🧪This is a one-person passion project, still very much in beta. Some roster data is off — fresh data drops Monday. Poke around, have fun, and bear with me!
NWSL NOTEBOOK

Data Sources

Where the numbers come from, what they cover, and how fresh they are.

Event Data

SPADL-converted event data covering every on-ball action — passes, carries, shots, tackles, interceptions, clearances, fouls, take-ons, crosses — for every NWSL match. The SPADL format (Soccer Player Action Description Language) standardizes event data from multiple providers into a unified action-based representation, making cross-season and cross-provider comparisons consistent.

Match & Schedule Data

Complete fixture records including final scores, half-time scores, venues, attendance figures, match week numbers, and competition stage. Covers regular season and playoff matches.

Match detail page showing score, AI insight, and tactical analysis
Match detail pages include AI-generated insights, pG momentum, key actions, and tactical contrast analysis

Player & Team Data

Player profiles including name, position, nationality, birth date, height, dominant foot, headshots, and jersey numbers. Team metadata including official names, short names, codes, and logos. Season rosters with appearances and goals.

Player profile page showing radar chart, impact metrics, form chart, and similar players
Player profiles combine VAEP, xT, radar charts, form trendlines, heatmaps, and similar player matching

Derived Analytics

Raw event data is transformed into higher-order metrics and models:

VAEP values — Per-action, per-match, and per-season aggregates of action value.
Expected Threat (xT) — Grid-based spatial value for carries and passes.
Team & player embeddings — 64/128-dimensional style vectors encoding tactical identity.
Rolling embeddings — Windowed embeddings for tracking tactical drift over time.
Archetype labels — GMM cluster assignments with AI-generated narratives.
CASLO signals — Coherence, friction, and pattern state time series.
Set piece analysis — Corner delivery quality and predicted shot probability.
Match insights — AI-generated headlines, verdicts, and tactical analysis.
Team profile page showing tactical identity, constellation map, and stats
Team profiles show tactical identity, archetype classification, embedding constellation maps, and season statistics

Coverage

Seasons2016 – 2025 (2026 in progress)
Matches~1,200+ regular season + playoffs
Actions~3M+ on-ball events
Players~1,500+
TeamsAll NWSL franchises

Data Freshness

Data pipelines run after each match week. The analytics database is populated by a separate pipeline system that processes raw event feeds, computes derived metrics, trains embedding models, and generates AI narratives. Some roster data and newer season data may lag — consistent with the beta nature of the project.

For details on how specific metrics are computed, see Metrics & Methodology.