This project analyzes recent performance trends of NBA teams and players in playin Tournament matchups using data from the nba_api. It focuses on two head-to-head comparisons:
- Golden State Warriors vs Memphis Grizzlies
- Sacramento Kings vs Dallas Mavericks
The goal is to understand which players are key contributors, especially in 4th quarters, and how their impact correlates with team performance (+/- stats). The reason for me to start this project is today is 2025/04/14 and tommorow is the first day pf the Play-in Tournament,I want to see who have a better chance to win and I want to give advise to the teams like there anaylsit and see the result after my prediction.
- Pulls full season game logs using
nba_api - Filters games where teams faced each other
- Extracts quarter-wise player scoring from Play-by-Play data
- Retrieves plus/minus stats from Box Scores
- Aggregates performance by:
- Total points
- Average points per game
- Average points in 4th quarters
- Average +/- rating
- Visualizes key players by plotting:
avg_4Q_ptsvsavg_plus_minus- Colored by team
- With player name annotations
- Python 🐍
nba_apipandasmatplotlib+seaborn
I look though the data from this season's team stats and filtered out the Warriors and Grizzlies' last 10 games.
from nba_api.stats.static import teams
from nba_api.stats.endpoints import teamgamelog
import pandas as pd
team_names = ['Golden State Warriors', 'Memphis Grizzlies', 'Sacramento Kings', 'Dallas Mavericks']
all_teams = teams.get_teams()
team_ids = {team['full_name']: team['id'] for team in all_teams if team['full_name'] in team_names}
def get_last_10_games(team_id, team_name):
gamelog = teamgamelog.TeamGameLog(team_id=team_id, season='2024-25', season_type_all_star='Regular Season')
df = gamelog.get_data_frames()[0]
df = df[['GAME_DATE', 'MATCHUP', 'WL', 'PTS', 'REB', 'AST']].head(10)
df['TEAM'] = team_name
return df
all_data = pd.concat([
get_last_10_games(team_ids['Golden State Warriors'], 'Warriors'),
get_last_10_games(team_ids['Memphis Grizzlies'], 'Grizzlies'),
get_last_10_games(team_ids['Sacramento Kings'], 'Kings'),
get_last_10_games(team_ids['Dallas Mavericks'], 'Mavericks')
])
print("Last 10 games stats：")
print(all_data)
print(all_data['MATCHUP'].unique())
summary = all_data.groupby(['TEAM', 'WL']).size().unstack(fill_value=0)
summary['WIN_RATE'] = summary['W'] / (summary['W'] + summary['L'])
print("\nWinRates：\n", summary)The result I got is the list below:
WinRates：
WL L W WIN_RATE
TEAM
Grizzlies 6 4 0.4
Warriors 3 7 0.7- we can see that the Warriors have a better win_rate than the Grizzlies in the last 10 games of this season
Here is the code I wrote to find out how there match up games turned out this season.
