US 20070072679 A1 Abstract A valuation system determines a win probability, which is a measure of the likelihood a participant in a particular game is to win the game given a current game state. The valuation system may determine the win probability from an analysis of events that have occurred in actual games that have occurred over a predetermined time period in the past. The valuation system may then provide the win probability to various media outlets for display or broadcast, for example, within the context of the game. The valuation system may also display or broadcast the win probability through its own broadcast source, such as its web site.
Claims(20) 1. A computer-readable medium whose contents cause a computing system to determine in real-time a win probability for a participant of an athletic competition during the athletic competition, the contents cause the computing system to:
determine a current state in the athletic competition, the current state determined from at least one factor that describes the state of the athletic competition; and determine the win probability for the participant based on the current state, the win probability determined from an analysis of a sufficient number of states that have occurred during past athletic competitions. 2. The computer-readable medium of
3. The computer-readable medium of
4. The computer-readable medium of
5. The computer-readable medium of
6. The computer-readable medium of
7. The computer-readable medium of
8. The computer-readable medium of
9. The computer-readable medium of
10. The computer-readable medium of
11. The computer-readable medium of
12. The computer-readable medium of
13. The computer-readable medium of
14. A method in a computing system to determine in real-time a win probability for a participant of an athletic contest during the athletic contest, the method comprising:
determining a current state in the athletic contest, the current state determined from factors that describes the state of the athletic contest; and determining the win probability for the participant based on the current state, the win probability determined from an analysis of a sufficient number of states that have occurred during past athletic contests. 15. The method of
16. The method of
17. A method in a computing system for determining in real-time values of game states occurring during a sports competition, the method comprising:
identifying a first game state in the sports competition, the first game state determined from at least one factor that describes the state of the sports competition; determining a first win probability based on the first game state; determining a second game state immediately succeeding the first game state; determining a second win probability based on the first game state; and determining a value for the second game state from the first win probability and the second win probability, wherein the first and second win probabilities are determined from an analysis of a sufficient number of states that have occurred during past sports competitions, the states determined from at least one factor that describes the state of the sports competition. 18. The method of
19. The method of
20. The method of
Description This application is a Continuation-In-Part (CIP) of U.S. patent application Ser. No. 11/186,502, entitled “Real-Time Play Valuation,” which was filed on Jul. 21, 2005, and identified by attorney docket number 430638001US, which is related to U.S. patent application Ser. No. 11/186,503, entitled “Athlete Valuation,” which was filed on Jul. 21, 2005, and identified by attorney docket number 430638002US, and U.S. patent application Ser. No. 11/186,602, entitled “Fantasy Single Sport Challenge,” which was filed on Jul. 21, 2005, and identified by attorney docket number 430638003US, and U.S. patent application Ser. No. 11/186,368, entitled “Fantasy Cross-Sport Challenge,” which was filed on Jul. 21, 2005, and identified by attorney docket number 430638004US, and U.S. patent application Ser. No. 11/186,469, entitled “Graphical User Interface for a Fantasy Sports Application,” which was filed on Jul. 21, 2005, and identified by attorney docket number 430638005US, the disclosures of which are incorporated by reference herein in their entireties. With the increasing number of available sporting event telecasts, the broadcasters of the sporting events are finding it increasingly difficult to attract and maintain their viewers. For example, many sporting enthusiasts who initially start viewing a telecast of one sporting event often lose interest in the telecast and “surf” or change channels to view a different telecast of another sporting event. In an effort to maintain the interest of viewers of telecast sporting events, broadcasters of sporting events have started to provide additional content, such as information regarding other sporting events, both in the same sport as the sporting event being telecast as well as in other sports, during the telecasts. For example, many sporting event telecasts provide a constant feed of updated scores of other sporting events. As another example, many sporting event telecasts incorporate in-game updates, which are highlights from other sporting events, into their telecasts. The viewing content currently being provided by the broadcasters of sporting events has not been totally effective in maintaining the interest of the viewers. An approach to measuring the value or significance of each play throughout the course of an athletic game or athletic competition in real-time is provided. The terms “game” and “competition” as used herein refer to the actual real-life athletic game, contest or competition, and are used interchangeably herein. Also, the term “real-time” as used herein refers to processing as the plays are occurring within the game or shortly after the completion of the game. In some embodiments, the value or significance of each play is based on the expectation of scoring (i.e., number of runs, points, or other scoring method) based on the state of the game (also referred to as “game state”). For example, at any given point in time during a game, a team has an expectation of the number of runs or points the team will score. In an individual sport, a player will have an expectation of the number of runs or points the player will score. The expectation of the number of runs or points is based on the state of the game at the given point in time during the game—i.e., prior to a play occurring during the game. The factors that determine the game states vary from sport to sport. For example, for baseball, the factors that determine a game state may include the current score, the current inning, the number of outs in the inning, the number of base runners and position of each of the base runners, etc. For football, the factors that determine a game state may include the current time in the game, the current down and distance, the spot or position on the field, the current score, weather conditions, etc. Net expected score (NES) refers to the future expected scoring, and a NES is calculated for every game state. The NES for each game state may be determined based on an analysis of the events that have occurred in actual games in a particular sport over a predetermined time period in the past, such as, by way of example, three years. The NES for each game state may also be determined using techniques such as, forward simulations, random value assignments, and the like. The number of events analyzed for each specific situation—i.e., game state—needs to be sufficient to provide an adequate event pool for deriving a NES equation that receives as input the factors that describe a game state and produces a NES for the described game state. In some embodiments, each of the events in the event pool is analyzed to determine an expectation of the number of runs that should be scored—i.e., NES—for each game state. The NESs determined from the analysis of the event pool and their games states are then analyzed, for example, using any of a variety of well-known regression techniques, to derive a NES equation (e.g., linear regression equation, cubic regression equation, etc.) that outputs a NES for any game state. The derived NES equation may include constants, coefficients and/or variables whose values are based on the factors that describe the game state. For example, in baseball, the number of events over a three year time span may amount to over 500,000 individual balls put in play and strikeouts. The 500,000-plus events are analyzed to determine the game states, and the game states are further analyzed to determine the NES for each game state. The determined NESs and their game states may be analyzed using any of the well-known regression techniques to derive a NES equation that is suitable for outputting a NES for each game state. A similar analysis can be performed for an adequate number of past events in other sports, such as football, basketball, etc., to determine a NES equation that is suitable for producing a NES for each game state that is based on the analyzed pool of past events. In some embodiments, a NES for a game state may be an expectation of the number of runs that should be scored during a particular segment of a game. The appropriate game segment is dependent on the characteristics of the particular sport, and is generally based on the determination of the effect a game state has on the remainder of the game. For example, for baseball, a NES for a game state may be an expectation of the number of runs that should be scored during the inning in which the particular game state occurs. Here, the game segment is one inning—i.e., the inning in which a game state occurs—and this is based on the premise that game states do not carry-over from one inning to another, succeeding inning. For football, a NES for a game state may be an expectation of the number of points that should be scored on a current drive—i.e., the current offensive possession in which the particular game state occurs. Here, the game segment is the current drive, which is based on the premise that game states are not likely to impact drives other than the drive in which the particular game state occurred. In some embodiments, the appropriate game segment for a game state may be determined based on the premise that the particular game state impacts