Stars Lab · Workshop · v1.38

Workshop

Deeper stats behind the Stars desk.

Calibration, trust, and method instruments — secondary to the day-to-day desk. Proof first; proxies stay labeled.

Back to desk Calibration Trust Thesis note

Live

12

modules with data · 1 building · 1 roadmap

Proof

1.16pp

market MAE %cap · n=349 · ~65% PR-verified · 2025-26

Proxy

xWAR

contribution · aging · outlook — labeled, not licensed GAR

01 · Lead

Proof

Measure error, refuse thin AAV, show the engine.

live

Calibration & misses

Market MAE 1.16 pp · n=349

On a market cohort, how wrong is fair AAV — and are residuals data or model?

Next · Weekly MAE changelog after each dial.

live

Trust & sample

League PR-verified ~65%

Which scores are too thin / estimated / fallback to shout about?

Next · Public weak-score audit that updates with every seed.

02

Market

UFA demand, comps, structure, and ex-ante deal quality.

live

UFA class

Fair %cap × projected FA-summer ceiling — model band, not rumor.

Which players entering the next FA summers carry the most projected cap demand — and what might they cost?

Next · Age-curve adjusted bands; peer comp anchors at signing.

live

Market comps

Peer %cap board — negotiation language, not vibes.

What do similar age / pace / position peers get paid — and should fair move?

Next · League comps explorer + residual vs peer-median %cap board.

03

Science

Aging, contribution proxies, goalies, outlook — labeled as such.

live

Aging curves

Age factor by position — term risk lives here.

How should fair AAV move with age by position — and do residuals blow up late?

Next · Role-conditioned curves (1F vs 3F), not just F/D/G.

live

xWAR proxy

Transparent contribution proxy — not EH GAR, labeled as such.

Can an open WAR-style proxy (not licensed GAR) stabilize mid-star fair AAV?

Next · Surface offense/impact/ST/two-way parts on player research view.

04

Ops

Aggregate value into roster decisions.

building

Team asset maps

Roster rollups — value research in ops language.

Where is roster surplus / deficit concentrated — and who is the team MVP by value?

Next · Richer Stars desk pressure strip (locked %, FA count, renewals).

How to work this

Start in Proof — improve market MAE or trust coverage before chasing new proxies.
Promote a building module only when it has a number you can defend.
Do not start roadmap modules until the 90-day ex-ante test has a write-up.

How Worth runs: methodology · Worth · misses · Stars desk. Charter: RESEARCH.md

live

Value Score

Engine v1.38

Given same-season impact and AAV, is this mid-contract deal rich, fair, or a bargain?

Next · Signal cards that expose note-parsed inputs (adjxG, TOI, rel).

live

Ex-ante deal quality

Gold-date as-of holdout n=321 — MAE 1.48 pp / $1.30M at signing.

At signing, could open data have flagged deals that later look bad?

Next · Expand remaining proxy dates + optional frozen method_version snapshots at as-of.

live

ELC / bridge / RFA

Separate CBA structure from skill before calling a bargain.

Which loud underpays are structural (ELC/bridge) vs true market bargains?

Next · Bridge vs UFA pricing residual study.

live

Goalie value

G cohort MAE 1.04 pp

Can GSAX + workload price goalies without one hot season → $12M fair?

Next · Regression-to-mean prior + multi-season GSAX blend.

live

Outlook (next year)

Pace bands + age-at-season path — decision support, not a crystal ball.

Given rates, age curve, and usage — how should this player look next season?

Next · Role/injury context for the largest age-aware band misses.

live

Usage & roles

Tough-minute skaters clear more fair %cap in matched production bands.

Do tough-minute players and sheltered scorers price differently at similar production?

Next · Test whether a tiny usage nudge improves market holdout MAE.

roadmapthesis

Playoff impact

Tiny samples, loud rates — capped skater multiplier plus stored official goalie playoff CSV rows.

How much should playoff rate/depth move fair AAV without overfitting hot runs?

Next · Audit skater + goalie playoff nudges vs regular-season-only scores.