Match Simulation

Introduction

This vignette covers bouncer’s Monte Carlo simulation capabilities. By simulating matches thousands of times, you can: - Estimate win probabilities with uncertainty ranges - Understand the distribution of possible outcomes - Simulate entire seasons to calculate playoff odds

Why Simulate?

A single prediction gives one number (e.g., “65% win probability”). Simulation gives you the full distribution: not just who wins, but by how much, and how often various scenarios occur.

Simulate a single match many times:

library(bouncer)

# Simulate India vs Australia 10,000 times
results <- simulate_match_ballbyball(
  team1 = "India",
  team2 = "Australia",
  format = "t20",
  n_sims = 10000
)

# View summary
print(results)

The results include: - Win percentage for each team - Score distribution (mean, median, percentiles) - Margin distribution

Quick Match Simulation

For faster results with less detail:

# Quick simulation (doesn't simulate ball-by-ball)
quick_results <- quick_match_simulation(
  team1 = "India",
  team2 = "Australia",
  format = "t20"
)
print(quick_results)

Simulation Configuration

Customize simulation parameters:

# Create custom simulation config
config <- create_simulation_config(
  n_sims = 5000,
  seed = 42,  # For reproducibility
  format = "t20"
)

# Use in simulation
results <- simulate_match_ballbyball(
  team1 = "Mumbai Indians",
  team2 = "Chennai Super Kings",
  format = "t20",
  config = config
)

Innings Simulation

Simulate a single innings:

# Simulate an innings
innings_result <- simulate_innings(
  batting_team = "India",
  bowling_team = "Australia",
  format = "t20",
  target = NULL  # First innings, no target
)
print(innings_result)

# Simulate a chase
chase_result <- simulate_innings(
  batting_team = "Australia",
  bowling_team = "India",
  format = "t20",
  target = 185
)
print(chase_result)

Season Simulation

Simulate an entire tournament:

# Simulate IPL season
season_results <- simulate_season(
  event = "Indian Premier League",
  season = "2024",
  n_sims = 1000
)

# View playoff probabilities
print(season_results$playoff_odds)

Running Multiple Seasons

For more accurate playoff odds:

# Run 10,000 season simulations
season_n <- simulate_season_n(
  event = "Indian Premier League",
  season = "2024",
  n_sims = 10000
)

# Each team's probability of:
# - Making playoffs
# - Finishing top 2
# - Winning the tournament
print(season_n$summary)

IPL Playoff Simulation

Simulate the IPL playoff bracket specifically:

# Simulate playoffs with the top 4 teams
playoff_results <- simulate_ipl_playoffs(
  qualifier1_teams = c("Gujarat Titans", "Chennai Super Kings"),
  eliminator_teams = c("Lucknow Super Giants", "Mumbai Indians"),
  n_sims = 10000,
  format = "t20"
)

# Championship probabilities
print(playoff_results)

Interpreting Results

Win Probability Confidence

From 10,000 simulations: - If Team A wins 6,500 simulations, we’re confident they win ~65% of the time - The uncertainty is roughly ±1% for 10,000 sims

Score Distribution

# Example output interpretation
# Mean score: 165.3
# Median score: 164
# 10th percentile: 142 (bad day)
# 90th percentile: 188 (good day)

Margin Distribution

Simulations also show likely winning margins: - Median winning margin - Probability of close games (< 10 runs or < 2 wickets) - Probability of blowouts

Setting Seeds for Reproducibility

# Same seed = same results
set.seed(42)
result1 <- simulate_match_ballbyball("India", "Australia", "t20", n_sims = 100)

set.seed(42)
result2 <- simulate_match_ballbyball("India", "Australia", "t20", n_sims = 100)

# result1 and result2 are identical

Computational Considerations

Simulations	Time (typical)	Use Case
100	< 1 second	Quick check
1,000	~5 seconds	Reasonable estimate
10,000	~30 seconds	Publication quality
100,000	~5 minutes	High precision research

Match Simulation

Introduction

Why Simulate?