NRCL-98 Publish sports product planning baseline

docs/meta.json

@@ -13,6 +13,7 @@
     "analysis",
     "trading",
     "workflows",
-    "examples"
+    "examples",
+    "research"
   ]
 }

docs/research/2026-09-09-paper-workflow-verification.mdx

@@ -0,0 +1,45 @@
+---
+title: Local paper workflow verification
+---
+
+# Local paper workflow verification — 2026-09-09
+
+Outcome: existing implementation passes the customer workflow at the installed Neural wheel / Vaticor bridge layer. No blocking defect found; no source changes needed.
+
+## Candidate and release state
+
+- Repository: [IntelIP/vaticor](https://github.com/IntelIP/vaticor)
+- Branch: `main`; clean before and after verification.
+- Exact HEAD: `8af4b3bd7e8f519adcfe274f93c643cac3a41449`.
+- GitHub PR #40, “NRCL-97 Prove saved paper replay and restart scenarios”: live GitHub reports MERGED at `2026-09-08T02:11:35Z`, merge commit matches HEAD. https://github.com/IntelIP/vaticor/pull/40
+- Installed Neural source commit: `9b0899d433eca225b586fb5d50d68e803b6aa22a`.
+- Installed wheel SHA-256: `423a666bb44b5f39171b4295d84c9c11322233449273e64501ec1b124722718e`.
+- Runtime ID: `4a06caf4f28c22acbfc1440d7642afd9f7aafbad546ef7d317d6d6459f958ef3`.
+
+## Direct verification
+
+Executed the existing integration checks:
+
+```sh
+bun test --timeout 60000 tests/unit/local-paper.test.ts tests/local-paper.integration.test.ts
+```
+
+Result: PASS; 15 tests; 159 assertions; zero failures; 42.94 seconds.
+
+Verified using isolated temporary queues and the installed Neural wheel:
+
+- Save and deduplicate immutable recordings; reject malformed, incomplete, sequence-gap and oversized input.
+- Validate a Kalshi price-rule strategy; submit a saved recording; execute through Neural; reopen persisted strategy, assumptions and results.
+- Clone original inputs with a strategy threshold change; preserve source results and fee/cash assumptions; compare two completed compatible experiments; reject incompatible comparison inputs.
+- Recover queued work after a child worker exits after simulation and before commit; publish one result; charge fixture fees once (two fills at 0.02 each); preserve history and inspect result from subsequent bridge processes.
+- Reconnect, expiry, insufficient depth and sequence-gap scenarios; origin/request validation and runtime identity rejection.
+
+The synthetic round trip yields cash 10.56 from starting cash 10 and realized PnL 0.56. This is fixture behavior, not measured trading performance.
+
+## Boundaries and next PM action
+
+- NRCL-97 is already implemented and merged. Reconcile its Plane state rather than assign duplicate implementation.
+- Frontend click-through and full Next.js server restart were not repeated this run. The direct check covers persistent workflow through separate bridge processes, including forced worker death, but is not a fresh browser usability acceptance.
+- Existing docs explicitly limit the simulator: Kalshi only, local and loopback-only, synthetic fixtures, fill-or-kill depth handling; no partial-fill simulation, live feed compatibility, queue priority, slippage or certified exchange fee accuracy.
+- The original verification performed no hosted deployment, live trading, authenticated venue call or paid experiment.
+- Verification required no product code changes. This report preserves the original evidence.