a multiple number of successive possessions—e.g., multiple possessions in basketball, multiple possessions in football, etc. For example, for football, based on the effect of a play on a concept generally known as “field position,” a change in game state may affect not only the current offensive team's drive in which the change in game state occurs but also the opponent's next drive, the next drive for the current offensive team, etc. Each NES is an expectation of the runs that each game state is worth—i.e., the value of each game state. Then, for any given play, a NES is determined for the game state existing prior to the play (pre-play NES) and a new NES is determined for the game state that is created after the play (post-play NES), and the value of the play is the difference between the NES for the ending game state (post-play NES) and the NES for the beginning game state (pre-play NES) plus the runs generated on or by the play. In a situation where a number of plays follow each other in sequence, the post-play NES for a just completed play becomes the pre-play NES for the immediately succeeding play. Even though the score valuation term may vary based on the sport (i.e., “points” for basketball and football, “runs” for baseball, “goals” for hockey and soccer, etc.), for ease of explanation, NES will be used herein to refer to the unit of measure for the value of a game state irrespective of the actual sport and the sport's scoring method. For example, in a particular baseball game between Team X and Team Y in Team Y's baseball stadium, Player A for Team X comes to the plate in the third inning of a tie game with Player P pitching for Team Y. There are no runners on base, and there is nobody out in the inning. Given this game situation and from the analysis of the past events, Team X can expect to score 0.54 runs in this inning from this point on. 0.54 is the NES for this game state. During this at bat—i.e., play,—Player A hits a single. At the end of the play, Player A is now on first base and there are still no outs in the inning. This is a positive result for Team X, and given this current game situation and from the analysis of the past events, Team X can now expect to score 0.91 runs in this inning from this point on. 0.91 is the NES for this new game state. Therefore, the value of the play—i.e., the total effect of the play—is +0.37 runs (i.e., 0.91−0.54). Continuing the example, the next batter, Player B, comes up to bat with Player A still on first base and no outs in the inning. The NES for this game state is the 0.91 determined above. Player B strikes out during this at bat, leaving Player A on first base. The new game state is a base runner on first base with one out. This is a negative result for Team X, and given this current game situation and from the analysis of the past events, Team X can now expect to score 0.55 runs in this inning from this point on. 0.55 is the NES for this new game state. Therefore, the value of the play—i.e., the total effect of player B's strikeout—is −0.36 runs (i.e., 0.55−0.91). The next batter, Player C, comes up to bat with Player A still on first base and one out in the inning. The NES for this game state is the 0.55 determined above. Player C hits a double to the gap between left field and center field scoring Player A from first base. As a result of the play, one run scores and a base runner, Player C, is on second base with one out. This is a positive result for Team X, and given this current game situation—i.e., a base runner on second base with one out in the inning—and from the analysis of the past events, Team X can now expect to score 0.72 runs in this inning from this point on. 0.72 is the NES for this new game state. Therefore, the value of the play—i.e., the total effect of player C's double—is +1.17 runs (i.e., 0.72−0.55+1 run scored). The NES concept can be similarly applied to other sports. For example, in a particular football game between the Rams and the Lions in the Lion's home stadium, the Ram's offensive unit may have possession of the football on first down-and-ten yards to go at their own 30 yard line in the first quarter. Given this game situation and from the analysis of the past events, the Ram's can expect to score 0.75 points during this offensive possession. 0.75 is the NES for this game state—e.g., a visiting team playing in the Lion's home stadium with the football on first down-and-ten yards to go at the visiting team's 30 yard line in the first quarter. On first-and-ten, the Ram's call a running play and their halfback runs the ball for 30 yards to the Lion's 40 yard line. As a result of this running play, the Ram's have the ball on first down-and-ten yards to go at the Lion's 40 yard line. This is a positive result for the Rams, and given this current game situation and from the analysis of the past events, the Ram's can now expect to score 1.4 points during this offensive possession. 1.4 is the NES for this new game state. Therefore, the value of the play—i.e., the total effect of the rush for 30 yards on first-and-ten to the opponent's 40 yard line—is +0.65 points (i.e., 1.4−0.75). In basketball, an offensive possession may be considered a play. In auto racing, each lap of a race may be considered a play. In some embodiments, a play may occur each time a change in the race standings occur during an auto race. Similarly, in sports such as cycling, boat racing, etc. where the competitors race around a track or course multiple times to determine a winner, each lap of the race may be considered a play, or each time the standings change during the race may be considered a play. In duration races, such as the Tour de France, each leg of the race may be considered a play. In rugby, each change of possession may be considered a play. In volleyball, each point or “side-out” may be considered a play. In track-and-field, each heat, event, or trial may be considered a play. In some embodiments, the valuation system determines in real-time values of plays occurring during games by identifying a play that occurred during a game. The system determines an expectation of scoring prior to the play, and an expectation of scoring that results from the play occurring—i.e., the expectation of scoring resulting after the play. The system then calculates a value of the play as the difference between the expectation of scoring that results from the play occurring and the expectation of scoring that existed prior to the play occurring, and any scoring generated by the play. The various embodiments of the techniques for measuring the value or significance of each play throughout the course of a game in real-time and their advantages are best understood by referring to In general terms, the valuation system provides real-time valuation of each play that occurs within a game as it relates to each play's effect on the outcome of the game. The valuation system analyzes a play-by-play game feed to calculate how each play impacts a team's scoring expectancy during the game or during appropriate segments of the game. The value of a play is the play's impact on the team's scoring expectancy. For example, if the team is expected to score 0.5 runs prior to a play, and the team is expected to score 0.7 runs after executing the play, the play's value is the difference in the team's scoring expectancy as a result of the play (i.e., 0.7−0.5=0.2). In some embodiments, the valuation system distributes a play's value amongst some or all of the athletes identified as being involved in the play. In some embodiments, the valuation system provides users the ability to create and/or participate in an interactive fantasy sports challenge in which participants act as “investors,” “traders,” “coaches,” “managers,” etc., to form their own portfolio of athletes from among active real-life athletes, and in which a winner of the fantasy sports challenge is based in part on the value of the portfolio and the real-life athletes' performance as determined by the play valuation as described herein. Athlete valuation and fantasy sports challenges are further described below. In some embodiments, the valuation system comprises a web server which functions to provide a web site that provides access to some or all of the features (i.e., services, functionality, data, etc.) provided by the valuation system. The web site may be identified and addressable by the hostname part of a uniform resource locator (URL). For example, the web site may provide a web page or multiple web pages, or other user interfaces (UIs) including graphical user interfaces (GUIs), through which users can obtain real-time valuation of plays during the course of a game. The users may also be able to obtain real-life athlete valuations based on the athlete's performance in a game based on the valuation of the plays occurring in the game. The web site may provide a web page or multiple web pages through which users can register and create/participate in interactive fantasy sports challenges, monitor the value of the portfolios, alter the composition of portfolios (e.g., buy/sell shares of financial instruments in real-life athletes), and perform other actions. The play-by-play feed provider provides the play-by-play game feed to the valuation system. In some embodiments, the play-by-play feed provider sends the valuation system messages that contain the play-by-play game feed. For example, the play-by-play feed provider may provide the valuation system the live, play-by-play game event feed for football games, baseball games, basketball games, etc. Only one play-by-play feed provider is shown in The client systems may include any type of computing system that is suitable for connecting to and accessing the valuation system. In some embodiments, each of the client systems has a web client computer program, such as any of a variety of well-known web browser programs suitable for connecting to and interacting with the valuation system. The network is a communications link that facilitates the transfer of electronic content between, for example, the attached valuation system, play-by-play feed providers, and the client systems. In one embodiment, the network includes the Internet. It will be appreciated that the network may be comprised of one or more other types of networks, such as a local area network, a wide area network, a point-to-point