from nba_api.stats.static import teams
from nba_api.stats.endpoints import teamgamelog
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
team_names = ['Golden State Warriors', 'Memphis Grizzlies', 'Sacramento Kings', 'Dallas Mavericks']
all_teams = teams.get_teams()
team_ids = {team['full_name']: team['id'] for team in all_teams if team['full_name'] in team_names}
def get_games(team_id, team_name):
gamelog = teamgamelog.TeamGameLog(team_id=team_id, season='2024-25', season_type_all_star='Regular Season')
df = gamelog.get_data_frames()[0]
df = df[['GAME_DATE', 'MATCHUP', 'WL', 'PTS', 'REB', 'AST']]
df['TEAM'] = team_name
return df
# Concatenate data for all teams
all_data = pd.concat([
get_games(team_ids['Golden State Warriors'], 'Warriors'),
get_games(team_ids['Memphis Grizzlies'], 'Grizzlies'),
get_games(team_ids['Sacramento Kings'], 'Kings'),
get_games(team_ids['Dallas Mavericks'], 'Mavericks')
])
# Pivot the table so that TEAM is the header
gsw_vs_mem_data = all_data[all_data['MATCHUP'].str.contains(r'GSW.*MEM|MEM.*GSW')]
gsw_vs_mem_Wariorsdata = gsw_vs_mem_data[gsw_vs_mem_data['TEAM'] == 'Warriors']
gsw_vs_mem_Memphisdata = gsw_vs_mem_data[gsw_vs_mem_data['TEAM'] == 'Grizzlies']
Kings_vs_Mavs_data = all_data[all_data['MATCHUP'].str.contains(r'SAC.*DAL|DAL.*SAC')]
Kings_vs_Mavs_kingsdata = Kings_vs_Mavs_data[Kings_vs_Mavs_data['TEAM'] == 'Kings']
Kings_vs_Mavs_mavsdata = Kings_vs_Mavs_data[Kings_vs_Mavs_data['TEAM'] == 'Mavericks']
print("\nGSW vs MEM Data with TEAM as Header in all_data:")
print(gsw_vs_mem_Memphisdata)
print(gsw_vs_mem_Wariorsdata)
print("\nKings vs Mavs Data with TEAM as Header in all_data:")
print(Kings_vs_Mavs_kingsdata)
print(Kings_vs_Mavs_mavsdata)the result:
GSW vs MEM Data with TEAM as Header in all_data:
GAME_DATE MATCHUP WL PTS REB AST TEAM
6 APR 01, 2025 MEM vs. GSW L 125 43 22 Grizzlies
46 JAN 04, 2025 MEM @ GSW L 113 43 29 Grizzlies
54 DEC 19, 2024 MEM vs. GSW W 144 60 38 Grizzlies
69 NOV 15, 2024 MEM @ GSW L 118 54 22 Grizzlies
GAME_DATE MATCHUP WL PTS REB AST TEAM
7 APR 01, 2025 GSW @ MEM W 134 49 29 Warriors
48 JAN 04, 2025 GSW vs. MEM W 121 29 32 Warriors
56 DEC 19, 2024 GSW @ MEM L 93 36 20 Warriors
70 NOV 15, 2024 GSW vs. MEM W 123 56 33 WarriorsWhat I see from the chart above is that
- Their score are pretty close so there is a big chance that they will play a very tight game from the first quarter to the end of the 4th.
- The Warriors can not stop the Grizzlies score, but the Grizzlies can't either. Their scores each games are high and the only lost for the Warriors was its score under 100 pts.
- The rebound can be a big problem for the Warriors but if they can secure their rebounds, they would have a better chance of winning, as for the Grizzlies,they can rebound pertty consistently comparing to the Warriors.
- Warriors is a better passing team overall, that requires the Grizzlies to be communicate more to have a succuessfull defence.
