Lab 09: AdX Two Day

This lab introduces multi-day advertising campaigns with strategic bidding across days.

Game Overview

Type: Multi-day real-time bidding advertising game Players: 2+ players Rounds: Two days with strategic bidding decisions Stages: Multi-stage with day 1 and day 2 campaigns

Games

AdX Two Day Game

Actions: Bid bundles for each day with strategic planning
State Space: Campaign information for both days
Key Concept: Multi-day strategic bidding and campaign planning

State Space

Observations

observation = {
    "day": 1,
    "campaign_day1": {
        "id": 1,
        "market_segment": "MALE_YOUNG_HIGH_INCOME",
        "reach": 500,
        "budget": 500.0
    },
    "campaign_day2": {
        "id": 2,
        "market_segment": "FEMALE_OLD_LOW_INCOME",
        "reach": 300,
        "budget": 300.0
    }
}

Actions

action = {
    "day": 1,
    "campaign_id": 1,
    "day_limit": 500.0,
    "bid_entries": [
        {
            "market_segment": "MALE_YOUNG_HIGH_INCOME",
            "bid": 2.5,
            "spending_limit": 100.0
        }
    ]
}

Rewards

Multi-day campaign performance:

# Day-specific profit calculation
reach_fulfilled = min(day_impressions, campaign.reach)
day_profit = (reach_fulfilled / campaign.reach) * campaign.budget - day_spent
reward = day_profit

Game Structure

Stage Type

Two-day simulation with strategic bidding decisions
Day 1 and Day 2 campaigns with different parameters
Strategic planning - balance performance across days

Learning Opportunities

Multi-day optimization - plan bidding across both days
Strategic allocation - distribute budget between days
Campaign coordination - optimize overall performance

Testing

Local Testing

from core.engine import Engine
from core.game.AdxTwoDayGame import AdxTwoDayGame
from core.agents.lab09.random_agent import RandomAgent

my_agent = MyAgent("MyAgent")
opponent = RandomAgent("Random")

engine = Engine(AdxTwoDayGame(num_players=2), [my_agent, opponent], rounds=2)
results = engine.run()

print(f"My total score: {results[0]}")
print(f"Opponent total score: {results[1]}")

Multi-day Analysis

def analyze_multi_day_performance(self):
    if hasattr(self, 'day_performance'):
        for day, performance in self.day_performance.items():
            print(f"Day {day}:")
            print(f"  Impressions: {performance['impressions']}")
            print(f"  Reach rate: {performance['reach_rate']:.2%}")
            print(f"  Profit: {performance['profit']:.2f}")

Next Steps

Implement a multi-day AdX agent using the common patterns
Study multi-day optimization to understand strategic planning
Test different bidding approaches against various opponents
Compete against other students

Focus on understanding multi-day strategic bidding and campaign coordination!