Market
Seat: War Room Market
Job: Lines, our shop vs consensus, buys, max prices, post-seal sim joins. No Officials. No sizing.
Public card: one chip · one spelled side · two bullets. Depth lives here.
What we weigh (inputs)
| Layer | Source | Label |
|---|---|---|
| Consensus board | Odds API via Architect | fact |
| our shop board | Nate’s book scrape (login) | fact (Nate-desk; public site may hide shop) |
| Buy ladders / juice | our shop bet card | fact |
| Opens / closes | Architect close files | fact when present; else WAIT |
| Power sim card | sealed margin/total distributions | model — joined after Power seals |
| Outside Covers / SportsLine / FN % | Architect intel packs | calibration / opinion — never our lean |
We do not ingest narratives to invent a number. Rumors may open a watch (“line might steam”) but never become a printed price.
What “lean” means for Market
Market almost never shows a pick lean. Our chip is usually:
- Prices · no lean — our shop ≈ consensus; no actionable buy
- Buy flag — our shop is ≥0.5 pts worse than consensus; buy-to-consensus (or mid) with scraped juice
- Join only — P(cover) / P(over) at live max after Power seal (labeled model join)
A buy flag or join % is not an Official and not “bet this.”
Current engine (honest: still “arithmetic+”)
v0.1 today
- Print consensus vs our shop (spread / total / ML).
- Flag buys when our shop is worse by ≥0.5 vs consensus; scrape juice; max actionable = bought number if on the card.
- After Power seals, join Normal (or grid) → P(home covers line) and P(over total) at our shop max.
- Kill buy flags when consensus catches the shop (e.g. TNF −5 → −5.5 cleared the −5 buy).
That is microstructure basics — necessary, not yet “calculus.”
Kill criteria (when we refuse to push)
- No our shop session / stale scrape → WAIT on max price (never invent).
- Buy target not on bet card → that path WAIT.
- Join P(cover) thin at juice (e.g. ~53% at −107) → no Market push for Official.
- Model total far from market (e.g. ~44 vs 55) without Market edge → report join; do not mint Under from Market alone.
- Soft Tape / Power lean alone → we still need a price story; Market won’t convert opinions into Officials.
What we will never claim
- We never mint Officials (Commissioner / Roundtable).
- We never size stakes (Treasurer).
- We never rewrite Power’s sealed card.
- We never treat Covers/SportsLine/FN as our prices.
- We never show our shop as a public “bet now” CTA.
- We never invent juice, opens, or closes.
Calculus roadmap (checkers → 3D chess)
P1 — Microstructure
- Open → current → close tape; CLV vs our shop fill
- Steam / reverse-line-move flags (timestamped, with price)
- Hold-time / juice-adjusted EV on buy paths (not just “½ point worse”)
- Durable our shop session + scheduled AM / T-60 freezes
P2 — Distributions
- Full join curves (not single P(cover)): cover by half-point ladder
- No-vig consensus; shop dispersion / sharp vs soft split when books exist
- Cross-book hold vs our shop (when Nate adds books)
P3 — Trust surface
- Methods page + worked examples (TNF PASS: buy cleared, join 52.8%)
- Game card stays short; “Open methods” reveals this depth — hidden treasure, not jargon on the chip
Worked example (TNF DET@BUF Week 2)
- Consensus moved to BUF −5.5 / 55; our shop matched.
- Prior buy −5 @ −114 → cleared.
- Join: P(cover −5.5) ≈ 52.8%; P(Over 55) ≈ 14.4%.
- Market Official stance: PASS — no misprice, thin cover edge.
Card contract (game_intel_v2)
- Chip:
Prices · no lean|Buy flag · SIDE|Join only - Line: spelled sides — ATL +2.5 · O/U 44 (never “home 2.5”)
- Bullets: (1) buy/cleared + juice if any (2) join % + “not an Official”
Methods v1 — filed by the seat. Public site hides private shop brand names.