Intro
The first thing you need when wanting to look at NHL data is NHL data. Unfortunately there isn’t too much documentation, but there is an API and resources (like https://github.com/Zmalski/NHL-API-Reference) which help along the way.
The main things I wanted to achieve was a table with all actions. For every shot, hit, goal etc. I wanted a table with all of the possible informatiion. This could then be supplemented with standings information, roster information etc.
Set Up
There are many different ways to do this. My preference is to run python script in JupyterLab within Anaconda:
You can then open a notebook:
The Code
Set Up
First, we need to import some libraries:
import requests
import pandas as pd
import datetime as dt
from datetime import datetime, timedelta
import json
import os
In the next cell, I like to create my variables. The two I need are one for if we choose to run this code on 1 season only (one_season_sel) and another if we wish to run the code on multiple seasons, specify the first (from_season_sel). Where these are used will become clear later:
one_season_sel = 20202021
from_season_sel = 20122013
Team Information
The first element I pulled was the team information in the NHL API:
# Make HTTP request to NHL API
team_url = "https://api.nhle.com/stats/rest/en/team"
team_response = requests.get(team_url)
# Convert response to dictionary
team_data = team_response.json()
# Extract team data
team_data_extract = team_data['data']
# Create empty list to store team data
team_list = []
# Loop through teams and extract data
for team in team_data_extract:
team_dict = {}
team_dict['id'] = team['id']
team_dict['franchiseId'] = team['franchiseId']
team_dict['fullName'] = team['fullName']
team_dict['leagueId'] = team['leagueId']
team_dict['rawTricode'] = team['rawTricode']
team_dict['triCode'] = team['triCode']
team_list.append(team_dict)
# Create DataFrame for team data
team_df = pd.DataFrame(team_list)
team_df = team_df.replace(',', '', regex=True)
There are other elements to the team information we could pull, however these are live/updated metrics which I may need to calculate myself on different time periods, rather than just seeing the latest information. If you want this, add the variables to the “for team in team_data_extract” loop in the code above. Possible variables include:
| summary | faceoffpercentages | daysbetweengames | faceoffwins | goalsagainstbystrength |
| goalsbyperiod | goalsforbystrength | leadingtrailing | realtime | outshootoutshotby |
| penalties | penaltykill | penaltykilltime | powerplay | powerplaytime |
| summaryshooting | percentages | scoretrailfirst | shootout | shottype |
| goalgames |
Roster Information
To pull all teams roster information, we need a list of the team codes:
df_teamcodes = team_df[['triCode']]
# Initialize an empty list to store data
df_rosterseasons = []
# Loop through team codes and retrieve roster data
for index, row in df_teamcodes.iterrows():
tri_code = row['triCode']
roster_url = "https://api-web.nhle.com/v1/roster-season/" + tri_code
# Make HTTP request to NHL API for team data
response = requests.get(roster_url)
roster_data = response.json()
# Append data for each team to the list
for season in roster_data:
df_rosterseasons.append({'triCode': tri_code, 'season': season})
# Create a DataFrame from the collected data
df_rosterseasons_forloop = pd.DataFrame(df_rosterseasons)
# Limit seasons run here
df_rosterseasons_forloop = df_rosterseasons_forloop[(df_rosterseasons_forloop['season'] >= from_season_sel)]
You will notice the final line here is where we can limit which seasons we run this data for.