I showed the stats for the players from these 2 teams when they are playing aginst each other, since they are playing a win or go home game, it is really import thing to show who is taking over the 4th quarter for their team and here is the code:
from nba_api.stats.static import teams
from nba_api.stats.endpoints import teamgamelog, playbyplayv2, boxscoretraditionalv2
import pandas as pd
import time
import matplotlib.pyplot as plt
import seaborn as sns
# 1. Get team ID
team_id_gsw = teams.find_teams_by_full_name("Golden State Warriors")[0]['id']
season = '2024-25'
# 2. Get all Warriors games and filter GSW vs MEM
games = teamgamelog.TeamGameLog(team_id=team_id_gsw, season=season).get_data_frames()[0]
gsw_vs_mem = games[games['MATCHUP'].str.contains('MEM')]
game_ids = gsw_vs_mem['Game_ID'].tolist()
# 3. Get per-quarter points for each player
all_player_stats = []
for game_id in game_ids:
pbp = playbyplayv2.PlayByPlayV2(game_id=game_id).get_data_frames()[0]
points_events = pbp[pbp['EVENTMSGTYPE'].isin([1, 3])]
points_events['PTS'] = points_events['EVENTMSGTYPE'].apply(lambda x: 2 if x == 1 else 1)
points_events['PTS'] = points_events.apply(
lambda row: 3 if row['EVENTMSGTYPE'] == 1 and '3PT' in str(row['HOMEDESCRIPTION']) else row['PTS'], axis=1
)
game_stats = points_events.groupby(['PERIOD', 'PLAYER1_NAME'])['PTS'].sum().reset_index()
game_stats['GAME_ID'] = game_id
all_player_stats.append(game_stats)
time.sleep(0.6)
df = pd.concat(all_player_stats)
# 4. Get PLUS_MINUS and TEAM from box score
plus_minus_list = []
for game_id in game_ids:
box = boxscoretraditionalv2.BoxScoreTraditionalV2(game_id=game_id)
pm_df = box.get_data_frames()[0][['PLAYER_NAME', 'TEAM_ABBREVIATION', 'PLUS_MINUS']]
pm_df['GAME_ID'] = game_id
plus_minus_list.append(pm_df)
time.sleep(0.6)
pm_df = pd.concat(plus_minus_list)
# 5. Merge scoring and plus-minus data
merged = df.merge(pm_df, left_on=['PLAYER1_NAME', 'GAME_ID'], right_on=['PLAYER_NAME', 'GAME_ID'], how='left')
# 6. Calculate avg 4th quarter points
q4 = merged[merged['PERIOD'] == 4]
avg_q4_pts = q4.groupby('PLAYER1_NAME')['PTS'].mean().reset_index()
avg_q4_pts.rename(columns={'PTS': 'avg_4Q_pts'}, inplace=True)
# 7. Full summary
summary = merged.groupby(['PLAYER1_NAME', 'TEAM_ABBREVIATION']).agg(
total_pts=('PTS', 'sum'),
avg_pts_per_game=('PTS', 'mean'),
avg_plus_minus=('PLUS_MINUS', 'mean'),
games_played=('GAME_ID', 'nunique')
).reset_index()
# 8. Merge 4th quarter points
summary = summary.merge(avg_q4_pts, on='PLAYER1_NAME', how='left')
# 9. Sort by impact
summary = summary.sort_values(by='avg_4Q_pts', ascending=False)
sum_GSW = summary[summary['TEAM_ABBREVIATION'] == 'GSW']
summary_MEM = summary[summary['TEAM_ABBREVIATION'] == 'MEM']
# 10. Display
print("\n🏀 Key Players vs MEM (GSW perspective):")
print(sum_GSW.head(10))
print(summary_MEM.head(10))
plt.figure(figsize=(10, 6))
sns.scatterplot(data=summary, x='avg_4Q_pts', y='avg_plus_minus', hue='TEAM_ABBREVIATION', s=100)
# Add player names as labels
for _, row in summary.iterrows():
plt.text(row['avg_4Q_pts'] + 0.1, row['avg_plus_minus'], row['PLAYER1_NAME'], fontsize=9)