dial-up connection, and the like. It will also be appreciated that the network may not be present. For example, the play-by-play feed provider and the valuation system may both reside on the same computing system and communicate via communication mechanisms, such as, interprocess communication, remote function call, internal communication buses, etc., typically supported on the computing system. The computer systems on which the valuation system, the play-by-play feed provider, the client systems, and other components described herein can execute may include a central processing unit, memory, input devices (e.g., keyboard and pointing devices), output devices (e.g., display devices), and storage devices (e.g., disk drives). The memory and storage devices are computer-readable media that may contain instructions that implement the server application, client application, and other components. In addition, the data structures and message structures may be stored or transmitted via a data transmission medium, such as a signal on a communications link. Various communication links may be used, such as the Internet, a local area network, a wide area network, a point-to-point dial-up connection, a cell phone network, and so on. Embodiments of the valuation system may be implemented in various operating environments that include personal computers, server computers, computing devices, hand-held or laptop devices, multiprocessor systems, microprocessor-based systems, programmable consumer electronics, digital cameras, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and so on. The computer systems may be cell phones, personal digital assistants, smart phones, personal computers, programmable consumer electronics, digital cameras, and so on. The valuation system may be described in the general context of computer-executable instructions, such as program modules, executed by one or more computers or other devices. Generally, program modules include routines, programs, objects, components, data structures, and so on that perform particular tasks or implement particular abstract data types. Typically, the functionality of the program modules may be combined or distributed as desired in various embodiments. The feed server comprises a plurality of play-by-play receivers 208, feed processor 210, and a messaging service 212. The play-by-play receivers provide connectivity to, and receive from the play-by-play feed provider the play-by-play messages—i.e., game feeds. The feed server may create an instance of a play-by-play receiver for each play-by-play game event feed that is being received. The feed processor receives from the play-by-play receivers the messages received from the play-by-play feed provider, and transforms the messages into an internal format, such as XML. The feed processor stores the received messages and the XML messages in the database, and sends a copy of the XML messages—i.e., the XML format of the play-by-play game feeds—to the messaging service. The messaging service sends the received XML messages describing the play-by-play game feed to the sport server for processing. In general terms, the sport server processes the XML play-by-play messages to determine a valuation for each play in real-time. The sport server comprises a feed receiver 214, a command manager 216, document handlers 218, play-by-play handlers 220, and a performance valuation component 222. The feed receiver receives the XML messages describing the play-by-play game feed from the messaging service component of the feed server. The command manager manages the multiple concurrent games by performing the necessary sequencing of the received XML play-by-play messages in order to ensure proper determination of game states. The document handlers parse the XML play-by-play messages. In some embodiments, the sport server maintains different document handlers for the different XML document types. For example, different document types may be used for the different types of play-by-play feeds—e.g., football, hockey, baseball, basketball, athlete statistics, etc. The play-by-play handlers process the parsed XML play-by-play messages to create the game event objects. For example, the play-by-play handlers aggregate the play-by-play data with the sport and athlete data from the database to create the plays, game states, and performance events for the athletes involved in the play-by-play game events. The play-by-play feeds, instead of expressly identifying the athletes by name and/or describing the play, may identify the athletes and/or describe the play using unique identifiers. For these feeds, the play-by-play handlers dereference the identifiers contained in the XML play-by-play play messages to identify the athletes and/or the play, as well as other information related to the description of the play that is contained in the XML messages. In some embodiments, the sport server maintains different play-by-play handlers for the different sports. The performance valuation component computes the value of each play that occurs within a game as it relates to each play's effect on the outcome of the game. In some embodiments, the value of each play is determined based on the game states of the current play and the previous play, and the corresponding NES for the two game states. The NES may be based on an expectation of the play based on an analysis of historical game states that occurred in actual games over a period of time. In some embodiments, the performance valuation component valuates the performance of the individual athletes involved in a play by assigning positive or negative performance points according to their participation in the play. The performance points assigned to an athlete can be thought of as a “dividend” generated by the athlete as a result of the athlete's participation in a play. For example, the performance valuation component identifies the athletes involved in a play, and based on each individual athlete's contribution and situational expectation of the play, the performance valuation component appropriately divides the value of the play amongst the identified athletes. The value attributed to an athlete is the athlete's performance points or generated dividend. In some embodiments, the performance valuation component breaks down a play into its intermediate states or components that represent the flow of the play, calculates a value for each intermediate state, identifies the athletes involved in each of the intermediate states and, for each intermediate state, appropriately distributes the value of the intermediate state amongst the athletes involved in the particular intermediate state. The intermediate states may correspond to athlete performance events which are discrete, measurable components of the play. For example, for baseball, the athlete performance events may include the interactions between pitcher and hitter (e.g., pitcher pitching and the batter hitting), the batted ball and fielder (e.g., the fielder fielding a batted ball), the fielder and base runner (e.g., the base runner advancing and the fielder preventing the base runner from advancing), etc. For football, the athlete performance events may include the quarterback and receiver (e.g., quarterback throwing a pass and the receiver catching the pass), the receiver and defensive back (e.g., the receiver running after catching a pass and the defensive back defending the pass and tackling the receiver), etc. The intermediate states have corresponding game states, and the series of game states represent the play from its beginning game state to its ending game state. The performance valuation component determines a value for an intermediate state from the intermediate NESs associated with the intermediate states. The performance valuation component determines an expectation of the each intermediate state based on the characteristics of the intermediate state. The expectation of the intermediate state is a probability of outcomes for the intermediate state determined from, for example, an analysis of the events and, in particular, an analysis of the intermediate states of the events that have occurred in actual games. The expectations of the intermediate states may be maintained in a table of relative probabilities in the database. The performance valuation component determines an intermediate NES based on the expectation of the intermediate state. The performance valuation component determines another intermediate NES based on the actual result of the intermediate state and calculates a value for the intermediate state using the two intermediate NESs. The performance valuation component identifies the athletes involved in the intermediate state and appropriately distributes the calculated value of the intermediate state amongst the athletes involved in the intermediate state. Each athlete's proportion of the calculated value is the athlete's performance points or dividend generated as a result of the athlete's participation in the intermediate state of the play. The performance valuation component similarly repeats this process for the subsequent intermediate states of the play as necessary. In some embodiments, the performance valuation component may calculate a value for a play without breaking the play down into its intermediate states. In this instance, the performance valuation component may distribute the calculated value amongst the athletes involved in the play without taking into consideration various aspects of the play, such as, by way of example, defense, base running, etc. This allows for a simpler distribution of the play's value amongst a potentially smaller number of athletes. In general terms, the fantasy game server hosts the interactive fantasy sports challenges. The fantasy game server allows registered users to act as “challenge creators” and create custom fantasy sports challenges. The fantasy game server may also provide fantasy sports challenges. For example, a fantasy sports administrator may act as a challenge creator and create fantasy sports challenges which the registered users can participate in. Registered users, including challenge creators, can then participate in one or more fantasy sports challenges. In some embodiments, participants of a fantasy sports challenge may be allowed to buy, sell, swap, discard, win, give away, steal, lose, trade