2025-26

Glossary

Definitions and formulas used throughout NetStats.

Player Metrics

Compound Archetype Profile

Definition: A three-part label summarizing a player across Scoring Type, Usage Type, and Passing Type — e.g. 'Mid+Paint / Ball-Dominant / Hub'. Each dimension is assigned independently by KMeans on real play-by-play and tracking data, so there is no constraint forcing a consistent 'big man' or 'guard' label across all three.

Pass Out Degree

Definition: Number of unique teammates a player passes to (with at least 6 passes).

Pass In Degree

Definition: Number of unique teammates a player receives passes from (with at least 6 passes).

Scoring Type

Definition: One of 6 shot-zone clusters (KMeans, k=6) describing where and how a player scores: Rim-Only, Paint-Dominant, Mid+Paint, Skip-Mid, Three-Level, or Perimeter. Clustered on shot zone percentages (rim %, paint non-RA %, mid %, above-break 3 %, corner 3 %) and zone efficiency across 2021-22 through 2024-25 seasons.

Usage Type

Definition: One of 3 ball-handling tiers (KMeans, k=3): Ball-Dominant (primary creator, handles heavily), Orchestrator (involved initiator — posts, passes, sets the table), or Catch-and-Go (quick-release receiver who barely dribbles). Clustered on touches per minute, dribbles per touch, time per dribble, and time of possession.

Passing Type

Definition: One of 4 network roles (KMeans, k=4): Hub (highest volume and assist rate — the true distributor), Initiator (moderate-to-high volume, pushes pace and initiates), Secondary (lower volume but contributes to ball movement), or Terminal (primarily a receiver and finisher, not a distributor). Clustered on passes made/received per minute, assist rate, and pass ratio.

Passing Influence (Network Centrality)

Definition: Eigenvector centrality — how connected you are to other high-volume passers. High means you're passing with the team's core ball-handlers. Low means you're on the periphery of the network regardless of how many passes you make. A player can make many passes but score low if those passes go to teammates who aren't themselves well-connected. The quintessential high-influence player is a point guard who exchanges heavily with other starters; a corner shooter who only receives from one creator scores low even with solid pass totals.

$$x_i = \frac{1}{\lambda} \sum_{j} A_{ij} \cdot x_j$$

Linking Role (Betweenness Centrality)

Definition: How often a player sits on the shortest passing path between two teammates who don't pass to each other directly. A high Linking Role score means the offense routes through this player to connect otherwise-disconnected parts of the network. Losing a high-betweenness player tends to fragment ball movement more than losing a high-volume but low-betweenness player.

$$C_B(v) = \sum_{s \neq v \neq t} \frac{\sigma(s,t|v)}{\sigma(s,t)}$$

Pass Variety (Entropy)

Definition: How evenly a player distributes passes across their teammates. High variety means the player spreads the ball unpredictably — hard to scout and creates defensive confusion. Low variety means passes are concentrated on one or two connections — easier to game-plan against but can indicate intentional role clarity (e.g., a point guard who consistently finds a specific shooter).

$$H = -\sum_{j} p_{ij} \cdot \log_2(p_{ij})$$

Team Metrics

Team Entropy

Definition: Shannon entropy of outgoing pass distributions. Measures unpredictability of ball movement.

$$H = -\sum_{i} \sum_{j} p_{ij} \cdot \log_2(p_{ij})$$

Flux

Definition: Weighted average change in shooting percentage from passer to receiver.

$$\text{Flux} = \sum_{i,j} p_{ij} \cdot (FG\%_j - FG\%_i)$$

Clustering Coefficient

Definition: Measures how interconnected teammates are via triangles, accounting for pass weights. Uses the geometric mean of edge weights between triples.

$$ c_u = \frac{1}{\deg(u)(\deg(u)-1)} \sum_{v,w} \left( \hat{w}_{uv} \hat{w}_{uw} \hat{w}_{vw} \right)^{1/3} $$

where the normalized edge weights are:

$$ \hat{w}_{uv} = \frac{w_{uv}}{\max(w)} $$

Degree Centralization

Definition: Measures how much passing activity is concentrated in a single player.

$$C_D = \frac{\sum_{i}(d_{\max} - d_i)}{(n-1)(n-2)}$$