We can then pull the roster data for these seasons:
# Initialize lists to store data
forward_data = []
defensemen_data = []
goalies_data = []
# Loop through each row in the DataFrame
for index, row in df_rosterseasons_forloop.iterrows():
tri_code = row['triCode']
season = row['season']
# Make HTTP request to NHL API for team data
roster_url = f"https://api-web.nhle.com/v1/roster/{tri_code}/{season}"
response = requests.get(roster_url)
# Convert response to dictionary
roster_data = response.json()
# Loop through forwards and extract relevant information
for forward in roster_data.get('forwards', []):
forward_info = {
'triCode': tri_code,
'season': season,
'playerID': forward.get('id', None),
'player_name': forward.get('firstName', {}).get('default', '') + ' ' + forward.get('lastName', {}).get('default', ''),
'sweater_number': forward.get('sweaterNumber', 'N/A'),
'position': forward.get('positionCode', ''),
'shoots_catches': forward.get('shootsCatches', ''),
'height_cm': forward.get('heightInCentimeters', ''),
'weight_kg': forward.get('weightInKilograms', ''),
'height_in': forward.get('heightInInches', ''),
'weight_lbs': forward.get('weightInPounds', ''),
'birth_date': forward.get('birthDate', ''),
'birth_city': forward.get('birthCity', {}).get('default', ''),
'birth_country': forward.get('birthCountry', ''),
'headshot_url': forward.get('headshot', '')
}
forward_data.append(forward_info)
# Loop through defence and extract relevant information
for defensemen in roster_data.get('defensemen', []):
defensemen_info = {
'triCode': tri_code,
'season': season,
'playerID': defensemen.get('id', None),
'player_name': defensemen.get('firstName', {}).get('default', '') + ' ' + defensemen.get('lastName', {}).get('default', ''),
'sweater_number': defensemen.get('sweaterNumber', 'N/A'),
'position': defensemen.get('positionCode', ''),
'shoots_catches': defensemen.get('shootsCatches', ''),
'height_cm': defensemen.get('heightInCentimeters', ''),
'weight_kg': defensemen.get('weightInKilograms', ''),
'height_in': defensemen.get('heightInInches', ''),
'weight_lbs': defensemen.get('weightInPounds', ''),
'birth_date': defensemen.get('birthDate', ''),
'birth_city': defensemen.get('birthCity', {}).get('default', ''),
'birth_country': defensemen.get('birthCountry', ''),
'headshot_url': defensemen.get('headshot', '')
}
defensemen_data.append(defensemen_info)
# Loop through goalies and extract relevant information
for goalies in roster_data.get('goalies', []):
goalies_info = {
'triCode': tri_code,
'season': season,
'playerID': goalies.get('id', None),
'player_name': goalies.get('firstName', {}).get('default', '') + ' ' + goalies.get('lastName', {}).get('default', ''),
'sweater_number': goalies.get('sweaterNumber', 'N/A'),
'position': goalies.get('positionCode', ''),
'shoots_catches': goalies.get('shootsCatches', ''),
'height_cm': goalies.get('heightInCentimeters', ''),
'weight_kg': goalies.get('weightInKilograms', ''),
'height_in': goalies.get('heightInInches', ''),
'weight_lbs': goalies.get('weightInPounds', ''),
'birth_date': goalies.get('birthDate', ''),
'birth_city': goalies.get('birthCity', {}).get('default', ''),
'birth_country': goalies.get('birthCountry', ''),
'headshot_url': goalies.get('headshot', '')
}
goalies_data.append(goalies_info)
# Create DataFrames for forwards, defensemen, and goalies
df_forwards = pd.DataFrame(forward_data)
df_defense = pd.DataFrame(defensemen_data)
df_goalies = pd.DataFrame(goalies_data)
# Union all of the data together
df_roster = pd.concat([df_forwards, df_defense, df_goalies], ignore_index=True)
Schedule Information
The next data we can pull is the schedule information, to know which team is playing who on each gameID:
# Initialize an empty list to store dictionaries for each game
games_data = []
# Loop through each row in the DataFrame
for index, row in df_rosterseasons_forloop.iterrows():
tri_code = row['triCode']
season = row['season']
# Make HTTP request to NHL API for team data
schedule_url = f"https://api-web.nhle.com/v1/club-schedule-season/{tri_code}/{season}"
response = requests.get(schedule_url)
# Convert response to dictionary
schedule_data = response.json()
games = schedule_data.get('games', [])
# Loop through each game and extract all columns
for game in games:
game_dict = {
'id': game.get('id'),
'season': game.get('season'),
'gameType': game.get('gameType'),
'gameDate': game.get('gameDate'),
'venue': game['venue'].get('default'),
'neutralSite': game.get('neutralSite'),
'startTimeUTC': game.get('startTimeUTC'),
'easternUTCOffset': game.get('easternUTCOffset'),
'venueUTCOffset': game.get('venueUTCOffset'),
'venueTimezone': game.get('venueTimezone'),
'gameState': game.get('gameState'),
'gameScheduleState': game.get('gameScheduleState'),
#'tvBroadcasts': game.get('tvBroadcasts'),
'awayTeamID': game['awayTeam'].get('id'),
'awayTeamPlaceName': game['awayTeam']['placeName'].get('default'),
'awayTeamAbbrev': game['awayTeam'].get('abbrev'),
'awayTeamLogo': game['awayTeam'].get('logo'),
'awayTeamScore': game['awayTeam'].get('score'),
'homeTeamID': game['homeTeam'].get('id'),
'homeTeamPlaceName': game['homeTeam']['placeName'].get('default'),
'homeTeamAbbrev': game['homeTeam'].get('abbrev'),
'homeTeamLogo': game['homeTeam'].get('logo'),
'homeTeamScore': game['homeTeam'].get('score'),
'periodType': game.get('periodDescriptor', {}).get('periodType'),
'lastPeriodType': game.get('gameOutcome', {}).get('lastPeriodType'),
'gameCenterLink': game.get('gameCenterLink'),
'winningGoalieID': game.get('winningGoalie', {}).get('playerId', 'n/a'),
'winningGoalie': game.get('winningGoalie', {}).get('lastName', {}).get('default', 'n/a'),
'winningGoalScorerID': game.get('winningGoalScorer', {}).get('playerId', 'n/a'),
'winningGoalScorer': game.get('winningGoalScorer', {}).get('lastName', {}).get('default', 'n/a'),
}
games_data.append(game_dict)
# Create DataFrame from the list of dictionaries
df_games = pd.DataFrame(games_data)
df_games['gameTypeDesc'] = df_games['gameType'].apply(lambda x: 'Preseason' if x == 1 else ('Regular' if x == 2 else ('Playoffs' if x == 3 else 'Unknown')))
df_games = df_games.drop_duplicates()
Standings Information
The next dataset we can pull is the standings data for every day.
