Submission Type

Presentation

Submission Title

Monte Carlo simulations of players on the slot floor: segmentation of slot machines and classification of players by historical behavior

Session Title

Session 1-1-D: Data and Player Behavior

Location

Caesars Palace, Las Vegas, Nevada

Start Date

28-5-2019 9:15 AM

End Date

28-5-2019 10:40 AM

Disciplines

Artificial Intelligence and Robotics | Gaming and Casino Operations Management | Numerical Analysis and Scientific Computing

Abstract

Player tracking systems collect vast quantities of data that can be used to study how players utilize and move between machines on a slot floor. These data can be used to build a Markov transition matrix for each segment of the player population, which describes the probability that a player occupying a slot machine will move to any other machine on the floor. The size of the transition matrix can be decreased and the statistical noise improved by segmenting slot machines into distinct types, based on characteristics such as denomination, volatility, manufacturer, game type, theme, etc., as well as location on the slot floor. Such a transition matrix can then be used to generate realistic Monte Carlo simulations. We will present simulations based on anonymized slot floor data, in which we investigate, characterize, and classify, and visualize player behavior. We will use artificial intelligence methods to solve a large optimization problem over a space of simulations, in which we minimize a statistical metric that quantifies segmentation error. In other words, we will search for the optimal way to segment both players, based on their observed behavior, and slot machines, based on their most important features determined from their tabulated characteristics.

Keywords

slot floor simulations segmentation optimization artificial intelligence

Author Bio

Dr. Jason Fiege is CEO of nQube Data Science Inc. and Associate Professor of astrophysics at the University of Manitoba. He is a scientific computing, data modeling, and optimization expert with over 20 years experience. He is the architect of nQube’s AI-guided optimization and data modelling platform, and leads their research in slot floor optimization, AI-based segmentation algorithms, predictive AI systems for time series analysis, and optimization of trading strategies in financial markets.

Funding Sources

This research is privately funded by nQube Data Science Inc., which is partially owned by the author, who is also CEO of the company. The research question addressed is of a curiosity-driven and exploratory nature, and not directly related to our current products or business activities.

Competing Interests

No competing interests beyond the company involvement noted above.

Share

COinS
 
May 28th, 9:15 AM May 28th, 10:40 AM

Monte Carlo simulations of players on the slot floor: segmentation of slot machines and classification of players by historical behavior

Caesars Palace, Las Vegas, Nevada

Player tracking systems collect vast quantities of data that can be used to study how players utilize and move between machines on a slot floor. These data can be used to build a Markov transition matrix for each segment of the player population, which describes the probability that a player occupying a slot machine will move to any other machine on the floor. The size of the transition matrix can be decreased and the statistical noise improved by segmenting slot machines into distinct types, based on characteristics such as denomination, volatility, manufacturer, game type, theme, etc., as well as location on the slot floor. Such a transition matrix can then be used to generate realistic Monte Carlo simulations. We will present simulations based on anonymized slot floor data, in which we investigate, characterize, and classify, and visualize player behavior. We will use artificial intelligence methods to solve a large optimization problem over a space of simulations, in which we minimize a statistical metric that quantifies segmentation error. In other words, we will search for the optimal way to segment both players, based on their observed behavior, and slot machines, based on their most important features determined from their tabulated characteristics.