bigdance A Python package for NCAA March Madness bracket simulation combining real-time ratings with customizable tournament simulations.

Overview

bigdance is a comprehensive Python package for simulating NCAA basketball tournament brackets. It provides tools for:

• Pulling real-time college basketball team ratings and matchups from Warren Nolan
• Creating and simulating hypothetical tournament brackets with in-season rankings
• Extracting and simulating real bracket pools from ESPN's Tournament Challenge
• Simulating tournament outcomes with adjustable "upset factors"
• Analyzing bracket pools to determine winning strategies
• Visualizing results and generating insights on optimal bracket selection
• Analyzing the importance of specific games in a tournament

Whether you're a fan looking to improve your bracket picks, a data scientist analyzing tournament patterns, or a researcher studying sports predictions, bigdance offers powerful, customizable tools to help you simulate and analyze the Big Dance of March Madness.

Installation

pip install bigdance

Quick Start

From the command line:

# Analyze a bracket pool from ESPN, pool ID found in the URL after "bracket?id="
bigdance espn --pool_id 77268ce6-7989-4e01-97dc-6681c63c6890

Example output:

                     name  avg_score  std_score win_prob
                    Tyler    137.760  21.439824    47.2%
Taylor's Educated Guesses    108.968  21.589281    28.3%
     Marclemore's Picks 1    124.376  20.652679    12.2%
        dsutt06's Picks 1    124.376  20.652679    12.2%
   Crazylegs329's Picks 1     93.536  20.058990     0.0%
     KyleStokes's Picks 1    107.376  20.652679     0.0%
        ddehart's Picks 1    120.760  21.439824     0.0%
      trev_wood's Picks 1    120.376  20.652679     0.0%

You can also run detailed game importance analysis:

bigdance espn --pool_id 77268ce6-7989-4e01-97dc-6681c63c6890 --importance

Example output:

=== GAME IMPORTANCE SUMMARY ===

GAME #1: Auburn vs Florida (Region: SOUTH)
  Max Impact: 0.4470 | Avg Impact: 0.1250
  Most affected entry: Tyler
    Win chances: 66.6% if Auburn wins vs 21.9% if Florida wins
    Currently at: 45.1% baseline win probability
    Difference: 44.7%

GAME #2: Duke vs Houston (Region: EAST)
  Max Impact: 0.5270 | Avg Impact: 0.1318
  Most affected entry: Taylor's Educated Guesses
    Win chances: 52.7% if Houston wins vs 0.0% if Duke wins
    Currently at: 29.9% baseline win probability
    Difference: 52.7%

=== END OF SUMMARY ===

Interactive Demo

A live web app is available at bigdance-bracket.streamlit.app — no installation or login required. It lets you pick your bracket, configure your pool size, and run simulations to estimate your win probability, all from the browser.

The app supports both the men's and women's NCAA tournaments and includes an Upset Strategy tab with pre-computed analysis of winning bracket patterns by pool size, comparing winners' and losers' upset counts and madness scores.

See app/README.md for details on running the app locally or deploying your own instance.

Key Features

Real-time Basketball Data

Pull current team ratings, rankings, and matchup predictions:

from bigdance import Standings, Matchups

# Get current team standings (with Elo ratings)
standings = Standings()

# Get predictions for today's games
today_games = Matchups()

# Get women's basketball ratings instead
womens_standings = Standings(women=True)

# Filter by conference
acc_teams = Standings(conference="ACC")

# Print top teams by Elo rating
print(standings.elo.sort_values("ELO", ascending=False).head(10))

Hypothetical Tournament Simulation Before Bracket Release

Create a bracket based on current Warren Nolan rankings and accounting for automatic conference bids:

from bigdance import create_teams_from_standings, Standings

# Get current standings
standings = Standings()

# Create bracket with automatic conference bids and seeding
bracket = create_teams_from_standings(standings)

# Simulate tournament once
results = bracket.simulate_tournament()

# Get the champion
champion = results["Champion"]
print(f"Simulated champion: {champion.name} (Seed {champion.seed})")

# Print all Final Four teams
for team in results["Final Four"]:
    print(f"{team.name} (Seed {team.seed}, {team.region} Region)")

To see each team's probability of reaching each round across many simulations:

from bigdance import simulate_round_probabilities, Standings

# Get current standings
standings = Standings()

# Simulate 1000 tournaments and compute round-by-round probabilities
df = simulate_round_probabilities(standings, num_sims=1000, upset_factor=0.25)

# Show top 20 teams by championship probability
print(df.head(20))

Output columns: Team, Seed, Region, First Round, Second Round, Sweet 16, Elite 8, Final Four, Championship — each showing the percentage of simulations in which that team reached that round.