plt.title('4th Quarter Points vs Plus/Minus')
plt.xlabel('Avg 4Q Points')
plt.ylabel('Avg Plus/Minus')
plt.legend(title='Team')
plt.grid(True)
plt.tight_layout()
plt.show()Here is the result:
🏀 Key Players vs MEM (GSW perspective):
PLAYER1_NAME TEAM_ABBREVIATION total_pts avg_pts_per_game avg_plus_minus games_played avg_4Q_pts
15 Jimmy Butler III GSW 26 6.500000 6.000000 1 10.0
17 Jonathan Kuminga GSW 39 6.500000 -7.000000 3 10.0
1 Brandin Podziemski GSW 39 4.875000 -4.250000 3 8.0
24 Moses Moody GSW 33 4.714286 -0.857143 3 7.5
5 Dennis Schröder GSW 25 5.000000 -5.800000 2 6.0
3 Buddy Hield GSW 46 4.600000 1.000000 4 6.0
29 Stephen Curry GSW 55 6.875000 6.750000 3 6.0
20 Lindy Waters III GSW 26 5.200000 -4.000000 3 5.0
19 Kyle Anderson GSW 24 3.000000 -0.500000 3 4.0
0 Andrew Wiggins GSW 52 6.500000 -7.625000 3 4.0
PLAYER1_NAME TEAM_ABBREVIATION total_pts avg_pts_per_game avg_plus_minus games_played avg_4Q_pts
21 Luke Kennard MEM 20 5.000000 3.000000 3 12.000000
10 Ja Morant MEM 45 6.428571 4.000000 2 9.000000
12 Jaren Jackson Jr. MEM 95 6.785714 1.142857 4 8.000000
4 Cam Spencer MEM 12 6.000000 -0.500000 2 6.000000
6 Desmond Bane MEM 72 5.142857 5.500000 4 5.333333
11 Jake LaRavia MEM 43 4.300000 10.300000 3 5.000000
14 Jaylen Wells MEM 41 4.100000 4.000000 4 5.000000
27 Santi Aldama MEM 44 4.888889 -4.444444 3 4.000000
32 Zach Edey MEM 38 3.454545 9.727273 4 3.333333
16 John Konchar MEM 6 3.000000 3.000000 1 3.000000
- We can see from the chart that for the Warriors, Jimmy Buttler played well and scored 10 pts in the 4th quarter, but since he just joined the team, he only played againest the Grizzlies once with the Warriors, beside him, the first man with a possitive average +/- rate and with the most averge 4th quarter points is Steph Curry.
- For the Grizzlies, Luke Kennard, Ja Morant and Jaren Jackson JR are the 3 main key player to watch out in the 4th, Desmond Bane has the second most total Points and positive +/- rate and Avg 4Q_pts as well.
- From the Scatter plot above, we also see that Draymond Green has better +/- rate than both Steph and Jimmy. Zach Edey also having one of the highest +/- rate with Jake LaRavia from the Grizllies.
If the Warriors want to win the game
- the big 3 have to be the best of them selfs, watch out for Luke Kennard and Ja Morant because they can score alot in the 4th quarter
- since 4 of the players that are negetive +/- rates were traded
- the new roaster can be better against the Grizzlies as the Warriors won 7 out of the last 10.
- Secure the rebound.
If the Grizzlies want to win the game
- Do what ever you can top stop the big 3 of the Warriors making impact for their team.
- Give more time to the players that have better +/- rate since you have more players with that.
Overall the Grizzlies recently just fired their head coach so it is a big impact, Warriors are 3-1 against the Grizzlies, Warriors have a better record the last 10 games, and they have a home court advantage , so I will Pick the Golden State Warriors to win this Playin Tournament.
