Betting ticket purchase support system, betting ticket purchase support method, and betting ticket purchase support program

Abstract

To provide a voting ticket purchase support system that inputs various indicators disclosed for a race, and presents voting tickets that a purchaser may desire with enhanced win / loss prediction precision to prompt the user to make a purchase.SOLUTION: A voting ticket purchase support system comprises: a race basic data acquisition part which acquires race basic data including past statistics information, odds information and race result information of race targets of a race as an object to be voted for; a reference order prediction data generation part which estimates ranking between a plurality of race objects to participate in a race and generates it as reference prediction order data; a before-race place prediction data generation part which estimate a ranking based upon the odds information for each race target and generates before-race prediction place data; a before-race predicted odds generation part which estimates before-race odds and generates before-race predicted odds data; and a probability prediction part which predicts a wind probability of a voting ticket set for a race target when the voting ticket is purchased, and generates probability prediction result data.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a betting ticket purchase support system, a betting ticket purchase support method, and a betting ticket purchase support program, and in particular to a betting ticket purchase support system, method, and program that finds advantageous conditions for purchasing betting tickets in a competition. [Background technology]

[0002] For example, in publicly managed sports events such as horse racing, bicycle racing, and boat racing, the ratio of the payout to the amount bet is called odds. For example, individual odds are displayed on betting tickets such as horse racing tickets before the event (race) takes place. Purchasers of betting tickets for sports get interested in hoping for a large payout for their bet or high odds (multiples).

[0003] In this case, the organizer of the race makes public a number of indicators, such as the odds for betting tickets, the popularity of the racehorses in the case of horse racing, the number of people who purchase betting tickets, etc. Furthermore, various indicators fluctuate even before the race begins. For this reason, it has not been easy for purchasers to predict the development of the race before the start and purchase the optimal betting ticket that meets their wishes.

[0004] Therefore, devices and systems have been proposed to assist purchasers of ballot tickets in making selections and purchasing ballot tickets (see Patent Documents 1 and 2, etc.). These devices, systems, etc. are designed to perform certain functions as a guide for purchasers in purchasing ballot tickets, etc.

[0005] However, the technologies represented by patent documents, etc., are limited in the types of indicators (data) they can handle, and therefore the accuracy of presenting betting tickets derived from indicators, data, etc. related to the competition is insufficient. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Laid-Open No. 2004-199479 [Patent Document 2] Japanese Patent Application Publication No. 2019-120977 Summary of the Invention [Problem to be solved by the invention]

[0007] The present invention has been made in consideration of the above points, and provides a betting ticket purchase support system, method, and program that incorporates multiple types of indicators disclosed in a competition to improve the accuracy of predicting the outcome of the competition, and presents betting tickets that can satisfy the purchaser's wishes and encourages them to purchase them. [Means for solving the problem]

[0008] In other words, the betting ticket purchase support system of the embodiment is characterized by comprising: a competition basic data acquisition unit that acquires basic competition data including statistical information on the competition objects, odds information for the competition objects, and competition result information for the competition objects for competitions that the competition object has previously conducted in the competition that the competition object is to be bet on, a reference ranking prediction data generation unit that estimates the rankings of multiple competition objects participating in the competition based on the competition basic data and generates this as reference predicted ranking data, a pre-competition ranking prediction data generation unit that estimates the pre-competition rankings of the competition objects in the competition that the competition object will soon conduct based on the reference predicted ranking data and odds information for each competition object in the competition that the competition object will soon conduct, and generates this as pre-competition predicted ranking data, a pre-competition prediction odds generation unit that estimates pre-competition odds for the competition objects in the competition that the competition object will soon conduct based on the competition basic data and generates this as pre-competition predicted odds data, and a probability prediction unit that predicts the probability of winning when a betting ticket set for the competition object in the competition that the competition object will soon conduct based on the pre-competition predicted ranking data and pre-competition predicted odds data, and generates probability predicted result data for the competition that the competition object will soon conduct.

[0009] Furthermore, the system may be provided with a betting ticket purchasing unit that generates betting ticket purchase data for purchasing betting tickets that have been decided to be purchased based on the probability prediction result data from the organizer of the competition in which the competition item that is the subject of the betting is held.

[0010] Furthermore, the competition basic data acquisition unit may acquire the competition basic data from the organizer of the competition in which the sport that is the subject of voting is held.

[0011] Furthermore, the basic competition data acquisition unit may include a statistical data generation unit that generates statistical data for each competition object from the statistical information of the competition object, odds information for the competition object, and competition result information for the competition object.

[0012] Furthermore, the basic competition data acquisition unit may include a feature amount generation unit that generates a competition object feature amount for each competition object in a competition that the competition object will be participating in.

[0013] Furthermore, the standard ranking prediction data generating unit, when estimating the rankings of a plurality of events participating in a competition based on the competition basic data, prioritizes the most popular event by taking into account odds information for the events. standard Generate predicted ranking data or exclude the most popular competitors standard Predicted ranking data may be generated.

[0014] Furthermore, the probability prediction unit may generate probability prediction result data that also includes winning feature amounts related to betting tickets for sporting items that have been winning in sporting events that have been held in the past.