or otherwise acquire or divest financial instruments (or “various instrument vehicles”) in athletes. For example, financial investments may be comprised of shares of individual athletes, mutual funds of athletes, index fund of athletes, and other types of financial instruments comprised of interests in athletes. Shares of an athlete represent an equity interest in the athlete and a right to dividends generated by the athlete's on-field performance. A mutual fund invests in a diversified group of athletes, and a share of the mutual fund effectively represents an equity interest in each of the underlying athletes. The mutual fund's underlying securities may be traded, realizing a gain or loss, and the fund collects the dividend. The investment proceeds are then passed along to the individual participant investors. An index fund may hold a portfolio of athletes' shares designed to match the price and dividend performance of the entire market or one of its sectors, and a share of the index fund effectively represents an equity interest in each of the underlying athletes. The index fund's underlying securities may be traded, realizing a gain or loss, and the fund collects the dividend. The investment proceeds are then passed along to the individual participant investors. In some embodiments, the fantasy game server interacts with the other components of the valuation system to enable the participants of fantasy sports challenges to establish their portfolios comprising of shares of financial instruments in real-life athletes, alter the composition of their portfolios by trading—i.e., buying and/or selling—the shares of the financial instruments, and perform other actions associated with participating in the interactive fantasy sports challenges. The fantasy game server also receives information from the other components of the valuation system to determine the value of the portfolios during the course of the interactive fantasy sports challenges and, at the end of a challenge, determines the ending value of the portfolios that are participating in the challenge. In some embodiments, the fantasy game server determines a value of a portfolio based on the market value of the shares of financial instruments contained in the portfolio and the dividends accrued by the shares of financial instruments during the duration of the interactive fantasy sports challenge. Interactive fantasy sports challenges are further discussed below. In general terms, the market exchange server facilitates the trading and selling of shares of financial instruments. In some embodiments, the market exchange server functions as a “market maker” in that the market exchange server buys/sells shares of financial instruments from/to the users of the valuation system, such as the participants of the interactive fantasy sports challenges. In some embodiments, the market exchange server establishes an initial public offering (IPO) price for a share of each financial instrument. The market exchange server may establish the IPO price for a share of a financial instrument in an athlete based on an expectation of that athlete's expected performance in games (e.g., the expected performance points or dividends) during the course of the athlete's career. In some embodiments, the market exchange server sits at the end of each buy/sell transaction and adjusts the price of the shares to create a constant trade balance between buying and selling (also referred to herein as “buy-sell pressure”). The market exchange server attempts to achieve a constant trade balance by adjusting the price of the shares accordingly. In general terms, the web server functions to provide a web site through which users can access some or all of the features provided by the valuation system. For example, the web server may provide web pages through which users can register and interact with the valuation system to: obtain/view real-time valuation of plays occurring in a game; obtain/view athlete valuations based on the athlete's performance in a game; participate in interactive fantasy sports challenges; monitor the value of their portfolios; alter the composition of their portfolios by buying/selling shares of financial instruments in real-life athletes; and perform other actions supported by the valuation system. In some embodiments, the valuation system may provide the NES to various media outlets, such as, television networks, cable networks, radio networks, satellite networks, and other media providers, for use by the media outlets. By way of example, a television network, such as, by way of example, ESPN, might enter into a license agreement to receive the NES from the valuation system and display the NES on its web site. ESPN may then display the received NES within the context of the game (e.g., video feed of the game) for which the NES applies. For example, during the video feed of a game, ESPN might display the NES prior to each play during the game to provide viewers a projected scoring expectation or value prior to each play actually occurring during the game. In some embodiments, the valuation system may broadcast the NES via its own web site in a similar manner. In some embodiments, the valuation system may provide or make available the NES on a subscription basis. For example, one or more individuals desiring to receive and use the NES may subscribe to receive the NES. For example, a subscriber may receive and use the NES to perform his or her own analysis of athletes. In general terms, the database functions as a repository for the data and information used by the components of the valuation system to facilitate the real-time valuation of each play that occurs within a game as disclosed herein. For example, the database may represent and/or store information and data such as the live play-by-play game feeds, the XML documents that correspond to the plays in the game feeds, historical game data, the game states and the factors corresponding to the game states, the text description of the plays, the NES coefficients that are used in the NES equation for athlete performance valuation in a given play, etc. In some embodiments, the database may be implemented using a SQL server. In some embodiments, the database functions as a repository for the data and information used by the components of the valuation system to facilitate the valuation of athletes. For example, the database may store the probabilistic models generated from the historical game data, the athlete valuation distribution tables, and other information used to assign athletes performance points or dividends. In some embodiments, the database functions as a repository for the interactive fantasy sports challenge data and the market data. The aforementioned components of the valuation system are only illustrative and are not intended to suggest any limitation as to the implementation of the illustrated components and/or the scope of use of the valuation system. For example, in various embodiments, one or more of the components of the feed server may be implemented as components of the sport server. Moreover, the valuation system may not include one or more of the depicted components or may include other components and modules not depicted. Furthermore, the functionality provided for in the components and modules may be combined into fewer components and modules or further separated into additional components and modules. In the discussion that follows, embodiments of the valuation system are described in conjunction with a variety of illustrative examples. It will be appreciated that the embodiments of the valuation system may be used in circumstances that diverge significantly from these examples in various respects. Play Valuation In block 304, the games states corresponding to the events in the event pool are determined, for example, from the conditions that describe the events. In block 306, the game states are analyzed to determine their results. For example, the result may be the number of points that is generated in the segment of the game affected by a game state. In block 308, the game states and their corresponding results are analyzed to derive one or more NES equations suitable for determining the net expected scoring for a game state. In some embodiments, the game states and their corresponding results may be analyzed using weighted least squares regression. For example, analyzing the game states and their results in a baseball event pool using the above regression technique may generate the following NES regression equation for baseball:
The NES for baseball is primarily dependent upon NUMOUTS (i.e., the number of outs in the inning) and the configuration of the runners on base (e.g., Runner on 1st, Runner on 2nd, Runner on 3rd all take the value of 1 if a runner is on that base and 0 otherwise). NES is also dependent, though less so, on the league the game is played in (e.g., League=1 if it is an American League game 0 otherwise), whether the home team is at bat (e.g., Home is 1 if the home team is at bat and 0 otherwise), the inning of the game (e.g., 1^{st }inning, 9^{th }inning and extra innings are all 1 if in that inning and 0 otherwise), the position in the lineup of the batter (e.g., Lineup position is equal to the number of the batter in the batting order), the number of runs scored by the offense (e.g., Offense score), whether the pitcher and batter are of opposite hand (e.g., opposite hand is 1 if they are opposite and 0 otherwise), whether the batter is a pinch hitter (e.g., 1 if the batter is a pinch hitter and 0 otherwise), the number of pitches thrown by the pitcher in the game (i.e., pitchcount) and the park that game is played in (e.g., each park has a unique score that either increases or decreases NES based on the correlation of the park with total runs scored). The values for the above variables are determined based on the factors that describe a game state, and inputting values for the variables as determined from the factors that describe the game state into equation [1] produces a NES for the particular game state. Table 1 below illustrates example NES outcomes generated using equation [1] for a particular combination of factors and the listed game states or situations (i.e., combination of the number of outs and the position of the base runners) for a particular baseball stadium or park.
