NBA Power Board
SBIN Analytics · Example dashboard · 2025–26 regular season

NBA Power Board

The standings show who has won. This page asks who is actually good. Set your own weights for the ranking formula, step through the season week by week, and see which teams won more or fewer games than their net rating predicts.

Power rankings

Each factor is converted to a z-score across all 30 teams, then blended using your weights. Click a team to follow it through every chart.

#±TeamScoreRecordMarginLast 10SOS

How the rankings moved

Each team's weekly rank under your current weights. Change a slider and the whole season is ranked again. Hover over a line to see the team.

Expected wins vs. net rating

Net rating is points scored minus points allowed per 100 possessions. The line shows how many wins that rating usually produces, fitted on these 30 teams. Teams above the line won more than their play suggests, usually by winning close games. Teams like that tend to regress.

Won more than expectedWon fewer than expected
Wins above expected
Teamfewer ← → moreDiff≤5 pts

Methodology

Data

All 1,230 regular-season games of 2025–26 (Oct 21, 2025 to Apr 12, 2026) and team net ratings from Basketball-Reference. The NBA Cup final is excluded because it does not count in the standings. Every team's win total was checked against its official record.

Composite score

For each week, each factor is standardized as z = (x − mean) / sd across the league. The score is Σ wᵢ·zᵢ / Σ wᵢ, so 0 is a league-average team and +1 is one standard deviation better. A team needs 5 games before it is ranked.

Expected wins

A least-squares line, wins = a + b × net rating, fitted on the season's 30 teams. The gap between actual and expected wins is mostly close-game results, which do not carry over well from one season to the next. The last column shows each team's record in games decided by 5 points or fewer.

What to borrow for your sport

  • Start from a question: "Who is actually good?"
  • Make the weights visible and adjustable, and be ready to defend your defaults.
  • Separate performance from results (xPts vs. points in soccer).
  • Add a time slider so the page tells the story of the season.
  • Next step: backtest. Do early-season rankings predict second-half records better than the standings do?