First, we need to limit the days which we pull this data for. I chose to do this one season at a time:
# Get list of dates
# Extracting the year portion from season_sel
year = int(str(one_season_sel)[:4]) # Assuming the format is consistent
# Creating start_date and end_date strings based on the extracted year
start_date = f"09-01-{year - 1}"
end_date = f"08-01-{year}"
# Generate a list of dates between start_date and end_date
date_list = pd.date_range(start=start_date, end=end_date)
# Create a DataFrame with the list of dates
df_dts = pd.DataFrame(date_list, columns=['Date'])
df_dts['Date'] = df_dts['Date'].astype(str)
You can then use this list of dates to create the standings for each day:
# Initialize an empty list to store standings data
standings_data_list = []
# Loop through each row in the DataFrame
for index, row in df_dts.iterrows():
Date = row['Date']
# Make HTTP request to NHL API for team data
standings_url = f"https://api-web.nhle.com/v1/standings/{Date}"
response = requests.get(standings_url)
standings_data = response.json()
# Extracting the 'standings' data
standings_list = standings_data.get('standings', [])
# Initialize an empty list to store extracted columns
extracted_columns = []
# Loop through each team's standing information
for team_info in standings_list:
extracted_info = {
'conferenceAbbrev': team_info.get('conferenceAbbrev'),
'conferenceName': team_info.get('conferenceName'),
'placeName_default': team_info.get('placeName', {}).get('default'),
'date': team_info.get('date'),
'divisionAbbrev': team_info.get('divisionAbbrev'),
'gamesPlayed': team_info.get('gamesPlayed'),
'goalDifferential': team_info.get('goalDifferential'),
'goalDifferentialPctg': team_info.get('goalDifferentialPctg'),
'goalAgainst': team_info.get('goalAgainst'),
'goalFor': team_info.get('goalFor'),
'goalsForPctg': team_info.get('goalsForPctg'),
'homeGamesPlayed': team_info.get('homeGamesPlayed'),
'homeGoalDifferential': team_info.get('homeGoalDifferential'),
'homeGoalsAgainst': team_info.get('homeGoalsAgainst'),
'homeGoalsFor': team_info.get('homeGoalsFor'),
'homeLosses': team_info.get('homeLosses'),
'homeOtLosses': team_info.get('homeOtLosses'),
'homePoints': team_info.get('homePoints'),
'homeRegulationPlusOtWins': team_info.get('homeRegulationPlusOtWins'),
'homeRegulationWins': team_info.get('homeRegulationWins'),
'homeTies': team_info.get('homeTies'),
'homeWins': team_info.get('homeWins'),
'l10GamesPlayed': team_info.get('l10GamesPlayed'),
'l10GoalDifferential': team_info.get('l10GoalDifferential'),
'l10GoalsAgainst': team_info.get('l10GoalsAgainst'),
'l10GoalsFor': team_info.get('l10GoalsFor'),
'l10Losses': team_info.get('l10Losses'),
'l10OtLosses': team_info.get('l10OtLosses'),
'l10Points': team_info.get('l10Points'),
'l10RegulationPlusOtWins': team_info.get('l10RegulationPlusOtWins'),
'l10RegulationWins': team_info.get('l10RegulationWins'),
'l10Ties': team_info.get('l10Ties'),
'l10Wins': team_info.get('l10Wins'),
'leagueHomeSequence': team_info.get('leagueHomeSequence'),
'leagueL10Sequence': team_info.get('leagueL10Sequence'),
'leagueRoadSequence': team_info.get('leagueRoadSequence'),
'leagueSequence': team_info.get('leagueSequence'),
'losses': team_info.get('losses'),
'otLosses': team_info.get('otLosses'),
'placeName_default': team_info.get('placeName', {}).get('default'),
'pointPctg': team_info.get('pointPctg'),
'points': team_info.get('points'),
'regulationPlusOtWinPctg': team_info.get('regulationPlusOtWinPctg'),
'regulationPlusOtWins': team_info.get('regulationPlusOtWins'),