Similar tables can be generated for the other combinations of the factors and for the other baseball stadiums or parks. In some embodiments, one or more variables (i.e., factors) listed above may be omitted in calculating the NES. In the case of football, a game state may be described by a number of primary factors and a number of other less significant factors. The primary factors may include the down, yards to go for a first down, and yards away from the goal, and their effects on the expectation of scoring are predictable—e.g., each is more favorable for points being scored when it is lower (first down is better than third down, and closer to the first down or the end zone is better than farther away). Other factors that are less central to the value of a play, but which may still contribute in determining the value of the play may include, for example, time remaining, score and score differential, whether or not the team is at home or away, what the surface of the field is (with turf being more conducive to scoring than grass), and the weather (measured through humidity and temperature). Any of a combination of these variables may be included in a regression equation to determine how many points the offensive team was expected to score prior to the play. After the play, the expectation is calculated again, using the new down, yards to go, and so forth. The expectation prior to the play is subtracted from the expectation after the play to arrive at the NES of that particular play. Analyzing the game states and their results in a football event pool using the above regression technique may generate the following NES regression equation for football:
The variables in equation [2] have the following meanings:
The values for the above variables are determined based on the factors that describe a game state, and inputting values for the variables as determined from the factors that describe the game state into equation [2] produces a NES for the game state. In equation [2], because the value of the N9 variable is 1 on normal plays, it has no effect on most plays. There are some plays, however, where players perform acts that are generally considered negative but, in context, are either beneficial or at least not harmful to the team. For instance, a hail mary at the end of a game may be the only way for a team to get the score it requires to tie or win the game. If this play is intercepted, negative value should not be assigned to the play nor should negative credit be given to the player—i.e., quarterback—who threw the interception, since this play did not hurt the team's chances of winning. Therefore, the N9 variable is 0 in this case, which makes the entire play value 0 as well. This is also the case for kneel downs at the end of the game to run the clock out and intentionally grounded passes—i.e., spikes—to stop the clock. The variables N3-N6 in equation [2] account for the concept of field position in football. While moving the ball from a team's own twenty yard line to midfield and then punting does not result in any points on this drive for the team, it lessens the opponent's chances of scoring on their ensuing drive. The effects of the current play on the opponent's following drive are taken into account by the variables N3 and N4 in equation [2]. Beyond the opponent's next drive, the effects of the current play on the current offensive team's next drive are taken into account by the variables N5 and N6 in equation [2] in a similar manner. Therefore, in the valuation system, the value of a football play is the play's cumulative effect on the expectation of scoring on the current drive (variable N2), the opponent's ensuing drive (variables N3 and N4), and the drive after that (variables N5 and N6). One factor in determining the value of the variables N2-N6 (the change in expected scoring) in equation [2] is the position on the field of play—i.e., the yards from goal. The yards from goal of the offensive team's current drive, which is needed to determine a value for variable N2, can be readily determined from the current game state. The expected yards from goal of the opponent's ensuing drive, which is needed to determine values for variables N3 and N4, and the current offensive team's next drive, which is needed to determine values for variables N5 and N6, may be determined from an analysis of the game states in the event pool. A regression analysis of the game states in the event pool may generate the following cubic regression equation for calculating the expected yards from goal of the opponent's next drive and the expected yards from goal of the current offensive team's next drive:
where x=the current drive's yards from goal when calculating the expected yards from goal of the opponent's next drive; or the opponent's expected yards from goal when calculating the expected yards from goal of the current offensive team's next drive. In addition to the yards from goal or expected yards from goal factor, the values for variables N3-N6 in equation [2] may be determined from one or more additional factors that describe the game states. For example, a regression analysis of the game states in the event pool may generate one or more regression equations for calculating values for these variables. By way of example, the regression formula for determining the value of N3 may include variables and coefficients for the variables such as, the number of yards from the goal line, the current down, the number of yards to go to gain a first down, the time remaining in the half or game, weather conditions (e.g., temperature and humidity), home field factor, type of playing surface, etc. One skilled in the art will appreciate that, for this and other processes and methods disclosed herein, the functions performed in the processes and methods may be implemented in differing order. Furthermore, the outlined steps are only exemplary, and some of the steps may be optional, combined with fewer steps, or expanded into additional steps without detracting from the essence of the invention. In some embodiments, the NES may be based on simulation. For example, a model may be created where individuals (athletes) or groups (teams) participating in a game would be represented by “agents.” These agents can then be given different characteristics that represent different attributes and skill levels. The outcome of a game or a game situation can then be simulated, for example, using a computing device, and this result may be used to determine the expected scoring. Football Play Valuation Example Team W is on offense and has the ball at its own 40 yard line in the first quarter of a tie game at home versus Team D. The current down and distance is first-and-ten. Based on this situation—i.e., game state—and from, for example, equation [2] above: Team W is expected to score 2.2 points on its current drive; Team D is expected to score 1.5 points on its next drive; and Team W is expected to score 1.6 points on its next drive. On Team W's next play, its quarterback throws a pass to its halfback for a seven yard gain, bringing up second-and-three from its own 47 yard line. Based on this new game state: Team W is now expected to score 2.6 points on its current drive; Team D is expected to score 1.3 points on its next drive; and Team W is expected to score 1.7 points on its next drive. The value of the just completed play—i.e., the seven yard reception—is the sum of the difference of the scoring expectations before and after the play for Team W's current drive, Team D's next drive, and Team W's next drive. In this example, the value of the play is +0.70 points ((2.6−2.2)+(1.5−1.3)+(1.7−1.6)). Of note is that the reduction in the scoring expectations of the opposing team, Team D, is considered favorable to the current offensive team, Team W. Baseball Play Valuation Example During a game between Team A and Team B being played in Team A's ballpark, Player JD of Team A comes up to bat in the third inning against Player MM. Player BM of Team A is on first base and there is one out in the inning. Based on this game state and from, for example, equation [1] above, the NES for this game state is 0.54 runs. During the at bat, Player JD hits a hard line drive just off the right field line resulting in a double. Player BM, who was on first base actually scores, and Player JD ends up on second base. As a result of the double by Player JD, the current game state is a man on second base with one out and one run scored. The NES for this game state is 1.64 runs (NES of 0.64 for the game state of a man on second base with one out +1 run scored). The value of the play is 1.1 runs (1.64 runs-0.54 runs). Distribution of Performance Points to Athletes Based on the Value of a Play In block 606, the sport server identifies the athletes who were involved in the play. The athletes may be identified in the XML play message. In block 608, the sport server distributes the value of the play calculated in block 604 amongst the identified athletes who were involved in the play as, for example, performance points or dividends generated by each share of a financial instrument in each identified athlete. In some embodiments, the value is distributed amongst the athletes according to relevant percentages as determined from an analysis of past events in the event pool using, for example, well known empirical and regression techniques. For example, for baseball, for a play involving a pitcher and a batter (e.g., strikeout, base on balls, or a home run), the empirical analysis of the past events may indicate that the pitcher is to be credited with 38% of the value of the play, and the batter is to be credited with 62% of the value of the play. Assuming the value of the play is 1.00 (e.g., a home run), the batter is credited with 0.62 performance points (a gain of 0.62 