WinRates：
WL L W WIN_RATE
TEAM
Kings 5 5 0.5
Mavericks 6 4 0.4Kings vs Mavs Data with TEAM as Header in all_data:
GAME_DATE MATCHUP WL PTS REB AST TEAM
22 MAR 03, 2025 SAC @ DAL W 122 37 34 Kings
29 FEB 10, 2025 SAC @ DAL W 129 50 28 Kings
49 DEC 30, 2024 SAC vs. DAL W 110 58 25 Kings
GAME_DATE MATCHUP WL PTS REB AST TEAM
20 MAR 03, 2025 DAL vs. SAC L 98 37 28 Mavericks
28 FEB 10, 2025 DAL vs. SAC L 128 33 31 Mavericks
49 DEC 30, 2024 DAL @ SAC L 100 36 17 Mavericks
from nba_api.stats.static import teams
from nba_api.stats.endpoints import teamgamelog, playbyplayv2, boxscoretraditionalv2
import pandas as pd
import time
import seaborn as sns
import matplotlib.pyplot as plt
# 1. Get team ID for Kings
team_id_kings = teams.find_teams_by_full_name("Sacramento Kings")[0]['id']
season = '2024-25'
# 2. Get all Kings games and filter SAC vs DAL
games = teamgamelog.TeamGameLog(team_id=team_id_kings, season=season).get_data_frames()[0]
sac_vs_dal = games[games['MATCHUP'].str.contains('DAL')]
game_ids = sac_vs_dal['Game_ID'].tolist()
# 3. Get per-quarter points for each player
all_player_stats = []
for game_id in game_ids:
pbp = playbyplayv2.PlayByPlayV2(game_id=game_id).get_data_frames()[0]
points_events = pbp[pbp['EVENTMSGTYPE'].isin([1, 3])]
points_events['PTS'] = points_events['EVENTMSGTYPE'].apply(lambda x: 2 if x == 1 else 1)
points_events['PTS'] = points_events.apply(
lambda row: 3 if row['EVENTMSGTYPE'] == 1 and '3PT' in str(row['HOMEDESCRIPTION']) else row['PTS'], axis=1
)
game_stats = points_events.groupby(['PERIOD', 'PLAYER1_NAME'])['PTS'].sum().reset_index()
game_stats['GAME_ID'] = game_id
all_player_stats.append(game_stats)
time.sleep(0.6)
df = pd.concat(all_player_stats)
# 4. Get PLUS_MINUS and TEAM from box score
plus_minus_list = []
for game_id in game_ids:
box = boxscoretraditionalv2.BoxScoreTraditionalV2(game_id=game_id)
pm_df = box.get_data_frames()[0][['PLAYER_NAME', 'TEAM_ABBREVIATION', 'PLUS_MINUS']]
pm_df['GAME_ID'] = game_id
plus_minus_list.append(pm_df)
time.sleep(0.6)
pm_df = pd.concat(plus_minus_list)
# 5. Merge scoring and plus-minus data
merged = df.merge(pm_df, left_on=['PLAYER1_NAME', 'GAME_ID'], right_on=['PLAYER_NAME', 'GAME_ID'], how='left')
# 6. Calculate avg 4th quarter points
q4 = merged[merged['PERIOD'] == 4]
avg_q4_pts = q4.groupby('PLAYER1_NAME')['PTS'].mean().reset_index()
avg_q4_pts.rename(columns={'PTS': 'avg_4Q_pts'}, inplace=True)
# 7. Full summary
summary = merged.groupby(['PLAYER1_NAME', 'TEAM_ABBREVIATION']).agg(
total_pts=('PTS', 'sum'),
avg_pts_per_game=('PTS', 'mean'),
avg_plus_minus=('PLUS_MINUS', 'mean'),
games_played=('GAME_ID', 'nunique')
).reset_index()
# 8. Merge 4th quarter points
summary = summary.merge(avg_q4_pts, on='PLAYER1_NAME', how='left')
summary_kings = summary[summary['TEAM_ABBREVIATION'] == 'SAC']
summary_mavs = summary[summary['TEAM_ABBREVIATION'] == 'DAL']
print(summary_kings)
print(summary_mavs)
# 9. Plot: 4Q points vs Plus/Minus
plt.figure(figsize=(10, 6))
sns.scatterplot(data=summary, x='avg_4Q_pts', y='avg_plus_minus', hue='TEAM_ABBREVIATION', s=100)
# Add player labels
for _, row in summary.iterrows():
plt.text(row['avg_4Q_pts'] + 0.1, row['avg_plus_minus'], row['PLAYER1_NAME'], fontsize=9)
