Quant Lab
as_of: 2026-09-17
status: DRAFT for Methods hub
seat: War Room Quant Lab
product_rule: Research packs only. Cards stay short. Depth lives here and in feature docs.
One sentence
Proprietary analytics seat: design features, algorithms, backtests, and calibration loops that Power / Market / Tape / Treasurer consume — never Officials, never seal rewrites.
Inputs
| Layer | Examples | Label |
|---|---|---|
| Architect facts | wait_core IR, odds facts, weather | fact |
| Historical results | Scorekeeper ledger, CLV | fact |
| Feature dictionary | proprietary_feature_dictionary_v1 | roadmap |
Never: our shop into Power blind path, Covers/FN as our edge, mid-week Officials.
Outputs
- Research packs under
data/features/ - Methods + backtest notes under
docs/ - Calibration reports with Scorekeeper
Kill criteria
- Feature not reproducible from labeled facts
- Pack would force a seat to break seal independence
- Arithmetic re-label of existing Power/Market output with no new signal
Calculus backlog (day one)
See ../proprietary_feature_dictionary_v1.md — prioritize opponent-adjusted EPA, joint sims, priced-gate curves, CLV/steam, Kelly correlation.
Methods v1 — filed by the seat. Public site hides private shop brand names.
Research packs
- opponent_adjusted_epa — Week 2 values READY · backtest WAIT · Power blind-seal · build plan · contract
Packs land here when Quant Lab files them. Outlines are not results.