performance points) and the pitcher is debited 0.38 performance points (a reduction of 0.38 performance points). In the prior football example where the pass play from the quarterback to the halfback resulted in a value of +0.70 points, the empirical analysis of the past events may indicate that the quarterback and the halfback are each to be credited with 50% of the value of the play. Here, the quarterback and halfback are each credited with 0.35 performance points. One skilled in the art will appreciate that distribution percentages for other sports, other types of plays, and/or other athletes may be obtained in a similar manner by analyzing the events in the event pool for the particular sport. In block 706, the sport server determines the intermediate states that represent the flow of the play. The intermediate states may divide the play into steps or stages that isolate the individual contributions of the athletes involved in the play. For example, for baseball, if the play is a batted ball, the intermediate states may be: (1) the estimated result of the batted ball; (2) the actual result of the batted ball; (3) the estimated base running result due to the batted ball; and (4) the final result of the batted ball, including the actual base running that occurred. In block 708, the sport server calculates the value of each intermediate state determined in block 706. In a manner similar to calculating a value of an entire play, a value of an intermediate state may be calculated from the NES of the prior intermediate state and the NES of the current intermediate state. In the case where an intermediate state involves an expected result, the NES of the intermediate state may be calculated using relative probabilities of the possible or expected outcomes or results of the intermediate state. In some embodiments, the relative probabilities of the possible results may be calculated empirically from an analysis of the past events in the event pool. For example, for baseball, the hit characteristics of a batted ball may include power, direction, type, and distance, as follows:
A probabilistic model that calculates expected at bat and base running outcomes based on the batted ball hit characteristics may be created from an analysis of the events in the event pool. Table 2 below illustrates an example batted ball probabilistic model for line drives landing between 200-250 feet.
Batted ball probabilistic models that calculate the expected outcomes for the other combinations of the characteristics of the batted ball may similarly be created from an analysis of the events in the event pool. Table 3 below illustrates an example base runner advancement probabilistic model illustrating the probability of scoring from third base on a fly ball of a particular distance.
Base runner advancement probabilistic models that calculate the expected base runner advancement outcomes for the other combinations of the characteristics of the batted ball and base runner positions may similarly be created from an analysis of the events in the event pool. In some embodiments, batted ball probabilistic models and the base runner advancement probabilistic models may be generated for each of the different base ball stadiums or parks. In block 710, the sport server identifies the athletes who were involved in each of the intermediate states of the play, and distributes the value of each intermediate state amongst the athletes involved in the intermediate state as, for example, performance points or dividends generated by each share of a financial instrument in each identified athlete. One skilled in the art will appreciate that probabilistic models similar to those illustrated above for baseball may be similarly created for the various other sports such as football, basketball, hockey, golf, auto racing, etc. For example, in golf, one probabilistic model may indicate the probabilities of obtaining various scores on a particular par 4 hole after hitting a second shot into a green-side bunker. In auto racing, one probabilistic model may indicate the probabilities of obtaining various finish positions for various states in the race (e.g., driver is currently in third position, 5 seconds behind the second position car and 12 seconds behind the first position car on lap 100 of a 200 lap race, etc.). In block 802, the sport server calculates a NES—i.e., a NES0—at the beginning of the at bat based on the relevant inputs (i.e., the game state before the play occurs) into the NES equation [1]. In block 804, using the batted ball characteristics, the sport server determines an expectation for the at bat based on an expected new NES—i.e., NES1. For example, the sport server determines from the tables the relative probabilities of a single, double, triple, or out occurring based on factors, such as hit distance, hit type, hit power, hit angle, etc. These probabilities are calculated empirically from observed data over a predetermined number of seasons. The sport server uses these probabilities to calculate an expected new NES (NES1) based on a hit of this type. At this stage, the sport server does not consider what actually happened on the play, but rather merely calculates an expected new NES based on a batted ball of this type. The sport server subtracts NES0 from NES1 to determine a change in NES caused by this hit. The sport server may divide this change in NES between the pitcher and hitter—i.e., the athletes involved in this intermediate state of the play—according to relevant percentages as determined from the analysis of past events. In block 806, the sport server calculates a NES change based on the actual result of the at bat, absent of additional base advancement. For example, based on the outcome of the batted ball (i.e. single, double, triple or out) the sport server calculates a new NES—i.e., NES2. This NES change (i.e., NES2−NES1) does not take into account the actual new game state, but rather only the game state that would have been directly impacted by the result of the batted ball. For example, if a man is on first base and the batter hits a single, only the NES change associated with the player advancing to second base is captured in this calculation regardless of whether the runner actually advanced to second base. The sport server divides this change in NES between the fielder(s) involved and the hitter, in the hitter's role as a base runner, according to relevant percentages as determined from the analysis of past events. Depending on the type of play (e.g., base on balls, strike out, home run), this calculation may not be necessary. In block 808, the sport server determines the expectation for base runner advancement based on the location of the batted ball. The sport server determines an expected base running outcome based on the outcome of the batted ball and the context of the game situation. For example, based on the hit characteristics and the result of the hit (e.g., a single to right, fly out to left, etc), the sport server calculates probabilities of advancement and an expected new NES—i.e., NES3. The sport server subtracts NES2 (calculated in block 806 above) from NES3 to determine a change in NES, and may divide this change in NES between the pitcher and hitter as a characteristic of the hit, for example, according to relevant percentages as determined from the analysis of past events. In block 808, the sport server determines a final NES based on the actual result of the play. The sport server subtracts NES3 from the final NES to determine a change in NES, and may divide this change in NES between the hitter and the fielders involved in the play, for example, according to relevant percentages as determined from the analysis of past events. By way of an example to illustrate the NES calculation and distribution example of Assuming Player JD strikes out on the eighth pitch of the at bat, no advancement by Player BM is anticipated. A NES2 value of 0.29 runs is calculated based on there being two outs and a runner on first base. The change in NES is actually 0.25 runs (i.e., the difference between NES0 and NES1) and this amount of performance points may be split with, for example, 38% or 0.095 performance points being added to the pitcher, and 62% or 0.195 performance points being subtracted from Player JD. The percentage allocation is an example, and may be determined from an analysis of the historical data. Assuming Player JD hits a hard line drive just off the right field line and the ball lands less than 150 feet from the infield, the data in the probabilistic tables may indicate that a hit ball of this type has: a 65% chance of being an out; a 12% chance of being a single; a 20% chance of being a double; and a 3% chance of being a triple. Using these probabilities, the expected new NES is calculated based on a 65% chance of there being a man on first base with two outs (e.g., expected new NES for this probable outcome=0.20), a 12% chance of men on first and second base with one out (0.10), a 20% chance of men on second and third base with one out (0.21), and a 3% chance of one run scored and a man on third base (0.06). Summing up all these probable percentages produces a new NES1 of 0.57 runs (NES1, block 804). The change in NES caused by this batted ball is then 0.57−0.54 or 0.03 runs. This amount of performance points may then be split between the pitcher and Player JD, with Player JD receiving positive performance points and the pitcher receiving negative performance points. Continuing the example, Player JD's line drive may actually result in a double down the line. Based on a two base hit, the new NES2 is expected to be men on second and third base with one out, which is 1.07 runs (NES2, block 806). The difference between NES2 and NES1 is 1.07−0.57 or 0.5 runs. This amount of performance points may then be split