Participant utility extraction for prediction market based on region of difference between probability functions
View Patent ↗The present disclosure provides a system that implements a prediction market, used for making forecasts and, more particularly, for extracting participant utility. One or more forecasts of a specific participant or subset of participants are taken, and a measure of divergence of these forecasts from a group as a whole (or for that matter, any group) is calculated. This divergence may then later be employed in the prediction market (e.g., for present or future forecasting). For example, if it is known that “sales managers” are consistently too liberal in forecasting product release dates than others within a company, this tendency may be compensated for or used in a variety of manners, such as automatically correcting predictions or forecasts for perceived error, automatically discarding or modifying predictions, or automatically generating “virtual” predictions for future events. Other actions may also be taken.
1. A method of conducting a prediction market that obtains prediction market wagers from respective individuals, each wager representing belief of the respective individual as to the outcome of an event, and of extracting a measure of participant utility utilizing the prediction market wagers, the method comprising:
receiving data reflecting a first aggregate probability distribution representing a group forecast for the event, including receiving prediction market forecasts from respective individuals, each prediction market forecast reflecting a forecast from the respective individual for the event, and mathematically combining the data representing the prediction market forecasts from the respective individuals to obtain the data representing the first aggregate probability distribution;
receiving data reflecting a second probability distribution representing a forecast of a participant for the event, where the participant forecast is placed as a prediction market wager priced in a manner dependent on the first aggregate probability distribution at the time of the prediction market wager by the participant;
updating the first aggregate probability distribution responsive to the participant forecast;
using a microprocessor based system to calculate a region of difference between a curve representing the second probability distribution and a curve representing the first aggregate probability distribution;
determining the measure of participant utility based on the region of difference; and
performing at least one of applying the measure of participant utility to a subsequent prediction market computation or formatting the difference data into a browser-readable page for output to indicate a measure of utility of the participant.
2. The method of claim 1 , where:
the method further comprises storing the measure in a database; and
the performing includes applying the measure to a subsequent prediction market computation, including in dependence on the measure, at least one of (a) adjusting a future forecast from the participant, (b) creating a virtual forecast from the participant for another event, (c) excluding the participant from a future prediction market, and (d) discarding a forecast from the participant for another event.
3. The method of claim 2 , where storing the measure includes storing a time stamp representing time of the participant forecast.
4. The method of claim 2 , where the database is a first database of a relational database, and where the method further comprises using a second database of the relational database, the second database identifying at least one affiliation for the participant.
5. The method of claim 4 , further comprising aggregating measures for multiple participants based on affiliations from the second database, to develop an aggregated measure representing a participant subset, the participant subset representing the multiple participants.
6. The method of claim 1 , where the method further comprises generating an aggregate forecast based on the updating of the first aggregate probability distribution, and providing data representing the aggregate forecast to at least one recipient for display via a web page.
7. The method of claim 1 , further comprising revealing information reflecting the group forecast to the participant prior to the placing of the prediction market wager representing from the participant.
8. The method of claim 1 , where:
the group forecast represents a first group;
the method further comprises repeating the receiving of data representing a participant forecast for each of plural different participants representing a second group;
using the microprocessor based system and determining the measure are performed such that the measure represents divergence of forecasts of the second group relative to the first aggregate probability distribution;
the method further comprises storing the measure representing divergence of forecasts of the second group relative to the first aggregate probability distribution.
9. The method of claim 8 , further comprising restricting access to the measure representing divergence of forecasts of the second group conditioned upon authentication from a security database.
10. The method of claim 8 , where performing comprises generating a report conditioned on receiving authorization to generate the report from a security database, the report dependent on the measure representing divergence of forecasts of the second group.
11. The method of claim 1 , where:
receiving the data representing the first aggregate probability distribution includes at least one of receiving the data in a manner representing a first probability model type or converting received data to the first probability model type; and
receiving the data representing the second probability distribution includes at least one of receiving the data in a manner representing the first probability model type or converting received data to the first probability model type.
12. The method of claim 11 , where the first probability model type is a student's-T distribution.
13. The method of claim 11 , where using the microprocessor based system to calculate the region of difference of the second probability distribution relative to the first aggregate probability distribution includes determining crossing points between the second probability distribution and the first aggregate probability distribution.
14. The method of claim 11 , where using the microprocessor based system to calculate the region of difference of the second probability distribution relative to the first aggregate probability distribution includes determining an area where the second probability distribution exceeds the first aggregate probability distribution.
15. The method of claim 11 , where using the microprocessor based system to calculate the region difference of the second probability distribution relative to the first aggregate probability distribution includes subtracting at least part of (a) one of the first aggregate probability distribution or the second probability distribution, from (b) the other of the first aggregate probability distribution or the second probability distribution.
16. The method of claim 5 , where the second database includes data that identifies a job function for each respective one of the multiple participants.
17. The method of claim 16 , embodied as a method of forecasting at least one event associated with an enterprise, where each of the multiple participants is an employee of the enterprise and where the second database identifies enterprise personnel.