[0015] Furthermore, machine learning may be used to generate reference predicted ranking data in the reference ranking prediction data generation unit, generate pre-competition predicted ranking data in the pre-competition ranking prediction data generation unit, generate pre-competition predicted odds data in the pre-competition prediction odds generation unit, and generate probability prediction result data in the probability prediction unit. [Effects of the Invention]

[0016] The betting ticket purchase support system of the present invention includes a competition basic data acquisition unit that acquires basic competition data including statistical information on the competition object, odds information on the competition object, and competition result information on the competition object for competitions that the competition object has previously participated in, in the competition that the competition object is to be bet on; a reference ranking prediction data generation unit that estimates the rankings of multiple competition objects participating in the competition based on the competition basic data and generates this as reference predicted ranking data; and a pre-competition ranking prediction data generation unit that estimates the pre-competition rankings of the competition objects based on the reference predicted ranking data and odds information for each competition object in the competition that the competition object will participate in, and generates this as pre-competition predicted ranking data. a pre-race prediction odds generation unit that estimates pre-race odds for a sporting object for an upcoming race based on the race basic data and generates pre-race prediction odds data, and a probability prediction unit that predicts the probability of a winning bet when a betting ticket set for a sporting object for an upcoming race based on the pre-race prediction ranking data and the pre-race prediction odds data is purchased, and generates probability prediction result data for the upcoming race of the sporting object, so that by incorporating a plurality of various indicators disclosed in the race, it is possible to improve the accuracy of predictions of the outcome of the sporting object in the race, and present betting tickets that satisfy the purchaser's wishes and encourage them to purchase them. Similar effects can also be obtained in betting ticket purchase support methods and programs. [Brief explanation of the drawings]

[0017] [Figure 1] 1 is a schematic diagram showing the configuration of a betting ticket purchase support system according to an embodiment. [Figure 2] 2 is a block diagram showing the configuration of functional units in a computer of the betting ticket purchase support system. FIG. [Figure 3] 1 is a schematic diagram showing the processing flow of a betting ticket purchase support system. [Figure 4] FIG. 10 is a flowchart showing the processing flow of the competition basic data acquisition unit. [Figure 5] FIG. 10 is a flowchart showing the flow of processing by a reference rank prediction data generating unit. [Figure 6]FIG. 10 is a flowchart showing the flow of processing by a pre-race ranking prediction data generating unit. [Figure 7] FIG. 10 is a flowchart showing the flow of processing by a pre-race predicted odds generation unit. [Figure 8] FIG. 10 is a flowchart showing the flow of processing by a probability prediction unit. [Figure 9] 10 is a block diagram showing the functional configuration of a betting ticket purchasing unit. FIG. [Figure 10] 10 is a flowchart illustrating a method for supporting the purchase of a betting ticket according to an embodiment. [Figure 11] 10 is a first graph showing the calculation results of the betting ticket purchase support system of the embodiment. [Figure 12] 10 is a second graph showing the calculation results of the betting ticket purchase support system of the embodiment. [Figure 13] 10 is a third graph showing the calculation results of the betting ticket purchase support system of the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0018] The betting ticket purchase support system of the embodiment is a system that supports purchasers in determining which betting tickets to purchase for which events and for which betting content in a competition in which the event being bet on is being held. The system also enables purchasers to purchase desired betting tickets from the event organizer. Because this betting ticket purchase support system includes a wide range of information accumulated in the past about the events being bet on, it lowers the barrier to entry for purchasing betting tickets, even for purchasers who are unfamiliar with the content of the competition, individual information about the events, and even with selecting betting tickets.

[0019] The term "competition" as used here refers to a competition in which multiple events compete for ranking, such as publicly managed competitions. Specific examples include "horse racing," such as central horse racing organized by the Japan Racing Association and local horse racing organized by the National Association of Racing Associations, as well as "boat racing" (motorboat racing) organized by local public organizations, and "keirin" or "auto racing" organized by local public organizations. Furthermore, to the extent permitted by legal systems, sports lotteries such as soccer are also included. Naturally, the term is not limited to the types of competition listed.

[0020] Furthermore, the term "game object" refers to the entity that is competing in the game. When the type of game is "horse racing," the game objects are horses and jockeys. Similarly, when the game is "boat racing," the game objects are boat racers, when the game is "cycle racing," the game objects are keirin racers, and when the game is "auto racing," the game objects are auto racers. Furthermore, in the case of sports lotteries such as soccer, the game objects are teams. In the explanation of the betting ticket purchase support system, method, and program of the embodiment, the "game" will be explained as "horse racing," the "game object" as "horses," and the "betting ticket" as "betting ticket." Furthermore, "odds" refers to the payout ratio for bets on publicly managed games, etc. Of course, changes to the various games and games objects mentioned above are naturally permitted.

[0021] 1 is a schematic diagram showing the configuration of a betting ticket purchase support system 1 according to an embodiment. In the betting ticket purchase support system 1, a venue 5 (a racecourse or track) where a race (horse race) is held, a computer 3 (server) of a race (horse race) organizer 2, and a computer 10 (server) of a betting ticket purchase support system operator 6 are connected via an internet line 4. Various information such as the location (racecourse) of the race to be held, the content (which race), the participating events (horses), and the results of the race are communicated between the organizer 2 and the venue 5, and the betting ticket purchase support system operator 6 obtains information made public by the organizer 2 from the organizer 2's computer 3 via the internet line 4.

[0022] The computer 3 of the organizer 2 and the computer 10 of the operator 6 of the betting ticket purchase support system are electronic computers (computing resources) such as well-known mainframes, workstations, cloud computing systems, etc. Note that the computer 10 also includes electronic computers (computing resources) such as personal computers, smartphones, tablet terminals, etc.