between the fielder's in the play (e.g., the first baseman who had a chance to field the play and the right fielder who could have held Player JD to a single) and Player JD. Continuing the example, based on a double hit to that area of the field, it may be determined that on average the base runner advances to third base 80% of the time (0.86) and scores 18% of the time (0.31) and is thrown out 2% of the time (0.01). The estimated NES3 caused by the expected base running advancement is 1.18 runs (NES3, block 808). The difference between NES3 and NES2 is 1.18−1.07 or 0.11 runs, and this amount of performance points may be split between Player JD and the pitcher. Continuing the example, on the play, Player BM may actually score and Player JD may end up at second base. The final NES associated with a run in and a man on second base with one out is 1.64 runs (final NES, block 810). The difference between the final NES and NES3 is 1.64−1.18 or 0.46 runs, and this amount of performance points may be split between the base runner (i.e., Player JD) and the fielders involved with the play. One skilled in the art will appreciate that similar NES calculations and distributions can be made amongst the athletes involved in the intermediate states of a play for other sports such as football, basketball, hockey, soccer, etc. For example, the intermediate states for football may break down a football play into states which isolate passing and pass defended, running and run defended, etc. Market Value for Shares of Financial Instruments in Real-Life Athletes In some embodiments, the market maker adjusts the market value of a share of a financial instrument in response to the buy orders and the sell orders as they arrive. In order to gauge the demand for a share of a financial instrument, the market maker relies on the trade balance, which represents the ratio of the total value of buy orders to the total value of sell orders over a period of time, and may be calculated as:
In equation [4] above, p_{t }is the market value at time t, / is the number of periods over which the trade balance is calculated, and T is the current period. In some embodiments, the market maker attempts to keep the trade balance as close to one (1) as possible by adjusting the market value in response to changes in the trade balance. For example, if the trade balance is below one (i.e., the value of the sell orders is higher than the value of the buy orders over a set period of time) then the market maker reduces the market value, in order to attract more buy orders. In some embodiments, the market maker adjusts the market value by following log normal distributions which approximate stock returns. In order to incorporate the log normal distribution into the stock adjustment process, the market maker may incorporate Brownian motion into the adjustment process. Brownian motion implies that at any given point, a stock price may move up or down according to:
In the equations above, μ is the riskfree rate of return, dt is the length of time between transactions, and σ is the volatility of the stock. In some embodiments, the market maker may modify the standard Brownian motion to better capture to magnitude of the difference between the buy and sell orders. For example, the Brownian motion may be modified by multiplying the gross upward price adjustment by the square root of the trade balance ratio, and multiplying the gross downward price adjustment by the reciprocal of the trade balance ratio as depicted in equation [8] below:
In some embodiments, the market maker establishes an IPO price for a share of each financial instrument that is available for purchase by the users of the valuation system, such as the participants in the fantasy sports challenges or other users of the valuation system who request valuations for plays, athletes, and other services provided by the valuation system. For example, for each real-life athlete supported by the market maker, the market maker may use any of a variety of well-known statistical models to perform a historical analysis of the real-life performance of the athlete. An estimate of the performance points the athlete will generate over a predetermined period of time may be determined from the historical analysis of the real-life performance of the athlete. These estimates can incorporate the projected performance as well as the likelihood the athlete will still be in the professional leagues. The market maker may then choose a discount rate used for a discounted cash flow analysis. Using the estimated performance over the period of time, the market maker then calculates the total discounted performance flow by multiplying each future cash flow by a yearly discount factor which is calculated from the discount rate. In some embodiments, the market maker establishes a floor price for each share of a financial instrument. The floor price is a minimum amount that will be charged for a share of a financial instrument. For example, the market maker may establish a floor price, such as $10.00 or 10 fantasy money units. By setting a floor or minimum price, the market maker minimizes the phenomenon of a drastic price increase which may be caused by purchases of shares of a financial instrument that are drastically undervalued. Athlete Valuation Athletes generate performance points based on their on-field performance as measured by the valuation system using the play valuation techniques and the techniques for distributing the value of the play amongst the athletes as discussed above. In some embodiments, performance points are accrued each time an athlete has a measurable impact on the game. In general, performance points are a comparison of the athlete's performance in a given situation versus the baseline average in that situation. This methodology assigns a point value to each performance. Win Probability In some embodiments, the valuation system determines a win probability, which is a measure of the likelihood (e.g., probability) a player or particular team is to win a particular game given a current game state. The valuation system may then provide the win probability to various media outlets for display or broadcast, for example, within the context of the game. The valuation system may also display or broadcast the win probability through its own broadcast source, such as its web site. In block 1104, the games states corresponding to the events in the event pool are determined, for example, from the conditions that describe the events. In block 1106, the game states are analyzed to determine the results of the games in which the game states occurred. In block 1108, the game states and their corresponding game results are analyzed to derive one or more win probability equations suitable for determining the probability of a game participant in the game. At any point during a game, the value of the current situation can be determined using historic value analysis. This analysis is based upon events that have occurred in actual games in a particular sport over a predetermined time period in the past, such as, by way of example, three years. Each different situation, or game state, is analyzed to determine its value in any number of different metrics. One such value metric may be win probability (i.e., a measure of how likely a particular game participant is to win a particular game given the current game state). The number of events analyzed for a specific situation needs to be sufficient to provide an adequate event pool for deriving an equation that receives as inputs the factors that describe a game state and produces a value metric for the described game state. The measures of value that are determined from the analysis of the event pool and their corresponding game states may be analyzed using, for example, any of a variety of well-known regression techniques (e.g., linear regression, cubic regression, logistic regression, etc.) to derive an equation for that particular measure of value. The derived equation may include constants, coefficients, and/or variables whose values are based on the factors that describe the game state or statistics that imply the strength of the participants that are competing in the game. For example, analyzing the game states and their results in a baseball event pool using the above regression technique may generate the following win probability regression equation for baseball:
In the above win probability equation for baseball, the win probability is primarily dependent on score differential (i.e., how many runs the offensive team is ahead or behind) and the current inning. Win probability is also dependent, though less so, on NUMOUTS (the number of outs in the current inning), whether the team at bat is the home team, the position in the lineup of the current hitter, the league in which the game is played, whether or not the pitcher and hitter use the same hand, and the configuration of the runners on base (e.g., Runner on 1^{st}, Runner on 2^{nd}, and Runner on 3^{rd }all take the value of 1 if a runner is on that base and 0 otherwise), and the park that the game is being played in (e.g., each park has a correlation with the scoring of runs, which will in turn effect the chances of the offensive team winning or losing given their current situation). The values for the above variables are determined based on the factors that describe a game state, and inputting values for the variables as determined from the factors that describe the game state into equation [9] produces a win probability for the particular game state. Table 4 below illustrates example win probability outcomes generated using equation [9] for a particular combination of factors and the listed game states or situations (i.e., combination of the score differential and the inning) for a particular baseball stadium.
