{
 "as_of": "2026-08-07",
 "source": "Audit of every live tool against the gate on index.html#tiers, performed 2026-08-07.",
 "tier": "labs",
 "graded": false,
 "note": "ONE REVIEWER, ONE DAY, reasoning from the code and the published data. This is a judgment, not a measurement — which is the gate’s known weakness and the reason the audit is published rather than held. Result: NO promotions, NO demotions. That is a suspiciously comfortable outcome and is flagged as such; it survives only because every collar came back with a condition attached. Treat these verdicts as falsifiable.",
 "headline_finding": "The Pound is empty. After ~170 commits, zero recorded retirements. Either nothing has ever failed, or failures are being abandoned rather than recorded — and the first is not plausible. An empty Pound is survivorship bias appearing in the one structure built to prevent it.",
 "gate": "Does it work, and can it be trusted? Forecasts need receipts against a benchmark chosen in advance; measurement tools need named sources and reproducible math. Nothing clears by feeling finished.",
 "what_would_change_our_mind": {
  "epa-stats": "An external cross-check that disagrees with the page.",
  "nfelo": "Nothing this season — the market comparison is unresolvable in one year by the site’s own pre-registered arithmetic.",
  "draft-rig": "A draft night where the rig fails under load, or a refusal to refresh the market snapshot before it.",
  "receipts": "An external, timestamped anchor would promote it. Nothing else would.",
  "any-labs-tool": "Receipts against a benchmark declared in advance."
 },
 "counts": {
  "audited": 11,
  "dawgs": 3,
  "labs": 6,
  "pound": 0,
  "promotions_recommended": 0,
  "demotions_recommended": 0,
  "outside_the_tier_system": 2
 },
 "data": [
  {
   "id": "epa-stats",
   "name": "NFL EPA Stats",
   "page": "/stats.html",
   "lane": "dawg",
   "verdict": "stands",
   "path": "instrument",
   "checked": "Aggregation independently re-implemented outside the page (tools/build-data.js) from the same encoded columns; outputs internally consistent and consistent with the seasons (SF 2023, BAL 2024, NE 2025 leading offensive EPA/play). Sources named, snapshot dated.",
   "not_checked": "No external cross-check against a published nflfastR table. Re-implementing the page’s own algorithm proves the maths is reproducible, not that it is right.",
   "owed": "One spot-check against a public source."
  },
  {
   "id": "nfelo",
   "name": "nfelo Power Ratings",
   "page": "/nfelo.html",
   "lane": "dawg",
   "verdict": "stands",
   "path": "instrument",
   "checked": "Pinned to nfelo commit 0d3f8418, model v4.3.0. Source named, derivation reproducible, dated.",
   "caveat": "The collar certifies the MIRROR, not the forecasts. Backtest on file: n=4,053, nfelo SU 0.6674 vs market 0.6677 — nfelo behind by 0.02pp against a 0.196pp standard error, i.e. indistinguishable, and in-sample-ish because it overlaps nfelo’s own optimisation window.",
   "owed": "A one-line qualifier on the page itself, matching the homepage disclaimer."
  },
  {
   "id": "draft-rig",
   "name": "Fantasy Draft Dashboard (the draft rig)",
   "page": "/dashboard.html",
   "lane": "dawg",
   "verdict": "stands",
   "path": "operational",
   "checked": "Ran a live 14-team draft end to end. For an operational tool that is the right evidence — \"does it work\" is answered by working under load.",
   "conditions": [
    "The Dashboard is a frame shell over board.html, dataviz.html, report.html and auction.html. Those are live standalone URLs with NO tier chip — same code, different verdict depending on how you arrive.",
    "The Grades view is a measurement dressed as a judgment: letter(0.5*z(surplus) + 0.5*z(starting-lineup value)). It measures buying value relative to the room, not whether the roster will win. Fix is labelling, not demotion.",
    "The rig runs on a 2026-07-29 market snapshot from a source that publishes through August. The tool works; the input is decaying. Collar is conditional on a refresh before draft night."
   ]
  },
  {
   "id": "receipts",
   "name": "Receipts",
   "page": "/receipts.html",
   "lane": "labs",
   "verdict": "stands",
   "path": "instrument",
   "checked": "Sources named (nfelo 0d3f8418, nflverse/nfldata, ESPN). Canonical string published and recomputes the locked hash — 272 rows, 6,697 bytes. Dated. Failure threshold pre-registered before any 2026 game.",
   "blocker": "The trust mechanism is SELF-ANCHORED. Data Dawgs hosts the ledger, the hash and the spec. Recomputation proves today’s file matches today’s hash; it cannot prove these were the predictions made before the games. A page whose purpose is verifiability cannot earn a collar on verification that is not independently checkable.",
   "promotion_path": "Anchor the hash externally and timestamp it. Nothing else needs to change."
  },
  {
   "id": "master-data",
   "name": "Master Data",
   "page": "/master.html",
   "lane": "labs",
   "verdict": "stands",
   "path": "instrument",
   "blocker": "Staleness, not method. Would clear on sources and reproducibility; the source updates daily and the file says 2026-07-29.",
   "promotion_path": "A refresh cadence. Cheapest promotion available."
  },
  {
   "id": "survivor",
   "name": "NFL Survivor",
   "page": "/survivor.html",
   "lane": "labs",
   "verdict": "stands",
   "path": "forecast",
   "blocker": "No receipts. Two disclosed defects: double-pick weeks are recorded but not simulated, so survival numbers after such a week are optimistic in a known direction; ownership is modelled, not observed."
  },
  {
   "id": "bozo",
   "name": "Bozo",
   "page": "/bozo.html",
   "lane": "labs",
   "verdict": "stands",
   "path": "forecast",
   "blocker": "Every price is self-reported and unverified. The simulation draws legs independently, knowably false for two legs on the same game. Worst CLV is unmeasured. The game is well designed; the numbers are not validated."
  },
  {
   "id": "dfs",
   "name": "DFS Solver & Contest Simulator",
   "page": "/dfs.html",
   "lane": "labs",
   "verdict": "stands",
   "path": "forecast",
   "blocker": "Built 2026-08-06, zero graded slates. Determinism is a strength but is not validation — an optimiser is exactly as good as the projections it is fed, and those are untested."
  },
  {
   "id": "guillotine",
   "name": "Guillotine Companion",
   "page": "/guillotine.html",
   "lane": "labs",
   "verdict": "stands",
   "path": "forecast",
   "blocker": "Nothing graded, nothing claimed."
  },
  {
   "id": "strategy",
   "name": "Strategy",
   "page": "/strategy.html",
   "lane": "labs",
   "verdict": "outside",
   "reason": "A synthesis of one analyst’s opinions, not a tool. Tiering it implies a validation question that does not apply."
  },
  {
   "id": "dawgs-page",
   "name": "Dawgs",
   "page": "/dawgs.html",
   "lane": "labs",
   "verdict": "outside",
   "reason": "Photographs. Not a tool."
  }
 ],
 "tier_meaning": "Pup — live and useful, not yet validated. It may compute real answers and still have open questions about calibration, assumptions, data quality or edge. Everything starts here.",
 "built": "2026-08-28",
 "canonical_url": "https://datadawgs216.com/data/tier-audit.json"
}