Lets-Go-Gambling

A research project exploring optimal Blackjack strategies using Monte Carlo Tree Search (MCTS) and Markov Decision Processes (MDP).

Overview

This project implements and analyzes advanced AI techniques for Blackjack, focusing on decision-making and strategy optimization. The agent leverages MCTS and MDPs to learn and play Blackjack, with results compared to standard baselines.

Features

• Monte Carlo Tree Search (MCTS) agent for Blackjack
• Markov Decision Process (MDP) modeling of the game
• Simulation and analysis of agent performance
• Playable Blackjack game engine
• Data and results for in-depth analysis

Project Structure

• main.py: Run simulations and adjust agent parameters.
• Helpers/AgentHelper.py: Core MCTS algorithm and agent logic.
• Helpers/DeckHelper.py: Deck and card utilities.
• Helpers/MetricsHelper.py: Metrics and evaluation tools.
• misc/Game_Engine.py: Playable Blackjack game.
• Simulation_Analysis.ipynb: Data analysis and visualization.
• data/: Simulation data and results.
• Blackjack_Simulation_with_MDP_MCTS.pdf: Full project report.

Getting Started

1. Install dependencies:

   pip install -r requirements.txt

2. Run a simulation

   python main.py
   

3. Play the game manually

   python misc/Game_Engine.py