Similar tables can be generated for the other combinations of the factors and for the other baseball stadiums. In some embodiments, one or more of the variables (i.e., factors) listed above may be omitted in calculating the win probability. For example, in a particular football game between Team X and Team Y in Team Y's football stadium, Team X has possession of the ball and leads 24-17. It is first down with ten yards to go, and the ball is at Team X's 20 yard line. It is in the fourth quarter with 11:05 on the clock. Given this current game state and from the analysis of the past events, Team X has a 78.0% chance of winning the game. During this first down play, Player A, a running back for Team X, runs the ball eight yards. At the end of the play, it is now second down with two yards to go from the 28 yard line. This is a positive result for Team X, and given this current game state and from the analysis of the past events, Team X now has a 79.2% chance to win the game. 79.2% is the win probability for this new game state. Therefore, the value of the play—i.e., the total effect of the play—is +1.2 basis points (i.e., 79.2−78.0). In another example, in a particular baseball game between Team W and Team V in Team V's baseball stadium, Player A for Team W comes to the plate in the third inning of a tie game with Player P pitching for Team V. There are no runners on base, and there is nobody out in the inning, and Team W has a 1-0 lead. Given this game state and from the analysis of the past events, Team W has a 61.3% chance to win the game. 61.3% is the win probability for this game state. During this at bat—i.e., play,—Player A hits a single. At the end of the play, Player A is now on first base and there are still no outs in the inning. This is a positive result for Team W, and given this current game state and from the analysis of the past events, Team W now has a 62.1% chance to win the game. 62.1% is the win probability for this new game state. Therefore, the value of the play—i.e., the total effect of the play—is +0.8 basis points (i.e., 62.1—61.3). In some embodiments, the value metric that is determined using historic value analysis may be a ranking of the team relative to other teams. The ranking of the team relative to other teams may be applicable to, for example, college football. In these instances, the variables might include not only in-game factors such as score, down and distance, and time remaining, but also win-loss record coming into the game, strength of schedule, previous ranking, conference affiliation, etc. In some embodiments, the probability that a team is going to win a particular game is dependent upon factors such as the score of the game as well as several variables that are particular to each sport. For example, in football, the probability that a team is going to win may be dependent upon variables such as, the points they have scored, the points they have allowed, whether they have possession of the ball, time remaining, down and distance, etc. These game state variables are particular to each sport, but the style of analysis to determine the win probability for sports other than football is similar. Some of the game state variables are related to the skill of the teams on the field (e.g., yards allowed) and some are not (e.g., score and time remaining). The win probability can be calculated on a team specific or team independent basis. Prior to any game, the odds that one team is going to win is rarely even. The actually probability of winning may be estimated based upon known variables such as the winning percentage for each team prior to the game, which team is at home, which team has scored more points in the season, which team has allowed more points, etc. These variables may be thought of as season state variables. These calculations are performed on a team specific basis, as the team independent win probability, prior to the start of a game, is equal for all teams or tournament participants. The same type of analysis may be used to calculate the probability that a team is going to win a series of games. For example, in baseball, the World Series is a best of seven series. Each strikeout and each homerun affect not only the probability that a team will win a given game, but also the probability that they will win the entire series. In some embodiments, the statistical analysis may be performed using binary regression techniques in which a win is coded as a “1” and a loss is coded as a “0.” The independent variables are the game and season state variables. For team independent analysis, only the non team specific independent variables may be used. Play by play data for the game state variables may create a large data set for the in game estimates. Sports with longer seasons may allow for larger data sets for the pre game estimates, but using data from previous seasons and adjusting for offseason changes allows for pre game estimates in sports such as football that have a short season and, thus, more volatile estimates of the season state variables. For example, for football, game state variables may include score, quarter, time remaining, team with possession, down, distance to go, passing yards, rushing yards, and valuation system statistics. Of these, passing yards, rushing yards and valuation system statistics are team specific variables. Season variables may include winning percentage, rushing yards, passing yards, valuation system statistics, coaching variables (e.g., years coached, lifetime win percentage, etc.), injuries, and home team. During playoff time, the analysis may be extended to include the probability that the team is going to win the championship (e.g., Super Bowl) based on their probability of winning the playoff game and the potential opponents. This analysis may include a mix of game and season state variables. For example, for baseball, the game state variables may include score, team at bat, runners on base, lineup position, inning, pitch count, pitcher (e.g., specific pitcher, starter vs. reliever, etc.), home team, league, and valuation system statistics. The season variables may include winning percentage, valuation system statistics, injuries, expected starting pitchers, and home team. In the playoffs, a combination of game state and season state variables may be used to calculate the probability a team is going to win the current series and/or the World Series. For example, for basketball, the game state variables may include score, quarter, time remaining, fouls, and valuation system statistics. The season variables may include winning percentage, home team, injuries, coaching variables, consecutive games, and road trip. In the playoffs, a combination of game state and season state variables may be used to calculate the probability a team is going to win the current series and/or the championship (e.g., NBA Championship). For example, for golf, the tournament state variables may include score, strokes behind, number of golfers ahead, hole, par on current hole, day of the tournament, and golfers remaining to play. The pre-tournament state variables may include total winnings, winning of the rest of the field, yardage of the course, and average score. In match play tournaments, tournament and pre-tournament variables may be used to estimate the probability of winning the entire tournament. For example, for tennis, the match state variables may include game score, set score, serve, faults, and serve velocity. The pre-match state variables may include ranking, opponent's ranking, major tournament, and head-to-head matchup history. Match state and pre-match state variables, along with related data on potential opponents, may be used to calculate the probability of winning the entire tournament. From the foregoing, it will be appreciated that embodiments of the invention have been described herein for purposes of illustration, but that various modifications may be made without deviating from the spirit and scope of the invention. Accordingly, the invention is not limited except in accordance with elements explicitly recited in the appended claims. Referenced by
Classifications
Legal Events
Rotate |