Upside Ranks→Publication QA
Monthly publish gate

Upside Ranks Publication QA Packet

Prevent stale numbers, weak thesis cards, free/paid leakage, and site/newsletter formatting mistakes before the monthly Upside Ranks package goes live.

View JSON QA packetScoring auditThesis template
Edition
July 2026
Overall status
editor review required
Gate mix
1 pass · 5 warn · 0 fail
Version
2026-06-10
Recommended monthly workflow

Run this before deploy/send

  1. 1Freeze the monthly data snapshot and edition label before writing copy.
  2. 2Run scoring audit for every candidate and demote any name that fails source confidence or risk disclosure.
  3. 3Complete one thesis card per ranked name using the stock-thesis template.
  4. 4Assemble site page, newsletter excerpt, and paid deep-dive package from the same approved data snapshot.
  5. 5Run this QA packet before deploy/send; unresolved fails block publication, warnings require editor signoff.
Automated data QA

Monthly JSON validator

Reads src/data/upside-ranks-july-2026.json, checks required thesis/scoring/source fields for every ranked name, and emits a deploy/send gate.

ready to publish
Stocks
20
List entries
35
Warnings
0
Fails
0

Required stock/thesis fields

pass

20 checked · 0 warnings · 0 fails

List/rank integrity

pass

35 checked · 0 warnings · 0 fails

Editorial/rank consistency

pass

20 checked · 0 warnings · 0 fails

Pillar score/rationale coverage

pass

120 checked · 0 warnings · 0 fails

Source URL coverage

pass

20 checked · 0 warnings · 0 fails

QA gates

What blocks monthly publication

Research

Data snapshot + edition lock

A monthly product breaks trust if ranks, prices, market caps, and edition labels disagree across pages.

warning

Checks

  • • Edition month/date appears consistently on landing page, tier pages, stock pages, API payloads, and newsletter copy.
  • • Current price, market cap, rank, probability, and quant score are pulled from one frozen data file.
  • • Archive links still point to prior editions and are not silently overwritten.

Evidence

  • • July 2026 is visible on the current landing page and existing audit/thesis packets.
  • • The current scoring audit uses the July 2026 JSON data record as its sample source.

Blocker rule

Any visible edition/date mismatch or duplicated rank across the final package.

Research

Scoring/ranking integrity

The ranking needs to feel explainable, not vibes-based.

pass

Checks

  • • Confirm current price, market cap, and list placement against the latest monthly data pull.
  • • Verify revenue growth, margin, guidance, and earnings-date claims from primary filings, investor relations, or trusted financial data.
  • • Write one sentence explaining why the catalyst matters this month, not someday.
  • • State the Goldilox target path and the single assumption most likely to invalidate it.
  • • Flag stale, missing, or low-confidence numbers before the name enters the publish package.
  • • Only mark ready when the thesis, key numbers, source URLs, and risk paragraph can survive a reader fact-check.
  • • Top-ranked names have rationale for the rank delta versus nearby names.
  • • A lower-ranked name is not written with stronger conviction than a higher-ranked name unless explicitly explained.

Evidence

  • • Worked sample CRDO audit score: 85.8/100.
  • • Worked sample status: Ready for editor review.

Blocker rule

Any final top-10 name lacks score, rank rationale, risk paragraph, or source-confidence pass.

Editorial

Thesis-card completeness

Upside Ranks should publish as a repeatable research product, not only a ticker table.

warning

Checks

  • • Include list, rank, probability, quant score, and edition month.
  • • One sentence only; no generic “AI winner” language.
  • • Must identify a near-term trigger: earnings, guidance, orders, product cycle, capex cycle, or technical setup.
  • • Use current market cap/price, score, target-path math, and at least two material operating numbers when available.
  • • Use the relevant +35%, 2x, or 5x path from the monthly data record; do not imply it is a base case.
  • • Must include the fastest thesis-breaker and whether the risk is valuation, execution, demand, concentration, or balance sheet.
  • • Name one upgrade trigger and one cut/demotion trigger.
  • • Minimum two credible links; prefer filings, company IR, transcripts, or high-quality financial data.

Evidence

  • • CRDO: thesis card shell generated with 4 source links.
  • • NVDA: thesis card shell generated with 3 source links.
  • • AEHR: thesis card shell generated with 4 source links.

Blocker rule

Any ranked stock goes live without why-now, key numbers, bull path, bear case, and what-would-change-our-mind fields.

QA

Source dates + embarrassment-risk check

The fastest way to damage the product is publishing old numbers as current facts.

warning

Checks

  • • Every material number has a source URL, source name, and source date in the working packet.
  • • Earnings/guidance numbers are checked against company IR, filings, or trusted financial data before publication.
  • • If a number is estimated, model-derived, or uncertain, the copy labels it that way instead of pretending precision.
  • • Risk language includes the fastest thesis-breaker, not a generic boilerplate disclaimer.

Evidence

  • • Existing thesis template requires two credible links per card.
  • • Existing scoring audit has a dedicated source-confidence pillar weighted at 12%.

Blocker rule

Any hard number in the newsletter/site package cannot be traced to a source in under 60 seconds.

Editorial

Free/paid boundary

The free issue should sell the product without giving away the full paid research package.

warning

Checks

  • • Free newsletter: publish framing, 1–3 teaser names, and product value proposition.
  • • Paid section/site: preserve full rankings, target-path detail, source packet, and thesis cards.
  • • Do not leak full top-30 tables, complete target math, or all stock-specific bear cases into free copy.

Evidence

  • • This gate is editorial/process only; no automated content split exists yet.

Blocker rule

Free copy contains enough detail that a reader does not need the paid/site package.

Publishing

Site + newsletter delivery QA

Last-mile formatting errors make good research look sloppy.

warning

Checks

  • • Landing page links to current edition, archive, scoring audit, thesis template, and publication QA packet.
  • • API payloads return HTTP 200 and valid JSON for any automation-facing packets.
  • • Newsletter copy has no broken links, no raw Markdown artifacts, and no stale archive references.
  • • Post-deploy smoke test verifies the live page, not only local build output.

Evidence

  • • This packet ships as a live page plus JSON endpoint so the QA workflow itself can be smoke-tested.

Blocker rule

Any core URL is missing, gated unexpectedly, returns non-200, or shows stale edition language.

Next automation slice

Wire this validator into the monthly refresh/deploy workflow so failures block publication automatically.