Home/API Docs

RAPM Data API

Programmatic access to Regularized Adjusted Plus-Minus data for every NBA player. Plug our player impact metrics directly into your models.

What is RAPM?

Regularized Adjusted Plus-Minus (RAPM) is a ridge-regression-based metric that isolates a player's individual contribution to their team's point differential per 100 possessions, controlling for teammates and opponents on the court.

Unlike box-score stats, RAPM captures defensive impact, spacing effects, and other contributions that don't show up in traditional statistics. Our implementation uses L2 regularization (ridge regression) with Bayesian priors derived from multi-season historical data.

Key Fields

blended — Weighted combination of rolling + prior
rolling — Current-season-only RAPM
prior — Multi-season historical RAPM
o_rapm — Offensive RAPM per 100 possessions
d_rapm — Defensive RAPM per 100 possessions
possessions — Sample size
lambda — Regularization weight (0–1, higher = more prior)

How It's Calculated

• Ridge regression on lineup-level point differential data
• Separate offensive and defensive targets
• Bayesian prior from 3+ seasons of historical data
• Lambda controls prior weight (low possessions → more prior)
• Updated weekly with latest game data

Endpoint

GEThttps://www.predictium.ai/api/rapm

Query Parameters

teamFilter by team abbreviation (e.g., PHI, LAL, BOS)

min_possessionsMinimum possessions threshold (e.g., 1000). Filters out low-sample players.

Example Request

curl "https://www.predictium.ai/api/rapm?team=PHI&min_possessions=1000"

Example Response

{
  "snapshot_date": "2026-02-09",
  "season": "2025-26",
  "player_count": 472,
  "players": [
    {
      "player_id": "203954",
      "player_name": "Joel Embiid",
      "team": "PHI",
      "position": "C",
      "rapm": {
        "blended": 4.2,
        "rolling": 3.8,
        "prior": 4.5,
        "o_rapm": 2.1,
        "d_rapm": 2.1,
        "possessions": 15000,
        "lambda": 0.88
      }
    }
  ]
}

Caching

Responses include Cache-Control: public, max-age=3600. RAPM data updates weekly, so caching for up to 1 hour is recommended.

Ready to integrate?

The RAPM Data API subscription gives you full programmatic access at $19.99/mo. Subscribe to both Predictions + Data API for $29.99/mo.

Get Started