Monopoly Agent Strategy Simulation

1
Monopoly Agent Strategy Simulation

• Nicholas Loffredo

2
Overview

• My project is to make a simulation of Monopoly
  with agents that can negotiate with each other,
  buy titles and houses/hotels, mortgage
  properties, etc. and find the optimal strategy
  for Monopoly.

3
Current State

• Computer buys/mortgages/unmortgages properties
  and buys/sells houses based on aggressiveness
  levels
• Monopoly, Matrix, Wins Display
• Mostly finished Monopoly rules/game
• Computer changes aggressiveness levels
• Multiple games in one run
• Read in aggressiveness values and wins from text
  file
• Agents trades properties for cash
• Agents auction properties

4
(No Transcript)
5
(No Transcript)
6
Results

• Player 0 (learning agent) is learning
  significantly
• Limited due to a number of factors, most likely
• Structure of aggressiveness chart
• Inflexible
• Does not allow for factoring in state of game,
  number of monopolies, etc.

7
Further Research

• Allow trading properties for other properties
• Fix a bug with trading/auctioning
• Test different structures of reinforcement
  learning
• Test an expert systems based AI vs. my
  reinforcement learning AI
• Attempt to implement aggressiveness chart idea
  for other games