style feature map generation method according to claim 22 ,
For each feature point in a current transaction style user determined above, the user obtains the trading results values of different other financial instruments and financial instruments to identify the transaction style, the trading performance value of the other financial products A method for generating a transaction style feature map, comprising a step of simulating another product to be displayed. The transaction style feature map generation method according to claim 27 ,
The other product simulation step compares the sales performance values of a plurality of other financial products, determines other financial products recommended to the user based on the comparison result, and displays the sales performance values of the other financial products. A transaction style feature map generation method characterized by:

Images