[0023] 2 is a block diagram showing the configuration of the functional units in computer 10 of operator 6 of the betting ticket purchase support system. Explaining computer 10 in detail, it is equipped with a CPU 11, ROM 12, RAM 13, storage unit 14, I / O 15 (input / output interface), etc. Naturally, computer 3 of organizer 2 has a similar configuration.

[0024] The functional units in the CPU 11 of the computer 10 are shown in the block diagram of Figure 2. The functional units include a competition basic data acquisition unit 110, a reference ranking prediction data generation unit 120, a pre-competition ranking prediction data generation unit 130, a pre-competition prediction odds generation unit 140, a probability prediction unit 150, and a betting ticket purchase unit 160. The operation and execution of the computer 10 is realized in software terms by a betting ticket purchase support program loaded into the main memory, etc.

[0025] When each functional unit of computer 10 in Figure 2 is implemented by software, computer 10 is implemented by executing instructions of a program, which is software that implements each function. The recording medium that stores this program can be a "non-transitory tangible medium," such as a CD, DVD, semiconductor memory, or programmable logic circuit. In addition, this program may be supplied to computer 10 of operator 6 of the betting ticket purchase support system via any transmission medium (communications network, broadcast waves, etc.) that can transmit the program.

[0026] The memory unit 14 of the computer 10 is a known storage device such as an HDD or SSD. The memory unit 14 may also be an external server (not shown). The memory unit 14 stores various data, information, a betting ticket purchase support program, various data necessary for executing the program, and the like. Furthermore, each functional unit that performs various calculations, computations, and other operations is an arithmetic element such as a CPU 11. In addition, input devices such as a keyboard and a mouse (not shown), a display unit (display device such as a monitor), an output device that outputs data, and the like may also be appropriately connected to the I / O 15 of the computer 10.

[0027] FIG. 3 is a schematic diagram showing the processing flow in each functional unit in the betting ticket purchase support system 1 of the embodiment. First, the competition basic data acquisition unit 110 (data generation) acquires and generates competition basic data required for subsequent processing. Next, the reference ranking prediction data generation unit 120 (fundamentals prediction) generates reference predicted ranking data that serves as a reference. Subsequently, the pre-race ranking prediction data generation unit 130 (technical prediction) generates pre-race predicted ranking data for the relevant competition. Furthermore, the pre-race predicted odds generation unit 140 (final odds estimation) generates competition predicted odds data. Then, the probability prediction unit 150 (betting ticket probability estimation) generates betting ticket purchase data. Based on the series of calculation results, the betting ticket purchase unit 160 (betting ticket purchase determination) generates specific betting ticket purchase information.

[0028] Thereafter, the planned purchase betting ticket data (information on planned betting ticket purchases) is accumulated in the planned purchase betting ticket database 513, and in processing batch 514, the desired betting tickets (betting tickets) are purchased from the race organizer via the Internet line 4. Each of these will be explained individually below.

[0029] The basic competition data acquisition unit 110 acquires basic competition data including statistical information on the object (horse), odds information for the object (horse), and competition results information for the object (horse) from past races (horse races) in which the object (horse) being bet on participates. Specifically, this information includes the weight of the object (horse) as a record of past races, wins and losses in the race, the horse's actual odds, information on betting ticket payouts, and other information about the racecourse where the race is held (length of run). Furthermore, the condition of the racecourse where the race is held (track), the characteristics and pedigree of the object (horse), etc. may also be added to the basic competition data.

[0030] FIG. 4 is a flow diagram showing the processing flow of the competition basic data acquisition unit 110 (data generation). The data acquisition unit 111 can periodically acquire information from the organizer of the competition that is the subject of betting, such as the Japan Racing Association's "JRA-VAN" or "JRDB." Information acquired includes the type of race and the horses that will be running, and is acquired as needed. Odds and indicators of horse popularity are acquired 10 minutes, 5 minutes, and at the time of the deadline for purchasing betting tickets for the competition (race).

[0031] The acquired basic race data is registered and updated in various databases (DBs) such as a race information racehorse database 501 and an odds popularity database 502.

[0032] The information acquired by the data acquisition unit 111 is transferred to the statistical data generation unit 112 (index statistical data generation). The statistical data generation unit 112 generates statistical data for each game object from the statistical information of the game object, odds information for the game object, and game result information for the game object.

[0033] The statistical data generation unit 112 (index statistical data generation) in Figure 4 generates various evaluation indices that are thought to affect the development of a race in a competition (horse racing). For example, the running ability (running speed), weight, body length, and race development in past races (leading style, catching up style, etc.) of the horse, which is the subject of the race. These are calculated as an average, standard deviation, etc. based on the results of past races. In addition, time information such as race time, winning margin time, leading 3F time, last 3F time, last 4F time from past races is also acquired, and from these, an average, standard deviation, etc. is calculated as an evaluation index of time.

[0034] The various statistical data generated by the statistical data generation unit 112 is registered and updated in various databases (DBs) such as the statistical index database 503 and the statistical database 504. The statistical data registered and updated in these databases (DBs) is basic competition data.

[0035] The various statistical data generated by the statistical data generation unit 112 is transferred to the feature generation unit 113 (feature data generation). The feature generation unit 113 generates competition object features for each competition object (horse) in a competition (horse race) that the competition object will soon participate in.