Customizing Upset Likelihood

Control how often upsets occur in your simulations:

from bigdance import create_teams_from_standings, Standings

# Get current standings
standings = Standings()

# Create bracket 
bracket = create_teams_from_standings(standings)

# Adjust upset factor for all games
# Range from -1.0 (chalk/favorites always win) to 1.0 (coin flip/50-50)
for game in bracket.games:
    # Values around 0.3 tend to match historical upset rates
    game.upset_factor = 0.3  

# Simulate tournament with adjusted upset factor
results = bracket.simulate_tournament()

ESPN Tournament Challenge Bracket Pool Simulation

Pull brackets directly from ESPN Tournament Challenge:

from bigdance.espn_tc_scraper import ESPNBracket, ESPNPool

# Create a bracket handler for men's tournament (use women=True for women's tournament)
bracket_handler = ESPNBracket()

# Get the current tournament bracket
bracket_html = bracket_handler.get_bracket()
actual_bracket = bracket_handler.extract_bracket(bracket_html)

# Load a pool and all its entries
pool_manager = ESPNPool()
pool_id = "1234567"  # ESPN pool ID found in the URL after "bracket?id="
entries = pool_manager.load_pool_entries(pool_id)

# Create a simulation pool from ESPN entries
pool_sim = pool_manager.create_simulation_pool(pool_id)

# Simulate and display top entries
results = pool_sim.simulate_pool(num_sims=1000)
print(results.head(10))

You can also compute per-team round probabilities using the real ESPN bracket:

from bigdance.espn_tc_scraper import ESPNPool
from bigdance import simulate_round_probabilities

pool_sim = ESPNPool().create_simulation_pool("1234567")

# Use the real bracket to compute each team's odds of reaching each round
df = simulate_round_probabilities(bracket=pool_sim.actual_results, num_sims=1000)
print(df.head(20))

Game Importance Analysis

Analyze which games have the most impact on a pool's outcome:

from bigdance.espn_tc_scraper import ESPNPool, GameImportanceAnalyzer

# Load a pool from ESPN
pool_manager = ESPNPool()
pool_sim = pool_manager.create_simulation_pool("1234567") # ESPN pool ID

# Create analyzer
analyzer = GameImportanceAnalyzer(pool_sim)

# Analyze the importance of each remaining game
importance = analyzer.analyze_win_importance()

# Print human-readable summary
analyzer.print_importance_summary(importance)

# Focus on impact for a specific entry
analyzer.print_importance_summary(importance, entry_name="My Bracket")

Advanced Analysis

Analyze winning strategies and optimal upset selections using a hypothetical bracket based on current Warren Nolan rankings:

from bigdance import Standings
from bigdance.bracket_analysis import BracketAnalysis

# Get current standings
standings = Standings()

# Create analyzer
analyzer = BracketAnalysis(standings, num_pools=100)

# Run simulations
analyzer.simulate_pools(entries_per_pool=10)

# Generate comparative visualizations
analyzer.plot_comparative_upset_distributions()

# Find optimal upset strategy
strategy = analyzer.identify_optimal_upset_strategy()
print(strategy)

# Find common underdog picks in winning brackets
underdogs = analyzer.find_common_underdogs()
print(underdogs)

# Save comprehensive analysis
analyzer.save_all_comparative_data()

Or after the bracket is released, you can integrate with ESPN Tournament Challenge to work with the real tournament bracket and analyze winning strategies:

from bigdance.bracket_analysis import BracketAnalysis

# Use ESPN data instead of Warren Nolan
analyzer = BracketAnalysis(use_espn=True, women=False, num_pools=100)

# For Second Chance brackets (starting from Sweet 16)
analyzer = BracketAnalysis(use_espn=True, second_chance=True, num_pools=100)

# Run simulations with ESPN data as the reference bracket
analyzer.simulate_pools(entries_per_pool=10)

Historical Scheduling and Results

Access game schedules and results:

from bigdance import Schedule
from datetime import datetime, timedelta

# Get last week's games
last_week = datetime.now() - timedelta(days=7)
today = datetime.now()
schedule = Schedule(
    start=last_week.strftime("%Y-%m-%d"),
    stop=today.strftime("%Y-%m-%d")
)

# View games from each day
for day_games in schedule.games_per_day:
    print(f"Games on {day_games.date.strftime('%Y-%m-%d')}:")
    print(day_games.matchups)

Command Line Tools

The package provides a unified bigdance CLI with subcommands:

# Get current team standings and ratings
bigdance standings

# Show each team's probability of reaching each round (hypothetical bracket)
bigdance simulate --num_sims 1000 --upset_factor 0.25 --top 20

# Women's tournament
bigdance simulate --gender women --num_sims 1000

# Analyze a bracket pool from ESPN
bigdance espn --pool_id 1234567

# Show each team's round probabilities using the real ESPN bracket
bigdance espn --pool_id 1234567 --team_probs

# Find most important remaining games
bigdance espn --pool_id 1234567 --importance

# Run bracket analysis with ESPN data
bigdance analyze --use_espn --num_pools 100

# Women's tournament analysis
bigdance analyze --gender women --num_pools 100

Use bigdance <command> --help for full options on each subcommand. The legacy python -m bigdance.<module> invocations also still work.

Development

To install the package for development:

git clone https://github.com/tefirman/bigdance
cd bigdance
pip install -e ".[dev]"

Run tests:

pytest

Documentation

For detailed documentation on all functions and classes, use Python's built-in help:

import bigdance
help(bigdance)

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Warren Nolan website for providing college basketball data (no affiliation)
• ESPN Tournament Challenge for tournament brackets (no affiliation)
• Andrew Sundberg for historical tournament data used in testing

Author

• Taylor Firman (@tefirman)