🏀 NBA Causal Intelligence

Overview

NBA game, season, draft, and career data spanning 1996–2023 from a 2.2 GB SQLite database. The logic engine interrogates 61,000+ game records across six diagnostic dimensions to trace exactly why a game or career turned.

Data Structure

Entity	Records	What It Captures
TeamGame	61,000	Per-game team performance: scoring, shooting, rebounds, turnovers
TeamSeason	772	Aggregated season records: win rates, offensive/defensive ratings
DraftPick	1,656	Draft record + career outcome: busts, overperformers, steals
PlayerCareer	1,941	Career summary: undrafted stars, journeyman patterns

Causal Reasoning Model

17 cause types across six diagnostic dimensions:

• Scoring: offensive collapse, poor shooting nights, free-throw drought
• Shooting: field goal %, three-point %, efficiency breakdown
• Turnovers: excessive giveaways, live-ball turnovers
• Rebounding: board differential, second-chance points allowed
• Home Court: road performance drop, hostile venue effects
• Draft & Career: bust indicators, late-round outperformers, undrafted gems

Sample Questions

• "Why did the Lakers offense vanish against the Celtics on 2022-12-13?"
• "What specific variable triggered the Warriors 2019-20 collapse?"
• "Was this draft pick outcome due to usage rate or shooting mechanics?"

Prescriptive Recommendations

Ask what to do: get action recommendations with cost-benefit analysis. The engine recommends shooting drills for teams with scoring collapses, turnover reduction programs for sloppy play, and defensive scheme changes for teams allowing too many points.

• "What should a team do to recover from a shooting collapse?"
• "What actions can reduce turnovers for teams that keep giving the ball away?"