'regulationWinPctg': team_info.get('regulationWinPctg'),
'regulationWins': team_info.get('regulationWins'),
'roadGamesPlayed': team_info.get('roadGamesPlayed'),
'roadGoalDifferential': team_info.get('roadGoalDifferential'),
'roadGoalsAgainst': team_info.get('roadGoalsAgainst'),
'roadGoalsFor': team_info.get('roadGoalsFor'),
'roadLosses': team_info.get('roadLosses'),
'roadOtLosses': team_info.get('roadOtLosses'),
'roadPoints': team_info.get('roadPoints'),
'roadRegulationPlusOtWins': team_info.get('roadRegulationPlusOtWins'),
'roadRegulationWins': team_info.get('roadRegulationWins'),
'roadTies': team_info.get('roadTies'),
'roadWins': team_info.get('roadWins'),
'seasonId': team_info.get('seasonId'),
'shootoutLosses': team_info.get('shootoutLosses'),
'shootoutWins': team_info.get('shootoutWins'),
'streakCode': team_info.get('streakCode'),
'streakCount': team_info.get('streakCount'),
'teamName_default': team_info.get('teamName', {}).get('default'),
'teamCommonName_default': team_info.get('teamCommonName', {}).get('default'),
'teamAbbrev_default': team_info.get('teamAbbrev', {}).get('default'),
'teamLogo': team_info.get('teamLogo'),
'winPctg': team_info.get('winPctg'),
'wins': team_info.get('wins'),
}
standings_data_list.append(extracted_info)
# Create a DataFrame
df_standings = pd.DataFrame(standings_data_list)
Game Action Information
The final set of data we can pull is the most important - the data for every action that occured in every game we specify. This can take some time to run so I opted to run this on a season-by-season basis. But you can pull this for specific games if you so choose:
# Limit To 1 Season
df_LimitSeasons = df_games[(df_games['season'] == one_season_sel)]
# Create a List of GameIDs for the above date range
df_gameIDs = df_LimitSeasons['id'].drop_duplicates()
df_gameIDs = pd.DataFrame(df_gameIDs, columns=['id'])
# Initialize lists to hold data
all_basic_info = []
all_plays_info = []
# Loop through each row in the DataFrame
for index, row in df_gameIDs.iterrows():
gameID = row['id']
# Make HTTP request to NHL API for team data
gameaction_url = f"https://api-web.nhle.com/v1/gamecenter/{gameID}/play-by-play"
response = requests.get(gameaction_url)
# Convert response to dictionary
action_data = response.json()
# Extracting basic information
basic_info = {
'id': action_data.get('id', 'n/a'),
'season': action_data.get('season', 'n/a'),
'gameType': action_data.get('gameType', 'n/a'),
'gameDate': action_data.get('gameDate', 'n/a'),
'venue': action_data.get('venue', {}).get('default', 'n/a'),
'neutralSite': action_data.get('neutralSite', 'n/a'),
'startTimeUTC': action_data.get('startTimeUTC', 'n/a'),
'easternUTCOffset': action_data.get('easternUTCOffset', 'n/a'),
'venueUTCOffset': action_data.get('venueUTCOffset', 'n/a'),
'venueTimezone': action_data.get('venueTimezone', 'n/a'),
'gameState': action_data.get('gameState', 'n/a'),
'gameScheduleState': action_data.get('gameScheduleState', 'n/a'),
'period': action_data.get('period', 'n/a'),
'periodDescriptor_number': action_data.get('periodDescriptor', {}).get('number', 'n/a'),
'periodDescriptor_periodType': action_data.get('periodDescriptor', {}).get('periodType', 'n/a'),
'away_id': action_data.get('awayTeam', {}).get('id', 'n/a'),
'away_name': action_data.get('awayTeam', {}).get('name', {}).get('default', 'n/a'),
'away_abbrev': action_data.get('awayTeam', {}).get('abbrev', 'n/a'),
'awayTeamPlaceName': action_data.get('awayTeam', {}).get('placeName', {}).get('default', 'n/a'),
'away_score': action_data.get('awayTeam', {}).get('score', 'n/a'),
