Model Transparency

Methodology

True North Performance Lab is built around readable hockey operations models. The goal is not to hide behind complexity, but to make model assumptions, decision logic, limitations, and use cases clear enough for front-office style evaluation.

Model Layer

True North Rate

A first-layer offensive production indicator.

TN Rate combines goal production and point-production pace into a readable offensive contribution signal. It is designed for quick comparison, leaderboard generation, and profile-level context rather than as a complete player-value metric.

Simplified Logic

Goals/Game × 1.5 + Points/Game × 1.0

Model Layer

Projection Logic

A forward-looking estimate of future production.

Projection outputs blend current scoring pace, age-curve assumptions, regression logic, and role context. The goal is not to predict a single exact outcome, but to create a reasonable decision-support range.

Simplified Logic

Current Production × Regression × Age Curve × Role Adjustment

Model Layer

Contract Value

A market-value estimate tied to projected production.

The contract model translates projected point production, age, and position into an estimated annual value range. This helps frame whether a player profiles as underpriced, fairly valued, or risky relative to expected output.

Simplified Logic

Projected Output + Age Context + Position Adjustment

Model Layer

Surplus Value

A simplified estimate of value above or below market cost.

Surplus value compares estimated player value against an assumed market cost. Positive surplus suggests a potentially efficient asset; negative surplus suggests possible contract or acquisition risk.

Simplified Logic

Estimated Value − Market Cost

Model Layer

Player Comparables

A similarity layer for player profile matching.

Comparable players are identified using production rate, projected points, age, team, and position context. The goal is to support scouting conversations, not to claim two players are identical.

Simplified Logic

Similarity = Production + Projection + Age + Position Context

Model Layer

Acquisition Flags

A front-office shorthand for player decision context.

Acquisition flags translate model outputs into plain-language scouting and roster-building signals such as development target, undervalued asset, contract risk, or core offensive piece.

Simplified Logic

Age + Projection + Surplus Value + Roster Context

Operating Philosophy

Decision-support, not decoration.

TNPL is designed to answer practical hockey operations questions: Which players are producing efficiently? Which profiles project forward? Which contracts may create surplus value? Which players resemble each other? Which acquisition targets deserve a deeper look? The interface exists to support those questions, not simply to display statistics.

Model Limitations

What this does not claim.

Current models prioritize transparency and readability over black-box complexity.

Publicly available data limits what can be inferred about deployment, tactics, injuries, and internal team evaluation.

Outputs should be treated as decision-support signals, not absolute truth.

Additional play-by-play, tracking, contract, and context-adjusted data would improve future model depth.