[0036] In the feature amount generation unit 113 (feature amount data generation) of FIG. 4, various statistical data are prepared into data that emphasizes features in order to be supplied to the reference ranking prediction data generation unit 120 (fundamental prediction) and the pre-race ranking prediction data generation unit 130 (technical prediction), which will be described below. For the statistical data supplied to the reference ranking prediction data generation unit 120 (fundamental prediction), feature amounts (race object feature amounts) are prepared from statistical data that does not change after the horse's weight is announced. For the statistical data supplied to the pre-race ranking prediction data generation unit 130 (technical prediction), feature amounts (race object feature amounts) are prepared from statistical data that mainly changes until the deadline for betting tickets (betting tickets) for the race. When averaging statistical data, for example, taking win rate as an example, the weighting rate can be increased for the rankings of the most recent 10 races.

[0037] The various features (game object features) generated by the feature generation unit 113 are registered and updated in various databases (DBs) such as a fundamentals prediction feature database 505 and a technical prediction feature database 506. The features (game object features) registered and updated in these databases (DBs) are basic game data.

[0038] The reference ranking prediction data generation unit 120 (fundamentals prediction) estimates the rankings of multiple sports objects (horses) participating in a competition (race) based on the competition basic data generated by the competition basic data acquisition unit 110 (data generation unit) and generates this as reference predicted ranking data. In particular, the reference ranking prediction data generation unit 120 is a basic ranking prediction that does not depend on popularity, odds, etc., and uses gradient boosting rank learning. The reference ranking prediction data generation unit 120 estimates the rankings of multiple sports objects (horses) participating in a competition based on the competition basic data.

[0039] FIG. 5 is a flow diagram showing the processing flow of the standard ranking prediction data generation unit 120 (fundamentals prediction). As shown in FIG. 5, the competition basic data generated by the competition basic data acquisition unit 110 (data generation unit) is stored in a race information racehorse database 501, a statistical index database 503, a statistical database 504, and a fundamentals prediction feature database 505. In the embodiment, the standard ranking prediction data generation unit 120 constructs statistical data for each individual racecourse, including JRA, local racecourses (Monbetsu, Kanazawa, Nagoya, Saga, Kochi), Hyogo Racecourse (Sonoda, Himeji), Southern Kanto Racecourse (Funabashi, Kawasaki, Urawa, Oi), Iwate Racecourse (Morioka, Mizusawa), and Obihiro Racecourse. The examples described below will show the results of calculations related to JRA.

[0040] The basic competition data stored in each database is incorporated into two types of models depending on the level of the return rate, and calculations are performed. The first is a return rate-focused model unit 122, and the second is a hit rate-focused model unit 123. Machine learning, which will be described later, is applied when performing calculations for both models. When performing calculations for both models, calculations are performed from various data stored in the race information racehorse database 501, statistical index database 503, statistical database 504, and fundamentals prediction feature database 505. In both models, the same data may be used, with only the return rate changed to a predetermined value (multiplied by a coefficient), or the type of original data may be changed for calculations.

[0041] The payout rate emphasis model unit 122 performs a calculation process to raise the rank of betting tickets (betting tickets) that are less popular so that the payout rate from betting is higher compared to the amount invested (high return). The payout rate emphasis model unit 122 is a so-called longshot model. Specifically, since the odds for the most popular event (horse) are low even though it is based on basic competition data, the most popular event (horse) is deliberately excluded and the second most popular event (horse) and even less popular events (horses) that have, for example, improved winning records in recent races are included, so that the gradient is adjusted.

[0042] The hit rate emphasis model unit 123 places emphasis on the winning rate of betting (low return), and performs calculation processing to raise the rank of the betting ticket (betting ticket) that is most likely to win in the race. The hit rate emphasis model unit 123 is a model that aims to predict the finishing order with certainty. Specifically, although the odds of the most popular event (horse) based on the basic competition data are lowered, emphasis is placed on the finishing order that increases the hit rate while maintaining the recovery rate, and a gradient adjustment is made so that the proportion of votes for the most popular event (horse) is preferentially increased.

[0043] The recovery rate emphasis model unit 122 or the hit rate emphasis model unit 123 is adjusted so as not to include features that are prone to overlearning or features that fluctuate after the announcement of horse weight. The fundamentals statistical data generation unit 124 generates reference predicted ranking data (fundamentals statistical data) that estimates the rankings between multiple events participating in the race to be provided to the pre-race ranking prediction data generation unit 130 (technical prediction) from the results of calculations performed by the recovery rate emphasis model unit 122 or the hit rate emphasis model unit 123.

[0044] Specifically, the reference predicted ranking data is a predicted finishing order of participating horses in a scheduled race that takes into account the results of calculations performed by the recovery rate emphasis model unit 122 or the hit rate emphasis model unit 123. At this stage, the ranking of the event is predicted based roughly on various information and data obtained before the race is held. The reference predicted ranking data is registered and updated in the database (DB) of the fundamentals prediction result database 507.

[0045] The pre-race ranking prediction data generation unit 130 (technical prediction) estimates the pre-race ranking of the event object in a race based on the reference predicted ranking data and odds information for each event object in the upcoming race, and generates pre-race predicted ranking data. The pre-race ranking prediction data generation unit 130 (technical prediction) predicts the race using feature quantities that change up until the race (race) deadline, such as the popularity of the event object (horse) and odds on betting tickets (betting tickets), as well as the results of the fundamental prediction and statistical data. In other words, this is added as a last-minute adjustment to improve the accuracy of betting ticket (betting ticket) selection.

