world-cup-knockout-simulation.py

import random
import pandas as pd

results_df = pd.read_csv('results_tally_blank.csv')
results_df.set_index("Country", inplace=True)

group_A = [{
  "team": "Netherlands",
  "rating": 1716
}, {
  "team": "Senegal",
  "rating": 1604
}]

group_B = [{
  "team": "England",
  "rating": 1756
}, {
  "team": "United States",
  "rating": 1653
}]

# ... and so on up to Group H

def play_match(team1, team2, round):
  
  r_diff = team2["rating"] - team1["rating"]
  pr_team1_win = 1 / (1 + 10**(r_diff / 400))

  if random.random() < pr_team1_win:
    new_rating = team1["rating"] + 60 * (1 - pr_team1_win)
    winner = {"team": team1["team"], "rating": new_rating}
    loser = team2
    
  else:
    new_rating = team2["rating"] + 60 * pr_team1_win
    winner = {"team": team2["team"], "rating": new_rating}
    loser = team1

  results_df.loc[winner["team"]][round] += 1
  results_df.loc[loser["team"]][round] += 1

  return winner

# https://www.worldcup2022football.co.uk/wc-2022-schedule.pdf

def simulate_knockout():

  quarter_finals_teams = [
    #Match 49
    play_match(group_A[0], group_B[1], "Round of 16"),

    #Match 50
    play_match(group_C[0], group_D[1], "Round of 16"),

    #Match 51
    play_match(group_A[1], group_B[0], "Round of 16"),

    #Match 52
    play_match(group_C[1], group_D[0], "Round of 16"),

    #Match 53
    play_match(group_E[0], group_F[1], "Round of 16"),

    #Match 54
    play_match(group_G[0], group_H[1], "Round of 16"),

    #Match 55
    play_match(group_E[1], group_F[0], "Round of 16"),

    #Match 56
    play_match(group_G[1], group_H[0], "Round of 16")
  ]

  QF = quarter_finals_teams

  semi_finals_teams = [
    #Match 57
    play_match(QF[0], QF[1], "Quarter Finals"),

    #Match 58
    play_match(QF[4], QF[5], "Quarter Finals"),

    #Match 59
    play_match(QF[2], QF[3], "Quarter Finals"),

    #Match 60
    play_match(QF[6], QF[7], "Quarter Finals"),
  ]

  SF = semi_finals_teams

  final_teams = [
    #Match 61
    play_match(SF[0], SF[1], "Semi Finals"),

    #Match 62
    play_match(SF[2], SF[3], "Semi Finals"),
  ]

  winner = play_match(final_teams[0], final_teams[1], "Final")

  results_df.loc[winner["team"]]["Win"] += 1

  return "one tournament done!"


import time
start_time = time.time()

num_sims = 50000
for i in range(0, num_sims):
  simulate_knockout()
print(results_df.to_string())

print("\n--- " + str(num_sims) + " simulations took " +
      str(round(time.time() - start_time, 2)) + " seconds ---")

results_df.to_csv("results_output.csv", encoding='utf-8')