docs/research/2026-09-09-sports-product-reconciliation.mdx

@@ -0,0 +1,46 @@
+---
+title: Sports product reconciliation
+---
+
+# Sports product reconciliation — September 9, 2026
+
+## Outcome
+
+The three authorized workstreams are complete: Plane reconciliation, local paper workflow verification, and second-venue research. No product code changes were necessary. This report records the PM decision and links evidence; it is not a release or deployment receipt.
+
+## Product goal and boundaries
+
+Help sports strategy developers compare matched propositions and settlement differences, test strategies reproducibly, and operate them across supported U.S. venues. Neural owns normalized contracts, recording, replay and execution; Vaticor owns user workflows and hosted operations. Initial user is provisionally the strategy developer. MLB pregame moneylines are a research sample, not a committed launch sport.
+
+## Plane reconciliation
+
+- NRCL-92–94 reconciled to Done after GitHub verified PR #38 merged, with existing local acceptance evidence.
+- NRCL-95–96 reconciled to Done after GitHub verified PR #39 merged, with saved-recording and clone evidence.
+- NRCL-97 reconciled to Done after PR #40 merge verification and current direct integration tests.
+- NRCL-80–85 canceled as superseded paper-to-action and retirement work. Historical records retained; no infrastructure, repository or website removed.
+- Project description updated to sports-first and conditional second-venue selection.
+- Missing native dates/estimates on NRCL-92–96 and NRCL-84–85 were explicitly assigned to the September 9 reconciliation scope (1 point); they do not represent reconstructed historical build estimates or original delivery promises. Existing populated metadata was preserved.
+- Confirmed final non-archived state: 18 Done, 7 Canceled, 3 Backlog, zero Started. All changed records were read back.
+
+## Work now captured
+
+| Story | State | Result or next outcome |
+|---|---|---|
+| [NRCL-98](https://app.plane.so/intelligent-intellectual-property/projects/7ffb1939-48c5-4a7a-a4be-6cd283fc8274/issues/64f56165-13e3-45bb-aad7-0cee0e1ff0aa) | Done | Second-venue feasibility brief and bounded public evidence |
+| [NRCL-99](https://app.plane.so/intelligent-intellectual-property/projects/7ffb1939-48c5-4a7a-a4be-6cd283fc8274/issues/730de44e-7f65-4c6b-95f7-3501f4e00d71) | Backlog | Match sports propositions and expose settlement differences |
+| [NRCL-100](https://app.plane.so/intelligent-intellectual-property/projects/7ffb1939-48c5-4a7a-a4be-6cd283fc8274/issues/8af2a8b1-5b4c-4ae2-8374-63ced7293206) | Backlog | Run an unchanged paper strategy on Polymarket US recordings |
+| [NRCL-101](https://app.plane.so/intelligent-intellectual-property/projects/7ffb1939-48c5-4a7a-a4be-6cd283fc8274/issues/9b7fc695-f593-4c09-9cec-eccb09723fde) | Backlog | Resolve venue access and compare pregame execution costs |
+
+NRCL-100 and NRCL-101 have native blocked-by links to NRCL-98 and NRCL-99. NRCL-98 is complete; rule matching remains the open dependency. Drafting access questions can proceed before the matched-price portion of NRCL-101. Future backlog was captured, not committed or implemented in this turn.
+
+## Direct verification
+
+[Paper workflow report](./2026-09-09-paper-workflow-verification.mdx): clean Vaticor main 8af4b3bd7e8f519adcfe274f93c643cac3a41449, installed Neural source 9b0899d433eca225b586fb5d50d68e803b6aa22a. Existing local tests: 15 passed, 159 assertions, zero failures. Covers saved recording reuse, strategy configuration, execution, clone/compare, reopening and worker crash recovery. No fresh browser walkthrough or full application restart repeated. Kalshi synthetic paper only; no hosted/live readiness claimed.
+
+## Venue decision
+
+[Research brief](./2026-09-09-sports-venue-decision.mdx): Polymarket US is the provisional second data/paper adapter on access and integration readiness. Novig remains an active candidate. Documented MLB forfeit and postponement differences mean equal teams/date do not establish equivalent economic contracts. No live venue winner or profitability claim.
+
+## Next pull
+
+Refine NRCL-99 for one agreed sport and exact contract sources. Once the rule contract is ready, NRCL-100 adapter work and NRCL-101 access/cost validation can run beside each other. Keep at most three started stories; keep hosted paper and live execution as subsequent bounded milestones.

