NBA games

DAL @ LAC

NBA Intuit Dome ·Final
96110
Final
Q1Q2Q3Q4T
DAL2625242196
LAC25312727110

Statlab take

baseline_gbm
Last 3 (LAC) / 2 (DAL) games this season
DAL away
home LAC
21% 79%

GBT team-winner trained on ~3 seasons of NBA games (68.8% holdout accuracy). Features: 10-game rolling form / points-for / points-against diff + rest days + back-to-back flags.

Box score

Player MIN PTS REB AST
Quentin Grimes 22 20 4 2
Jaden Hardy S 26 16 3 4
Jazian Gortman S 28 13 2 0
P.J. Washington S 19 10 3 0
Brandon Williams 18 9 1 2
Dereck Lively II S 24 8 7 5
Olivier-Maxence Prosper 10 7 1 1
Naji Marshall 27 6 4 3
Daniel Gafford 17 4 7 1
Spencer Dinwiddie 16 3 1 1
Player MIN PTS REB AST
Kevin Porter Jr. 19 18 5 6
Norman Powell S 22 13 1 3
Kai Jones 24 12 4 0
Derrick Jones Jr. S 23 12 4 0
Ivica Zubac S 24 11 12 1
James Harden S 23 10 2 12
Amir Coffey 18 9 2 1
Kris Dunn 20 9 2 0
Terance Mann S 20 6 2 1
Cam Christie 7 4 1 1

First basket

scorer · ROC 0.80
Trained on 67736 player-games
Actual
Dereck Lively II · Free Throw (1 pts) · 11:27 Q1
"Dereck Lively II makes free throw 2 of 2"
Most likely first scorer (top 10)
1 James Harden LAC
24%
2 Ivica Zubac LAC
23%
3 Amir Coffey LAC
19%
4 Derrick Jones Jr. LAC
18%
5 Terance Mann LAC
16%
6 Jordan Miller LAC
0%
7 Alondes Williams LAC
0%
8 Cam Christie LAC
0%
9 Kai Jones LAC
0%
10 Kevin Porter Jr. LAC
0%
Method (given top scorer)
3PT
34%
Dunk
22%
Free Throw
16%
Jumper
16%
Layup
13%

Per-player binary classifier — features: starter status, recent PPG/FGA/3PM, starter-rate, home/away. Probabilities normalized so the field sums to 100%.

Statlab projections

Trained on 68258 player-games; PTS MAE 4.7 (naive 6.8)
# Player Team Pred PTS Pred REB Pred AST
1 James Harden LAC 15.2 3.4 6.4
2 Jordan Miller LAC 11.1 4.1 2.2
3 Kai Jones LAC 10.0 7.2 1.7
4 Kobe Brown LAC 8.0 6.6 1.9
5 Kevin Porter Jr. LAC 7.8 3.0 3.2
6 Ivica Zubac LAC 7.3 5.4 1.4
7 Derrick Jones Jr. LAC 6.0 2.5 0.9
8 Amir Coffey LAC 5.9 1.8 0.8
9 Terance Mann LAC 5.6 3.4 0.9
10 Alondes Williams LAC 5.6 2.6 3.6
11 Cam Christie LAC 3.8 2.0 0.8

GBT MultiOutput regressor per player — 9 targets (PTS/REB/AST/STL/BLK/TO/MIN/FG/3PT). Features: rolling form last 10 games + opponent allowed averages + rest days. Sorted by predicted PTS.