[0046] 6 is a flow diagram showing the processing flow of the pre-race ranking prediction data generation unit 130 (technical prediction). In the pre-race ranking prediction data generation unit 130 (technical prediction), the technical learning model unit 132 executes calculations for ranking prediction based on various data stored in the race information racehorse database 501, fundamentals prediction result database 507, technical prediction feature database 506, odds popularity database 502, and statistics database 504.

[0047] The technical learning model unit 132 calculates the first, second, and third place rates (which horses will come in what place) for the competition items (each participating horse) in the upcoming race. For this purpose, in addition to the results of fundamental predictions and statistical data, the odds five minutes before the voting deadline are used as features. Note that the past odds and popularity data used as features during learning in the technical learning model unit 132 are data up to five minutes before the start deadline. By limiting the data used by dividing the time period, learning is performed to prevent new calculation results from being affected by leaks of the original calculation results. The model division and machine learning method used in the technical learning model unit 132 are the same as the calculation method used in the standard ranking prediction data generation unit 120 (fundamental prediction) described above.

[0048] From the predicted results of the finishing order calculated by the technical learning model unit 132, the technical statistical data generation unit 133 generates pre-race predicted ranking data (technical statistical data) of the predicted results to be used for predicting betting tickets to be purchased (recommended betting tickets). The pre-race predicted ranking data is registered and updated in the database (DB) of the technical prediction result database 508.

[0049] The pre-race predicted odds generation unit 140 (final odds estimation) estimates pre-race odds of the event object in the upcoming event based on the basic race data and generates them as pre-race predicted odds data. In this embodiment, in addition to the basic race data, the base predicted ranking data and the pre-race predicted ranking data are included.

[0050] 7 is a flow diagram showing the processing flow of the pre-race predicted odds generation unit 140 (final odds estimation). In the pre-race predicted odds generation unit 140 (final odds estimation), the final closing odds estimation unit 142 calculates final odds based on various data stored in the race information racehorse database 501, the technical prediction result database 508, the fundamentals prediction result database 507, and the odds popularity database 502.

[0051] The final win and place odds for the event (each horse participating in the race) are estimated from the odds 10 minutes before the race (race), the odds 5 minutes before the race, the results of the predictions mentioned above, and statistical data. For example, if 10 horses are scheduled to run in a race, from horse 1 to horse 10, the odds for horses 1 to 10 will be published separately. For example, if horse 4 is predicted to come in first, horse 7 to come in second, and horse 3 to come third, the odds for those horses will be generated as pre-race predicted odds data, such as 1.4 for horse 4, 3.6 for horse 7, and 6.1 for horse 3. Of course, the estimated odds also take into account odds based on how the betting ticket is purchased, such as win or place.

[0052] Based on the pre-race odds calculated by the final closing odds estimation unit 142, the purchase statistical data generation unit 143 generates purchase statistical data relating to the specific types of betting tickets (betting tickets) recommended for purchase in the race. The purchase statistical data is registered and updated in the database (DB) of the final odds prediction result database 509. The purchase statistical data makes it possible to search for the estimated final odds of each participating horse in the race.

[0053] The probability prediction unit 150 predicts the winning probability of a betting ticket when purchasing a betting ticket set for a sporting object in an upcoming sporting event based on the pre-race predicted ranking data and the pre-race predicted odds data, and predicts the payout amount attributable to the purchase amount of the betting ticket from the winning probability, thereby generating probability prediction result data for the upcoming sporting event of the sporting object. The probability prediction unit 150 also generates the probability prediction result data by including winning feature amounts related to betting tickets for the sporting object that have won in past sporting events.

[0054] 8 is a flow diagram showing the processing flow of the probability prediction unit 150 (betting probability prediction). In the probability prediction unit 150 (betting probability prediction), a calculation is performed in the hit rate prediction unit 152 (betting win rate prediction) to predict whether the betting ticket (betting ticket) in the race will be a hit, based on the various data stored in the race information racehorse database 501, the technical prediction result database 508, the fundamentals prediction result database 507, and the final odds prediction result database 509.

[0055] The winning probability prediction unit 152 (betting winning probability prediction) compares predictions with winning race results for races held in the past, and repeats learning between the predictions and winning betting tickets. The relationship found between past prediction results and actual winning betting tickets is calculated as a winning feature. Then, based on the accumulation of past data, a prediction calculation is performed to determine whether each betting combination for an upcoming race will be a winning ticket, and a winning probability prediction for each betting ticket (betting ticket) is calculated.

[0056] Based on the hit probability prediction data calculated by the hit rate prediction unit 152, the payout rate calculation unit 153 (statistical data + payout rate calculation) generates probability prediction result data regarding the amount of money (payout rate) that can be obtained if the betting ticket is a hit if the betting ticket is purchased. Specifically, the payout rate of the betting ticket (horse racing ticket) is calculated from the hit probability of the betting ticket (horse racing ticket) calculated from the hit probability prediction data and the product of the hit probability and the payout for the betting ticket (horse racing ticket) at the time of prediction. The payout rate here is the quotient obtained by dividing the payout amount by the purchase price.