docs/research/2026-09-09-sports-venue-decision.mdx

@@ -0,0 +1,62 @@
+---
+title: Sports venue decision brief
+---
+
+# Sports venue decision brief — Neural / Vaticor
+
+Research date: September 9, 2026 America/New_York. Public API observations occurred September 10, 2026 UTC (still September 9 Eastern). Decision: choose the next bounded data/paper integration alongside the existing Kalshi work. MLB full-game pregame moneylines are a provisional research sample, not a committed product-sport decision.
+
+## Recommendation
+
+**Advance Polymarket US as the provisional second data/paper adapter. Keep Novig as an active candidate whose API access and commercial fit must be resolved before a live adapter commitment.** This is an integration-readiness recommendation, not a finding that Polymarket US offers better prices or profitability.
+
+Reasons: Polymarket US offers working unauthenticated sports discovery and books; Neural already references Polymarket US normalization examples; its current docs explicitly describe an ISV route for hosted customer workflows. Novig has meaningful observed MLB activity and attractive documented pregame trading fees, but the current public evidence does not establish a Vaticor customer-account automation model. Final live venue selection remains open.
+
+Local fit verified in `README.md`: stable product is a deterministic, dependency-free kernel beneath Vaticor; Kalshi data uses decimal prices and fractional depth; existing Polymarket US examples are listed. `docs/trading/recorded-paper.mdx` states recordings currently support Kalshi only. No implementation or live readiness inferred from those descriptions.
+
+## Facts that change the plan
+
+| Dimension | Polymarket US | Novig | Implication |
+|---|---|---|---|
+| Read access | Public gateway exposes events, sports, books and display-price history without a key; direct discovery/book requests worked | Public daily market/trade CSVs require no authentication; NBX trading credentials must be requested | Polymarket US has the clearer immediate live-data path |
+| Hosted customer model | Documented ISV/IB integration: firm acts for onboarded retail participants; signed agreements and RSA/JWT credentials; preproduction test funds | Affiliate documentation supports odds display and deeplinks; NBX credentials alone do not establish hosted customer-account rights | Ask which commercial model fits Vaticor; do not assume personal keys authorize a hosted service |
+| Pregame straight-contract fee | Current taker formula: `0.06 × contracts × p × (1-p)`; $1.50 for 100 at $0.50 | Straight fees apply only to fills matched during `OPEN_INGAME`; no trading fee for a pregame fill | Compare executable cost, not fee schedule alone |
+| History | Display-price endpoint plus public time-and-sales and end-of-day reports | Daily market census and executed-trade CSVs | Neither price history nor daily volume provides historical queue/depth replay |
+
+Sources: [Polymarket US API](https://docs.polymarket.us/api-reference/introduction), [ISV eligibility](https://docs.polymarket.us/partners/partner-types/isvs), [partner responsibilities](https://docs.polymarket.us/partners/your-role), [partner onboarding](https://docs.polymarket.us/partners/get-connected/onboarding), [Novig authentication](https://docs.novig.com/api-reference/authentication), [Novig affiliate integration](https://docs.novig.com/affiliates/overview), [Polymarket US fees](https://docs.polymarket.us/fees), [Novig fees](https://docs.novig.com/fees).
+
+Polymarket US's partner route is more concrete than our earlier blanket statement that hosted access was unknown for both candidates. It is **not** evidence that IntelIP is approved or that unattended customer strategy execution and connection to pre-existing accounts are included. Its documented partner flow also includes customer onboarding and funding integration, which may exceed our initial model. Separately, institutional market-data onboarding requires a Market Data Agreement; public reachability does not establish storage, redistribution, replay or resale rights. [Data onboarding](https://docs.polymarket.us/data-guide/onboarding)
+
+## Contract equivalence has a demonstrated blocker
+
+The current Novig **MLB Winner Series**, public Appendix A, differs from Polymarket US's sports guidance:
+
+- **Forfeit without on-field result:** Novig settles to void; Polymarket US awards the winner of the forfeit.
+- **Postponement:** Novig's ordinary window is 48 hours, or 45 days for postseason, subject to formal date updates and specified exceptions. Polymarket US generally uses rescheduling before contract expiration, described as typically two weeks.
+- **Exceptional payout:** Novig voids use its defined fair-value process. Polymarket US uses last fair market price for specified contingencies. Neither should be modeled as an automatic refund or always-$0/$1 settlement.
+
+Sources: [Novig contract directory — MLB Winner Series, Appendix A pages 2–6](https://support.novig.com/en/articles/16083642-contracts), [Polymarket US sports FAQ](https://docs.polymarket.us/faqs/sports-faqs).
+
+**Matching teams, date and winner proposition is insufficient to claim economic equivalence.** First milestone should surface matched sporting propositions with explicit rule differences. Strict equivalence requires reviewing the exact listed contracts on both venues, including Kalshi; this study did not establish a strictly equivalent three-venue pair. Do not label resulting price gaps risk-free arbitrage.
+
+## Bounded public-data evidence
+
