X-NAV — one number for trade value

Front offices do not trade goals or Corsi — they trade assets. X-NAV (Extended Net Asset Value) exists because comparing a 24-year-old RFA winger against a 31-year-old defenseman with two years of term requires everything to be priced in the same currency: production, defense, gravity, age curve, contract surplus against a projected market, deployment difficulty, and team control.

The model is deterministic application code, not a language model. Every component is visible on a player's Value Breakdown, so a number can always be interrogated: how much is on-ice value, how much is the contract, how much is upside. Unsigned players are projected onto their fair-market AAV rather than valued against a zero cap hit — a pending RFA is not an infinite bargain.

Player Gravity — a modelled territorial field

Player Gravity v3 is a position-relative territorial influence index. It combines on-ice chance impact, transition proxies, and defensive suppression into an offensive-zone well, neutral-zone well, and defensive-zone dome. The rink field is a model visualization of those components, not a literal tracking map.

The current field force is a bounded display composite, not expected goals and not a direct measurement of defender attention. The public scope is MIXED SITUATIONS because v3 combines all-situations, 5v5, 5-on-4, 4-on-5, and regular-season EDGE aggregate inputs. Signal Stability is based mainly on agreement between current and baseline on-off values, with a legacy defenseman pair-driver adjustment; it is not a fitted portability model. Reliability is a 0–100 coverage/stability index, not a probability. Missing evidence shrinks the estimate toward neutral and lowers reliability.

X-NAV receives only the transition portion of the current Gravity model. Direct offensive production and defensive suppression are valued elsewhere. Weighting and calibration are Ledger-defined and versioned against the available league population.

Territorial Gravity v4 is a separate 5v5 expected-goal contract and diagnostic path. In v4 output, positive wells and domes represent estimated expected-goal impact in each phase of play. It will remain display-only and behind a disabled feature flag until an authorized event/shift dataset supports fitting, uncertainty intervals, held-out validation, and season-specific calibration. It does not currently affect X-NAV.

STRAND DNA — identity, not grades

Two 70-point wingers can be entirely different players. STRAND (Stylistic Trait & Rating Analysis for NHL Development) profiles how a player creates value — scoring pace, chance creation, suppression, usage trust, deployment difficulty — as an identity strand rather than a single grade, so roster fit and role redundancy become visible at a glance.

STRAND traits are normalized by position and rendered with league context. It answers the question a grade cannot: what does this player actually do?

The GM Audit — plausibility, not just math

A trade can be mathematically balanced and still be one no general manager would make. The GM Audit is the rule layer that thinks like a front office: clause and cap legality, retention mechanics, roster slots, timeline alignment, surplus gaps, and whether the deal fits each team's contention window.

The audit publishes its objections as named flags rather than a silent score, so a rejected deal always says why — and a verdict can be argued against on the record.

The Simulation Engine — consequences on the record

Armchair GM's three-year Cup Run exists because trades are hypotheses and seasons are experiments. The engine rolls the whole league forward — aging, retirements, breakouts, drafts, cap growth, AI cap compliance — so a deadline rental or a rebuild pivot is tested against consequences, not vibes.

Seeded randomness keeps runs replayable: the same decisions in the same season produce the same league, so outcomes are attributable to choices.

Data pipeline & acknowledgements

Nightly snapshots combine NHL API rosters and NHL EDGE tracking with MoneyPuck's public analytics and CapWages contract data, layered over multi-season baselines so single-season noise never masquerades as signal. In fixed-weight composites such as Gravity v3, missing evidence contributes no term, which shrinks the estimate toward neutral and lowers the reported reliability.

The Ledger stands on the shoulders of the public hockey-data community. Sincere thanks to the NHL, MoneyPuck, CapWages, and Hockey-Reference — their work makes independent analysis like this possible. Full source credits with links are in the Glossary's Data & Sources section.