[0057] The probability prediction result data is registered and updated in the database (DB) of the probability prediction result database 510. When generating the probability prediction result data, information on payout data 511 (via the database DB) provided by the organizer of the race that is the subject of betting may be added. The probability prediction result data makes it possible to search for the winning probability, payout rate, and payout amount of each betting ticket for each horse running in the race. Ultimately, a matching betting ticket (betting ticket) that falls within the set target winning probability and target payout probability range is searched for.

[0058] The betting ticket purchasing unit 160 (betting ticket purchase decision) generates betting ticket purchase data for purchasing a betting ticket that has been decided to be purchased based on the probability prediction result data from the organizer of the competition in which the competition to be bet is held.

[0059] 9 is a flow diagram showing the processing flow of the betting ticket purchasing unit 160 (betting ticket purchase determination). Based on the various data stored in the race information racehorse database 501, user information and settings database 512, and probability prediction result database 510, the betting ticket purchase determination unit 162 (betting ticket purchase determination logic) performs calculations to specifically identify the types and combinations of betting tickets (betting tickets) for the race in question, and generates betting ticket data to be purchased. The user information and settings database 512 stores information as setting values ​​about the betting ticket purchasing method desired by the user of the betting ticket purchase support system 1, specifically, the selection between the recovery rate emphasis model unit 122 and the hit rate emphasis model unit 123, and the respective proportions when combining both.

[0060] The betting ticket purchase determination unit 162 (betting ticket purchase determination logic) determines a combination of betting tickets (betting tickets) in accordance with the user's setting values ​​so that the composite odds, composite hit rate, composite return rate, etc. are close to the target values ​​set by the user, and generates this as betting ticket data to be purchased. The betting ticket data to be purchased is registered and updated in the database (DB) of the betting ticket database 513. In the betting ticket data to be purchased, it is possible to search for betting tickets (betting tickets) of the type determined and selected by the betting ticket purchase determination unit 162 (betting ticket purchase determination logic) for the race in question as betting tickets to be purchased.

[0061] Machine learning is used for the calculations and computations in each of the functional units described above. Specifically, machine learning is performed based on various data stored in the respective databases when generating reference predicted ranking data in the reference ranking prediction data generation unit 120 (fundamentals prediction) in Fig. 5, generating pre-race predicted ranking data in the pre-race ranking prediction data generation unit 130 (technical prediction) in Fig. 6, generating pre-race predicted odds data in the pre-race prediction odds generation unit 140 (final odds estimation) in Fig. 7, generating probability predicted result data in the probability prediction unit 150 (betting probability estimation) in Fig. 8, and generating planned betting ticket data in the betting ticket purchase unit 160 (betting ticket purchase determination). Machine learning is also performed when generating statistical data in the statistical data generation unit 112 of the competition basic data acquisition unit 110 (data generation unit) in Fig. 4 and when generating event object features for each event object in the feature generation unit 113.

[0062] Machine learning uses techniques such as support vector machines (SVMs), model trees, decision trees, neural networks, multiple linear regression, locally weighted regression, and probabilistic search methods. In the embodiment, a fourth-order high-order equation is calculated as an approximate formula using the multiple linear regression technique, as described in the examples below.

[0063] Now, using the flowchart in Figure 10, we will explain both the betting ticket purchase support method and the betting ticket purchase support program in the betting ticket purchase support system 1 of the embodiment. The betting ticket purchase support method is executed by the CPU 11 of the computer 10 based on the betting ticket purchase support program. The betting ticket purchase support program causes the computer 10 of Figure 1 to execute various functions, including a competition basic data acquisition function, a reference ranking prediction data generation function, a pre-competition ranking prediction data generation function, a pre-competition predicted odds generation function, a probability prediction function, and a betting ticket purchase function. Each of these functions is executed in the order shown in the figure. Note that each function overlaps with the description of the betting ticket purchase support system described above and the flow diagrams in Figures 4 to 9, so details will be omitted.

[0064] 10, the processing of the CPU 11 of the computer 10 includes various steps such as a step of acquiring basic competition data (S110), a step of generating predicted reference ranking data (S120), a step of generating predicted pre-competition ranking data (S130), a step of generating predicted pre-competition odds (S140), a step of predicting probability (S150), and a step of purchasing betting tickets (S160). Of course, the various steps necessary for the operation of the processing itself of the CPU 11 are naturally included.

[0065] The basic competition data acquisition function acquires basic competition data including statistical information on the competition object, odds information for the competition object, and competition result information for the competition object for the competition that the competition object is the target of betting on (S110; basic competition data acquisition step).The reference ranking prediction data generation function estimates the rankings of multiple competition objects participating in the competition based on the basic competition data and generates reference predicted ranking data (S120; reference ranking prediction data generation step).

[0066] The pre-competition ranking prediction data generation function estimates the pre-competition ranking of the object in a competition based on the reference predicted ranking data and odds information for each object in the competition that the object will compete in, and generates this as pre-competition predicted ranking data (S130; pre-competition ranking prediction data generation step).The pre-competition predicted odds generation function estimates the pre-competition odds of the object in the competition that the object will compete in, based on the competition basic data, and generates this as pre-competition predicted odds data (S140; pre-competition predicted odds generation step).