18. A prediction market apparatus adapted to obtain prediction market wagers from respective individuals, each wager representing belief of the respective individual as to the outcome of an event, and to extract a measure of participant utility, the prediction market apparatus comprising:
an interface to receive data representing a group forecast for the event, and an interface to receive data representing a participant forecast for the event, where the group forecast is dependent on prediction market forecasts from multiple individuals, that each reflect a forecast from a respective one of the multiple individuals for the event, where the group forecast is expressed as a first aggregate probability distribution that represents a mathematical combination of data representing the prediction market forecasts to obtain the group forecast, and where the participant forecast is placed as a prediction market wager priced in a manner dependent on the first aggregate probability distribution;
a calculation engine to
update the first aggregate probability distribution responsive to the participant forecast for the event, and
calculate a region of difference between a curve representing the first aggregate probability distribution and a curve representing a second probability distribution corresponding to the participant forecast for the event; and
an application engine to perform at least one of applying the measure of participant utility to a subsequent prediction market computation or providing a least one of viewing or reporting of a measure of utility of the participant;
where
the calculation engine is implemented by a microprocessor-based system, and
the measure of utility of the participant is determined dependent on the region of difference.
19. The prediction market apparatus of claim 18 , where the interface to receive data representing the participant forecast includes a tool to obtain the prediction market wager from the participant.
20. The prediction market apparatus of claim 18 , where the interface to receive data representing the group forecast includes a tool to obtain forecast data from each of multiple individuals representing the respective individual's forecast for the event and where the calculation engine is to convert the forecast data for the multiple individuals into the first aggregate probability distribution.
21. The prediction market apparatus of claim 20 , where the application engine is to provide at least one of viewing or reporting of data representing the first aggregate probability distribution as updated.
22. The prediction market apparatus of claim 18 , where the calculation engine includes code to convert at least one of the data reflecting the group forecast or the data representing the participant forecast to a common probability distribution type shared by both the first aggregate probability distribution and the second probability distribution.
23. The prediction market apparatus of claim 22 , where the common probability distribution type is a student's T distribution, and where the calculation engine is to combine a first set of data identifying a first student's T distribution representing to the group forecast, with a second set of data identifying a second student's T distribution representing to the participant forecast.
24. The prediction market apparatus of claim 18 , where:
the prediction market apparatus further comprises storage to store data representing the first aggregate forecast as updated; and
the application engine is further to provide a least one of viewing or reporting of the aggregate forecast as updated using the data stored by the storage.
25. The prediction market apparatus of claim 24 , where:
the prediction market wager is a first prediction market wager;
the apparatus further comprises a storage to store at least a time stamp identifying the time of the participant forecast;
the interface to receive data responsive to the participant forecast includes a tool to obtain the data representing the participant forecast as the first prediction market wager;
the calculation engine is to price the first prediction market wager dependent upon the group forecast at a time indicated by the time stamp; and
the calculation engine is to price a second prediction market wager subsequent to the first prediction market wager in a manner dependent upon the group forecast and the first prediction market wager.
26. The prediction market apparatus of claim 18 , embodied as a computer with instructions that are stored on machine readable media.
27. The prediction market apparatus of claim 18 , where the calculation engine and the application engine are embodied as instructions stored on machine readable media.
28. The prediction market apparatus of claim 18 , embodied as plural network machines including at least one server.
29. An apparatus comprising instructions stored on non-transitory machine-readable media, the apparatus for use with a machine that is to construct a prediction market that obtains prediction market wagers from respective individuals, each wager representing belief of the respective individual as to the outcome of an event having an interface to receive data representing a group forecast for an event, and that is to extract a measure of participant utility utilizing the prediction market wagers, the apparatus having an interface to receive data representing to a participant forecast for the event, the participant forecast being placed as a prediction market wager priced in a manner dependent on a first aggregate probability distribution representing the group forecast, the instructions when executed by a machine to cause the machine to:
calculate a region of difference between a curve representing the first aggregate probability distribution and a curve defining a second probability distribution representing the participant forecast for the event;
store difference data dependent on the region of difference in storage readable by the machine;
update the first aggregate probability distribution responsive to the participant forecast; and
perform at least one of applying the measure of participant utility to a subsequent prediction market computation or format the difference data into a browser-readable page for output to indicate a measure of utility of the participant.
30. An improvement in a system for conducting a prediction market that obtains prediction market trades from respective individuals, each trade representing belief of the respective individual as to the outcome of an event, that combines the prediction market trades for the event to obtain a first aggregate probability distribution representing group belief as to outcome of the event, and that generates a forecast in dependence on the first aggregate probability distribution, where the system includes circuitry and software to interact with each respective individual to establish a price associated with the respective prediction market trade and to confirm a decision by the respective individual to place the prediction market trade subject to the price, the improvement comprising:
for at least one new prediction market trade from a participant, computing utility of a participant making the at least one new prediction market trade, by
converting data representing the given prediction market trade to a second probability function, and
determining a region of difference between a curve representing the second probability function and a curve representing the first aggregate probability distribution dependent on at least one of (i) points where the first aggregate probability distribution crosses the second probability function, (ii) a result of subtracting the first aggregate probability distribution and the second probability function, or (iii) a difference in area between the curve representing the first aggregate probability distribution and the curve representing the second probability function;
updating the first aggregate probability distribution responsive to the new prediction market trade from the participant; and
performing least one of applying the measure of participant utility to a subsequent prediction market computation or formatting the difference data into a browser-readable page for output to indicate a measure of utility of the participant.
31. The improvement of claim 30 , where determining difference includes determining difference using each of (i), (ii) and (iii).
32. The apparatus of claim 18 , where:
the apparatus further comprises a database; and
the at least one comprises one or more of (a) adjusting a future forecast from the participant, (b) creating a virtual forecast from the participant for another event, (c) excluding the participant from a future prediction market, and (d) discarding a forecast from the participant for another event.