+**Novig:** Manifest lists both trade and market files for all seven dates September 2–8. September 8 `markets.csv` downloaded successfully (6,453,454 bytes). Filtering exact `reportTicker == MLB-MONEY` produced **33 listed market IDs, 31 with positive daily volume, and 4,591,596.11 contracts traded**. Sum uses decimal `dailyVolume`, one row per market. This includes pregame and live activity; the census contains markets beyond a single day's games and lacks event/team/start-time metadata. It demonstrates activity, not executable liquidity or pregame volume. The trade CSV exceeded the bounded 30 MB request cap, so no trade-level result is claimed. [Manifest](https://data.novig.com/reporting/trade-data/index.json), [September 8 market report](https://data.novig.com/reporting/trade-data/2026-09-08/markets.csv), [schema](https://docs.novig.com/api-reference/trade-data)
+
+**Polymarket US:** The September 9 research session successfully accessed unauthenticated MLB discovery and a full-game winner book. This is a technical reachability observation only. No immutable book response was retained; exact historical quotes and depth are therefore omitted. The mutable endpoints cannot reproduce that observation or establish comparative liquidity. [Discovery endpoint](https://gateway.polymarket.us/v2/leagues/mlb/events?limit=3), [API documentation](https://docs.polymarket.us/api-reference/introduction)
+
+History distinction: Polymarket US `longPrice`/`shortPrice` are book-derived display prices, may sum above one, and are not trade prints. Its time-and-sales report has price/size/time/symbol without aggressor side. Novig trades have maker and taker rows; count only taker rows for trade count/notional volume. None is sufficient alone for fill-quality backtests. [Price history](https://docs.polymarket.us/api-reference/price-history/get-price-history), [time-and-sales](https://docs.polymarket.us/faqs/execution-tape), [daily report](https://docs.polymarket.us/faqs/eod-reporting)
+
+## Concrete work to pull next
+
+1. **Sports proposition matching and settlement differences.** Customer can compare the same game/outcome while seeing material rule differences. Define versioned event/team identities, doubleheader game number, full-game versus segment, extra innings, listed rule source/version, and `compatible / different / unknown` result. Acceptance fixtures must cover ordinary winner, forfeit, postponed game, duplicate team matchup and partial-game exclusion. Unknown rules prevent an equivalence claim, not market discovery.
+2. **Polymarket US recording adapter for the existing paper workflow.** Reuse existing discovery/normalization code where suitable. Preserve raw venue IDs, decimal prices/depth, observation/exchange timestamps, book quality, and schema version. Explicitly handle one-sided/missing/stale books. Keep historical display prices separate from trades/books. Acceptance: one sports recording runs through the existing strategy contract and Vaticor experiment comparison without strategy changes; unsupported fill assumptions remain visible. Start with synthetic offline fixtures. Capturing or retaining venue books for replay requires applicable recording, retention and replay permissions first; public endpoint access does not satisfy that gate.
+3. **Second-venue commercial and matched-market trial decision.** Record answers for API eligibility, customer-owned/existing-account model, unattended automation, hosting/geography, data retention/display/replay rights, fees and credential scopes. Draft questions can proceed now; sending messages was not authorized in this stream. Once access is available, compare matched propositions at the same pregame observation times and fixed 10/100/1,000-contract sizes, reporting price plus fees, depth and freshness, with rule differences beside the result. Choose or defer the live adapter from those results.
+
+Novig-specific adapter requirements if selected: OAuth renewal; separate QA configuration; quantity conversion (`100` wire units per contract versus contract units in CSVs); fees rounded to ledger precision; snapshot plus stream recovery; venue lifecycle transitions and cancellations; accepted order versus reconciled fill distinction. Preserve raw capability differences instead of expanding the stable kernel into a broad broker abstraction. [Fees and units](https://docs.novig.com/fees), [order-book semantics](https://docs.novig.com/api-reference/WSS/orderbook-channel), [order placement](https://docs.novig.com/api-reference/orders/place-order)
+
+## Remaining decision evidence
+
+Unknown: comparative executable prices across matched venues; stable pregame depth; exact contract equivalence with Kalshi; commercial permissions and onboarding cost for Vaticor; Novig credential eligibility; production adapter behavior. Synthetic sports matching and offline paper fixtures may proceed. Captured venue recordings and replay datasets remain gated on applicable recording, retention and replay rights, independently of any future live-trading decision.
+
+The research stream used public read-only sources; it made no trading or account changes.

docs/research/README.mdx

@@ -0,0 +1,13 @@
+---
+title: Sports product planning baseline
+---
+
+# Sports product planning baseline
+
+These dated reports distinguish observed implementation, product decisions and future work. They do not establish hosted or live trading readiness.
+
+- [Sports product scope and reconciled roadmap](./2026-09-09-sports-product-reconciliation.mdx)
+- [Novig and Polymarket US venue decision](./2026-09-09-sports-venue-decision.mdx)
+- [Local paper workflow verification](./2026-09-09-paper-workflow-verification.mdx)
+
+Plane owns current delivery state. These documents preserve the September 9, 2026 baseline; later implementation and acceptance should cite its own exact candidate.

docs/research/meta.json

@@ -0,0 +1,9 @@
+{
+  "title": "Research",
+  "pages": [
+    "README",
+    "2026-09-09-sports-product-reconciliation",
+    "2026-09-09-sports-venue-decision",
+    "2026-09-09-paper-workflow-verification"
+  ]
+}