'away_sog': action_data.get('awayTeam', {}).get('sog', 'n/a'),
'away_logo': action_data.get('awayTeam', {}).get('logo', 'n/a'),
'home_id': action_data.get('homeTeam', {}).get('id', 'n/a'),
'home_name': action_data.get('homeTeam', {}).get('name', {}).get('default', 'n/a'),
'home_abbrev': action_data.get('homeTeam', {}).get('abbrev', 'n/a'),
'homeTeamPlaceName': action_data.get('homeTeam', {}).get('placeName', {}).get('default', 'n/a'),
'home_score': action_data.get('homeTeam', {}).get('score', 'n/a'),
'home_sog': action_data.get('homeTeam', {}).get('sog', 'n/a'),
'home_logo': action_data.get('homeTeam', {}).get('logo', 'n/a'),
'timeRemaining': action_data.get('clock', {}).get('timeRemaining', 'n/a'),
'secondsRemaining': action_data.get('clock', {}).get('secondsRemaining', 'n/a'),
'running': action_data.get('clock', {}).get('running', 'n/a'),
'inIntermission': action_data.get('clock', {}).get('inIntermission', 'n/a'),
'periodType': action_data.get('periodDescriptor', {}).get('periodType', 'n/a'),
'gameOutcome_lastPeriodType': action_data.get('gameOutcome', {}).get('lastPeriodType', 'n/a')
}
all_basic_info.append(basic_info)
# Extracting 'plays' data
plays_data = action_data.get('plays', [])
for play in plays_data:
details = play.get('details', {})
play_info = {
'id': action_data.get('id', 'n/a'),
'eventId': play.get('eventId', 'n/a'),
'period': play.get('period', 'n/a'),
'timeInPeriod': play.get('timeInPeriod', 'n/a'),
'timeRemaining': play.get('timeRemaining', 'n/a'),
'situationCode': play.get('situationCode', 'n/a'),
'homeTeamDefendingSide': play.get('homeTeamDefendingSide', 'n/a'),
'typeCode': play.get('typeCode', 'n/a'),
'typeDescKey': play.get('typeDescKey', 'n/a'),
'sortOrder': play.get('sortOrder', 'n/a'),
'eventOwnerTeamId': details.get('eventOwnerTeamId', 'n/a'),
'losingPlayerId': details.get('losingPlayerId', 'n/a'),
'winningPlayerId': details.get('winningPlayerId', 'n/a'),
'xCoord': details.get('xCoord', 'n/a'),
'yCoord': details.get('yCoord', 'n/a'),
'zoneCode': details.get('zoneCode', 'n/a'),
'shotType': details.get('shotType', 'n/a'),
'scoringPlayerId': details.get('scoringPlayerId', 'n/a'),
'scoringPlayerTotal': details.get('scoringPlayerTotal', 'n/a'),
'assist1PlayerId': details.get('assist1PlayerId', 'n/a'),
'assist1PlayerTotal': details.get('assist1PlayerTotal', 'n/a'),
'assist2PlayerId': details.get('assist2PlayerId', 'n/a'),
'assist2PlayerTotal': details.get('assist2PlayerTotal', 'n/a'),
'goalieInNetId': details.get('goalieInNetId', 'n/a'),
'awayScore': details.get('awayScore', 'n/a'),
'homeScore': details.get('homeScore', 'n/a'),
'playerId': details.get('playerId', 'n/a'),
'shootingPlayerId': details.get('shootingPlayerId', 'n/a'),
'awaySOG': details.get('awaySOG', 'n/a'),
'homeSOG': details.get('homeSOG', 'n/a'),
'reason': details.get('reason', 'n/a'),
'hittingPlayerId': details.get('hittingPlayerId', 'n/a'),
'hitteePlayerId': details.get('hitteePlayerId', 'n/a'),
'blockingPlayerId': details.get('blockingPlayerId', 'n/a'),
'descKey': details.get('descKey', 'n/a'),
'duration': details.get('duration', 'n/a'),
'committedByPlayerId': details.get('committedByPlayerId', 'n/a'),
'drawnByPlayerId': details.get('drawnByPlayerId', 'n/a')
}
all_plays_info.append(play_info)
# Create DataFrames from the extracted data
GameAction_BasicInfo = pd.DataFrame(all_basic_info)
GameAction_Plays = pd.DataFrame(all_plays_info)
Conclusion
This puts all the data into dataframes. You can then export or work the data however you wish. I chose to output this data to csv (old school) and work from there.
Good luck!
#32
Last Updated: 2024-04-14