plt.title('SAC vs DAL — 4th Quarter Points vs Plus/Minus')
plt.xlabel('Avg 4Q Points')
plt.ylabel('Avg Plus/Minus')
plt.legend(title='Team')
plt.grid(True)
plt.tight_layout()
plt.show()here is the result:
PLAYER1_NAME TEAM_ABBREVIATION total_pts avg_pts_per_game avg_plus_minus games_played avg_4Q_pts
0 Alex Len SAC 4 2.000000 0.000000 1 2.0
4 De'Aaron Fox SAC 34 8.500000 17.000000 1 10.0
5 DeMar DeRozan SAC 72 6.545455 15.545455 3 9.0
6 Domantas Sabonis SAC 37 4.111111 7.888889 2 3.0
7 Doug McDermott SAC 4 4.000000 -2.000000 1 4.0
9 Isaac Jones SAC 6 6.000000 -6.000000 1 6.0
11 Jake LaRavia SAC 8 2.000000 -5.000000 2 2.0
13 Jonas Valančiūnas SAC 28 4.666667 7.000000 2 6.0
15 Keegan Murray SAC 28 4.666667 15.000000 3 2.0
16 Keon Ellis SAC 13 2.600000 -9.400000 2 2.0
18 Kevin Huerter SAC 3 3.000000 11.000000 1 NaN
21 Malik Monk SAC 43 3.909091 11.545455 3 2.0
22 Markelle Fultz SAC 6 6.000000 1.000000 1 6.0
29 Trey Lyles SAC 24 3.428571 14.000000 3 5.0
30 Zach LaVine SAC 34 5.666667 17.000000 2 NaN
PLAYER1_NAME TEAM_ABBREVIATION total_pts avg_pts_per_game avg_plus_minus games_played avg_4Q_pts
1 Brandon Williams DAL 10 5.000000 -6.000000 1 6.0
2 Daniel Gafford DAL 12 4.000000 -5.000000 2 NaN
3 Danté Exum DAL 14 2.800000 -11.000000 2 3.0
8 Dwight Powell DAL 2 2.000000 -10.000000 1 NaN
10 Jaden Hardy DAL 19 3.166667 -2.666667 3 NaN
12 Jazian Gortman DAL 6 3.000000 10.000000 1 4.0
14 Kai Jones DAL 22 5.500000 -9.000000 1 6.0
17 Kessler Edwards DAL 12 3.000000 -11.000000 2 4.0
19 Klay Thompson DAL 22 4.400000 -17.200000 2 3.0
20 Kyrie Irving DAL 38 6.333333 -6.833333 2 11.0
23 Max Christie DAL 25 3.571429 8.428571 2 2.5
24 Naji Marshall DAL 29 3.222222 1.444444 2 3.0
25 Olivier-Maxence Prosper DAL 25 4.166667 -13.000000 2 4.0
26 P.J. Washington DAL 26 8.666667 0.000000 1 NaN
27 Quentin Grimes DAL 8 2.666667 -18.000000 1 4.0
28 Spencer Dinwiddie DAL 61 6.100000 -1.800000 3 6.0
From those stats, I can tell that both team has very big injury isusse and a lot of roster change this season, for example, the best choice for the kings are De'Aaron Fox and DeMar DeRozan, but Fox got traded away so there's only DeRozan left. Another good player for the kings is Malik Monk, but he is injuried. But overall, most of the players have a big positive average +/- rate comaparing to the Mavs.
The kings wins easily, Demar DeRozan is the key player to secure the win, he is just unstopable to the MAVS. As for the MAVS, the only unknown X factor is Anthony Davis, he is a new mavs but got injured for most of his time with the Mavs, so his return and how he preforms would be the main reason why Mavs could have a chance to win.
Through this project, I demonstrated my ability to build a complete data pipeline—from collecting live sports data using APIs to performing detailed exploratory analysis and visualizing player and team performance trends. I combined technical skills in Python, pandas, and matplotlib with domain knowledge to extract actionable insights and make data-driven predictions. The project showcases my capacity to think like an analyst, communicate strategic recommendations, and deliver value under tight deadlines — all skills I can bring to a data-driven team.