Get ESPN's Matchup Predictions for 2017 NFL Seasons

matchups.py

"""
Hacky parse through ESPN's NFL schedule page.
"""
import pickle
import requests
from collections import namedtuple
from bs4 import BeautifulSoup as bs

base_url = 'http://www.espn.com'
Stats = namedtuple('Stats', ['home', 'home_val', 'away', 'away_val', 'score'])
FPI = { # 2017 FPI Rankings
    'NE': 9.2, 'GB': 4.8, 'PIT': 4.7, 'SEA': 4.2, 'ATL': 3.8, 'DAL': 3.8,
    'OAK': 3.7, 'KC': 3.4, 'CAR': 1.8, 'BAL': 1.3, 'NYG': 1.2, 'DEN': 0.8,
    'MIN': 0.8, 'CIN': 0.3, 'PHI': 0.2, 'TB': 0.1, 'NO': 0.1, 'IND': -0.2,
    'TEN': -0.6, 'DET': -0.7, 'LAC': -0.7, 'WSH': -0.8, 'MIA': -1.1, 'HOU': -1.1,
    'ARI': -1.5, 'BUF': -1.9, 'JAX': -4.0, 'CHI': -5.2, 'LAR': -5.3,
    'NYJ': -6.0, 'CLE': -7.6, 'SF': -7.6
}

def get_matchups():
	weeks = {}

	for week_number in [j+1 for j in range(17)]:
		r = requests.get(base_url+'/nfl/schedule/_/week/'+str(week_number))
		soup = bs(r.content, 'html.parser')

		matchups = []
		for link in soup.find_all('a'):
			href = link.get('href')

			if '?gameId' in href:
				m = requests.get(base_url+href)
				matchup_soup = bs(m.content, 'html.parser')

				for s in matchup_soup.find_all('span'):
					class_name = s.get('class')
					if not class_name:
						continue

					if 'home-team' in class_name:
						away = str(s.string)
					elif 'away-team' in class_name:
						home = str(s.string)
					elif 'value-home' in class_name:
						home_val = round(float(s.string.strip('%'))/100., 3)
					elif 'value-away' in class_name:
						away_val = round(float(s.string.strip('%'))/100., 3)
						score = round(abs(FPI[home] - FPI[away]) * abs(home_val - away_val), 3)

				matchups.append(Stats(home, home_val, away, away_val, score))

			weeks[week_number] = matchups

		return weeks

if __name__ == '__main__':
	weeks = get_matchups()