[0067] The probability prediction function predicts the probability of winning when a betting ticket set for a sport object in an upcoming sport event is purchased based on the pre-race predicted ranking data and pre-race predicted odds data, and generates probability prediction result data for the upcoming sport event (S150; probability prediction step).The betting ticket purchase function generates planned purchase betting ticket data for purchasing a betting ticket determined to be purchased based on the probability prediction result data from the organizer of the sport event to be bet on (S160; betting ticket purchase step).

[0068] The computer program of the present invention described above may be recorded on a processor-readable recording medium, and the recording medium may be a "non-transitory tangible medium" such as a tape, disk, card, semiconductor memory, programmable logic circuit, etc.

[0069] The computer program can be implemented using, for example, a scripting language such as ActionScript or JavaScript (registered trademark), an object-oriented programming language such as Objective-C or Java (registered trademark), or a markup language such as HTML5. [Example]

[0070] The inventors conducted a simulation of the betting ticket purchase support system of the present invention by obtaining information on horse races and participating horses from the Japan Racing Association from 2017 to 2021. The results are shown in Tables 1 to 15, and Figures 11, 12, and 13.

[0071] In Tables 1 to 5, the base is one horse, the opponent is a minimum of four horses, and the opponent is a maximum of six horses. And we bought equally.

[0072] [Table 1]

[0073] [Table 2]

[0074] [Table 3]

[0075] [Table 4]

[0076] [Table 5]

[0077] In Tables 6 to 10, the main horse is one, the opponent is a minimum of four, and the opponent is a maximum of six. Then, for all races, a bias allocation purchase was made based on the deviation value of the main horse calculated from the predicted results and the specified bias ratio. The purchase rate for the weighted allocation was calculated using the following formula. The purchase amount was 100 yen per item x purchase rate. The weighted ratio was set to 3.0. (1) If the overall deviation value is greater than or equal to the threshold Purchase rate = 1.0 + {(total deviation value - threshold) / slope ratio} (2) If the overall deviation value is less than the threshold Purchase rate = 1.0 - {(overall deviation value - threshold) / slope ratio}

[0078] [Table 6]

[0079] [Table 7]

[0080] [Table 8]

[0081] [Table 9]

[0082] [Table 10]

[0083] In Tables 11 to 15, the main horse is one, the opponent is a minimum of four, and the opponent is a maximum of six. Then, within the race, a bias allocation purchase was made based on the deviation value of the main horse calculated from the predicted result value and the specified bias ratio. The purchase rate for the weighted allocation was calculated using the following formula. The purchase amount was 100 yen per item x purchase rate. The weighted ratio was set to 3.0. (1) If the overall deviation value is greater than or equal to the threshold Purchase rate = 1.0 + {(total deviation value - threshold) / slope ratio} (2) If the overall deviation value is less than the threshold Purchase rate = 1.0 - {(overall deviation value - threshold) / slope ratio}

[0084] [Table 11]

[0085] [Table 12]

[0086] [Table 13]

[0087] [Table 14]

[0088] [Table 15]

[0089] From the results of Tables 1 to 15, we found that the return rate on the investment amount is higher when betting on the predicted rank order than when betting on the order of popularity of the participating horses (sports).

[0090] Next, we verified the prediction accuracy. In the graph of Figure 11, the vertical axis (y-axis) is the probability of winning, and the horizontal axis (x-axis) is the hit prediction result value + payback prediction result value. Then, the approximate formula was set as "y=ax^4+bx^3+cx^2+dx+e". Accuracy is determined by the coefficient of determination, R 2 " was determined as the accuracy using the least squares method. The data range is information on horse racing races and participating horses held by the Japan Racing Association from 2013 to the present day in 2021. In the graph of FIG. 12, the vertical axis represents the probability of coming in second, and in the graph of FIG. 13, the vertical axis represents the probability of coming in third; otherwise, the graph is the same as the graph of FIG.

[0091] Each graph and its coefficient of determination R 2 As a result of the simulation, an extremely high correlation was confirmed. [Explanation of symbols]

[0092] 1. Voting ticket purchase support system 2 Competition organizers 3 Computer (server) 4. Internet connection 5. Venue 6. Operators of voting ticket purchase support systems 10. Computers 11 CPU 12 ROM 13 RAM 14 Storage section 15 I / O 110 Basic Competition Data Acquisition Department 120 Standard ranking prediction data generation unit 130 Pre-competition ranking prediction data generation unit 140 Pre-race prediction odds generation section 150 Probability Prediction Section 160 Voting Ticket Purchase Department

Claims

1. a competition basic data acquisition unit that acquires basic competition data including statistical information about the competition, odds information for the competition, and competition results information for the competition that the competition object has previously participated in, in the competition that the competition object is being bet on; a reference ranking prediction data generation unit that estimates the ranking of a plurality of competing objects participating in the competition based on the competition basic data by gradient boosting rank learning and generates reference predicted ranking data; a pre-race ranking prediction data generation unit that estimates a pre-race ranking of the event object in a competition based on the reference predicted ranking data and odds information for each event object in a competition that the event object will play in the future, and generates the pre-race predicted ranking data; a pre-race predicted odds generating unit that estimates pre-race odds of the event object in a competition that the event object will play in the future based on the competition basic data and generates the pre-race predicted odds data; a probability prediction unit that predicts the winning probability of a betting ticket when a betting ticket set for a competition to be held in the future based on the pre-competition predicted ranking data and the pre-competition predicted odds data, and generates predicted probability result data for the competition to be held in the future, The competition basic data acquisition unit a statistical data generating unit that generates statistical data for each game object from the statistical information of the game object, odds information for the game object, and game result information for the game object; a feature generating unit that generates a game object feature for each game object in a game that the game object will play, The reference rank prediction data generation unit A recovery rate-focused model unit that performs calculations to raise the rank of voting tickets that are less popular; a hit rate-focused model unit that performs calculations to raise the rank of the ballot ticket that is most likely to win in the competition; and a fundamentals statistical data generating unit that estimates the rankings of a plurality of contestants participating in a contest from the results of calculations performed by the recovery rate emphasis model unit or the hit rate emphasis model unit, and generates reference predicted ranking data. A betting ticket purchase support system characterized by:

2. 2. The betting ticket purchase support system according to claim 1, further comprising a betting ticket purchasing unit that generates betting ticket purchase data for purchasing betting tickets determined to be purchased based on the probability prediction result data from the organizer of the competition in which the competition to be bet is held.

3. 3. The betting ticket purchase support system according to claim 1, wherein the competition basic data acquisition unit acquires the competition basic data from a sponsor of a competition in which the sport to be bet on is held.

4. 4. The betting ticket purchase support system according to claim 1, wherein the reference ranking prediction data generation unit, when estimating the rankings of a plurality of competition objects participating in a competition based on the competition basic data, generates the reference predicted ranking data by prioritizing the most popular competition object in consideration of odds information for the competition object, or generates the reference predicted ranking data by excluding the most popular competition object.

5. 5. A betting ticket purchase support system as described in any one of claims 1 to 4, wherein the probability prediction unit generates the probability prediction result data including winning features related to betting tickets for sports that have won in past sports events.

6. 6. A betting ticket purchase support system as described in any one of claims 1 to 5, wherein machine learning is used in generating the reference predicted ranking data in the reference ranking prediction data generation unit, generating the pre-competition predicted ranking data in the pre-competition ranking prediction data generation unit, generating the pre-competition predicted odds data in the pre-competition prediction odds generation unit, and generating the probability prediction result data in the probability prediction unit.

7. The computer a step of acquiring basic competition data including statistical information on the event, odds information on the event, and competition results information on the event in the past that the event to be bet on has been held in the competition that the event to be bet on is held in; a reference ranking prediction data generation step of estimating the rankings of a plurality of competing objects participating in the competition based on the competition basic data by rank learning using gradient boosting and generating reference predicted ranking data; a pre-race ranking prediction data generation step of estimating a pre-race ranking of the event object in a race based on the reference predicted ranking data and odds information for each event object in a race that the event object will be running in the future, and generating the pre-race predicted ranking data; a pre-race predicted odds generation step of estimating pre-race odds of the event object in a race that the event object will play in the future based on the race basic data and generating the pre-race predicted odds data; a probability prediction step of predicting the winning probability of a betting ticket when a betting ticket set for a competition to be held by the competition object is purchased based on the pre-competition predicted ranking data and the pre-competition predicted odds data, and generating probability prediction result data for the competition to be held by the competition object; The basic athletic data acquisition step includes: a statistical data generating step of generating statistical data for each game object from the statistical information of the game object, odds information for the game object, and game result information for the game object; a feature generating step of generating a game object feature for each game object in a game that the game object will play; The reference rank prediction data generating step includes: A recovery rate-focused model step that performs calculations to raise the rank of voting tickets that are less popular; a hit rate-oriented model step that performs calculations to raise the rank of the ballot ticket that is most likely to win in the competition; a fundamentals statistical data generation step of estimating the rankings of a plurality of contestants participating in a contest from the results of the calculations performed in the recovery rate emphasis model step or the hit rate emphasis model step, and generating reference predicted ranking data; A method for supporting the purchase of betting tickets.

8. The computer a basic competition data acquisition function that acquires basic competition data including statistical information on the competition, odds information on the competition, and competition results information on the competition that the competition to be voted on has previously held; a reference ranking prediction data generation function for estimating the ranking of a plurality of competing objects participating in the competition based on the competition basic data and generating the ranking as reference predicted ranking data; a pre-competition ranking prediction data generation function that estimates a pre-competition ranking of the object in a competition based on the reference predicted ranking data and odds information for each object in the competition that the object will be competing in by using gradient boosting rank learning, and generates pre-competition predicted ranking data; a pre-race predicted odds generation function for estimating pre-race odds of a sporting object in a sporting event that the sporting object will be holding based on the basic sporting event data and generating the pre-race predicted odds data; a probability prediction function that predicts the winning probability of a betting ticket when a betting ticket set for a competition to be held by a competition object is purchased based on the pre-competition predicted ranking data and the pre-competition predicted odds data, and generates probability prediction result data for the competition to be held by the competition object; The basic competition data acquisition function is a statistical data generation function that generates statistical data for each game from the statistical information of the game, odds information for the game, and game result information for the game; A feature generation function that generates a feature of each game object in a game that the game object will play in the future is realized. The reference ranking prediction data generation function is A recovery rate-focused model function that performs calculations to raise the rank of voting tickets that are less popular, and A hit rate-focused model function that performs calculations to raise the ranking of the most likely winning betting ticket in the competition; and a fundamentals statistical data generation function that estimates the rankings of a plurality of contestants participating in a competition from the results of calculations performed in the recovery rate emphasis model function or the hit rate emphasis model function, and generates reference predicted ranking data. A voting ticket purchase support program characterized by:

Images