Game Simulation

Run complete game simulations for testing and analysis.

Tools

run-game - Full Game Simulation

Simulates complete prediction market games with agents, betting, and outcomes.


# Single game
bun run command:simulate
 
# With specific outcome
bun run command:simulate --outcome=YES
 
# Batch mode (100 games)
bun run command:simulate:batch
 
# Save results
bun run command:simulate --save=game.json --verbose

generate-world - Narrative Generation

Generates complete game narratives with NPC actions and events.


# Default (SUCCESS outcome)
bun run command:world
 
# Specific outcome
bun run command:world --outcome=FAILURE
 
# Verbose narrative
bun run command:world:verbose
 
# Save to file
bun run command:world --save=world.json

run-game Options

Flag	Description	Example
`--outcome=YES\|NO`	Set predetermined outcome	`--outcome=YES`
`--count=N`	Run N simulations (batch mode)	`--count=100`
`--save=file`	Save game data to JSON	`--save=game.json`
`--fast`	Skip verbose logging	`--fast`
`--verbose`, `-v`	Show all events	`--verbose`
`--json`	Output JSON only	`--json`

generate-world Options

Flag	Description	Example
`--outcome=SUCCESS\|FAILURE`	Set predetermined outcome	`--outcome=SUCCESS`
`--save=file`	Save world to JSON	`--save=world.json`
`--verbose`, `-v`	Show all events	`--verbose`
`--json`	Output JSON only	`--json`

Use Cases

Testing Game Mechanics


# Run single game with full logging
bun run command:simulate --verbose

View:

Agent decision-making
Betting patterns
Market dynamics
Winner calculation

Batch Analysis


# Run 1000 games for statistics
bun run command:simulate --count=1000 --fast --json > results.json
 
# Analyze with jq
cat results.json | jq '.results[] | {outcome, winners}'

Narrative Testing


# Generate complete world narrative
bun run command:world:verbose
 
# See full event timeline:
# DAY 1
# Actor1: Announces AI safety initiative
# Actor2: Responds with skepticism
# News: Tech Summit Reveals Surprising Consensus
# ...

Performance Benchmarking


# Measure game simulation performance
time bun run command:simulate:batch
 
# Output:
# 100 games completed
# { totalTime: '45s', avgTime: '450ms/game' }

Output Formats

Console (Default)

Human-readable with colors:


BABYLON GAME SIMULATION
==========================
Question: Will AI regulation pass?
Predetermined Outcome: YES
Agents: 5
 
Agent1: Bet YES (100 tokens)
Agent2: Bet NO (50 tokens)
Market: 65% YES / 35% NO
 
Outcome revealed: YES
Winners: Agent1, Agent3
Game complete { duration: '2.5s', events: 145 }

JSON Output

Machine-readable:


bun run command:simulate --json


{
 "id": "game123",
 "question": "Will AI regulation pass?",
 "outcome": true,
 "agents": [
 {"id": "agent1", "tokens": 200},
 {"id": "agent2", "tokens": 50}
 ],
 "market": {
 "yesOdds": 65,
 "noOdds": 35
 },
 "events": [...],
 "winners": ["agent1", "agent3"]
}

File Export


bun run command:simulate --save=game-$(date +%s).json
# Creates: game-1698765432.json

Simulation Components

Game Simulation (run-game)

Market Init - Starting odds (50/50)
Agent Deploy - 5 agents with initial tokens
Clue Distribution - Agents receive clues over time
Agent Betting - Agents bet based on clues
Market Updates - Odds adjust based on bets
Outcome Reveal - Final outcome shown
Winner Calc - Tokens distributed to winners

World Generation (generate-world)

Question Setup - Define prediction question
NPC Cast - Create 8 NPCs with roles
Timeline Sim - Simulate 30 days of events
Event Generation - Generate actions, conversations, news
Feed Posts - Create social media posts
Outcome Reveal - Final outcome announcement

Performance

run-game (Single)

Duration: 1-3 seconds
Events: 100-200
Memory: ~50MB

run-game (Batch)

Duration: ~500ms per game
100 games: ~50 seconds
Memory: ~100MB

generate-world

Duration: 5-15 seconds (depending on verbosity)
Events: 400-600
Memory: ~75MB

Testing Patterns

Outcome Balance


# Run 100 games, check 50/50 split
bun run command:simulate:batch | grep "outcomes"
# yesOutcomes: 50 (50%)
# noOutcomes: 50 (50%)

Agent Behavior


# Run single game, observe betting
bun run command:simulate --verbose 2>&1 | grep "Bet"
# Agent1: Bet YES (100 tokens)
# Agent2: Bet NO (50 tokens)

Market Dynamics


# Export game data, analyze market
bun run command:simulate --json --save=market-test.json
cat market-test.json | jq '.market'