Finance & Investment Banking earnings-analysisequity-researchvaluationdcfinvestment-memoposition-sizingsec-filingsanti-bias

Tech Earnings Deep Dive Analyzer

Turns a one-line ask like "analyze NVDA's latest earnings" into an institutional-grade, multi-perspective, decision-ready investment memo.

FollowSkills review · FSRS-2.0
Use with care
47/ 100 5-point scale 2.4 / 5
1 2 3 4 5 6
1Trust13 / 25 · 2.6/5

Pure methodology prompt: no code, no permission requests, no external side effects; transparent data flow; explicitly demands primary-source tracing, forbids fabricated citations, includes a disclaimer. Deductions: it mandates injecting a promotional footer (with external links) into every output — an unconfirmed output hijack deviating from least intervention; authors are individuals, not registry-verified. Below red-line threshold.

2Reliability8 / 20 · 2.0/5

Clear flow structure (Steps 0-5, modules A-P, output template, writing discipline); key paths statically inferable. Deductions: SKILL.md depends on references/valuation-models.md, investing-philosophies.md, bias-checklist.md, none supplied in evidence; no tests, no failure-feedback design for abnormal input; output quality hinges on live agent retrieval with low reproducibility.

3Adaptability9 / 15 · 3.0/5

Trigger description is very thorough, with broad topic terms and example prompts; bilingual output rules explicit; skill boundaries and synergies declared. Deductions: deliberately wide triggering ('should I keep holding MSFT') risks over-triggering on simple questions; non-fit boundaries (non-tech, non-earnings) undeclared; core data (SEC filings, 13F, market data) depends on overseas sources with no degraded-mode behavior specified.

4Convention8 / 15 · 2.7/5

Well-layered docs: positioning, execution flow, modules, template, writing discipline, language rules, skill coordination; MIT license explicit; version v3.0. Deductions: no changelog, no stated maintenance responsibility or update path, no FAQ/known-limitations section; mandating a promotional footer is an improper convention; referenced files not provided in evidence.

5Effectiveness6 / 15 · 2.0/5

Offers a complete reusable framework and output template; the 16-module + 6-perspective + anti-bias combination plausibly beats ad-hoc manual analysis. Deductions: static review cannot verify real report quality or direct usability; forced promotional footer requires manual cleanup; the workload (full-depth 16 modules) raises cost/benefit concerns for a single session.

6Verifiability3 / 10 · 1.5/5

Methodologically verifiable design: Tier 1-3 evidence grading, anti-fabrication rule, evidence source list in output. Deductions: no execution or reproduction from static read; no CI, tests, or third-party execution evidence; whether citations are actually traceable depends entirely on runtime behavior, with no committed corroboration.

Evidence confidence:Low Reviewed Sep 09, 2026 Reviewed revision 562c14b0c0bc
Before you use it
  • The skill mandates a promotional footer on every output; clean it before use
  • The three referenced references/ files were not provided in this evidence set — verify their presence and quality after install
  • Analysis depends entirely on live agent retrieval of SEC/market data; results are non-reproducible and may contain hallucinated citations — verify primary sources before acting
  • Core data sources are US-based and may be unreachable from mainland-China networks; no fallback is specified
  • Outputs include investment-decision content (Action Price, position sizing) — research reference only, not investment advice
Review evidence [1][2][3]
See the full review method →

What does this skill do, and when should you use it?

tech-earnings-deepdive is an earnings-analysis skill designed for the "large retail investor" — self-funded, holding tech positions on a quarterly basis. Anchored on 1-3 key forces, it runs 16 analysis modules (A-P), 6 investment philosophy perspectives, a multi-method valuation matrix, and an anti-bias/pre-mortem pass before outputting an actionable memo with an Action Price, position-sizing cadence, and a long-term monitoring checklist. It enforces primary-source evidence standards (SEC filings, insider transactions, hiring and customer signals) and demands a Variant View that challenges market consensus. It is pure Markdown prompting — no scripts or external tools.

On receiving an earnings-analysis request, it executes a five-step flow: Step Zero identifies the 1-3 forces that will determine the company's value; Step One runs the 16 modules (revenue quality, margins and GAAP/Non-GAAP variance, cash flow and capital allocation, guidance, competitive landscape, industry-specific KPI dashboards, products and new businesses, partner ecosystem, executives and governance, macro/policy, valuation model selection, ownership structure, monitoring variables, R&D efficiency, accounting quality, ESG screening), requiring core data traced to Tier 2 sources like SEC filings and at least 3 Tier 1 primary sources per report; Step Two elicits Long/Short/Pass calls from six philosophies (Buffett/Munger, Baillie Gifford/ARK, Tiger Cubs, Klarman/Marks, Tepper/Ackman, Druckenmiller); Step Three builds a valuation matrix with sensitivity analysis and probability-weighted IRR iron rules (long ≥15%, short ≥20-25%); Step Four runs bias self-checks and a pre-mortem; Step Five outputs the full memo with Variant View, kill conditions, action triggers, and monitoring checklists, written in the user's input language.

  1. A quarterly tech-stock holder who, after earnings drop, needs a structured add/trim/hold decision instead of a gut call
  2. An independent investor wanting a forced Variant View and anti-bias checklist to test whether they're anchored to consensus
  3. A long/short analyst building a short thesis from ownership distribution, insider selling, and short-interest data (Module L)
  4. An investment group tracking one company across quarters with a standardized template (key forces → modules → valuation → decisions)
  5. A retail investor wanting institutional-grade output discipline via Tier 1-3 evidence sourcing rules

What are this skill's strengths and limitations?

Pros
  • Complete methodology: 16 modules + 6 philosophies + multi-method valuation + anti-bias framework, covering evidence collection through execution
  • Key-forces-driven depth allocation avoids checklist-style breadth with no depth
  • Tiered evidence standard explicitly bans fabricated citations and requires tracing core data to SEC filings
  • Actionable output: Action Price, entry pacing, add/trim/exit triggers, and monitoring checklists rather than vague bullish/bearish calls
  • Pure Markdown prompting with no script dependencies — portable to any Agent Skills-compatible client
Limitations
  • Output quality depends entirely on the model's ability to retrieve real, current financials online; the skill ships no data sources or scripts, so offline or outdated retrieval degrades reliability
  • No test suite or benchmark evaluation provided
  • A full run spans 16 modules and 6 perspectives — high token usage and long runtime
  • Parameters like IRR thresholds are hard-coded by the author; adjusting for personal risk preference requires editing SKILL.md by hand
  • Long multi-step instructions rely on the model following them faithfully; smaller-context models may execute the flow incompletely

How do you install this skill?

Option 1 (Skills.sh auto-discovery): run npx skills add https://github.com/star23/Day1Global-Skills --all — skills install to .agents/skills/ and auto-activate on matching topics. Option 2 (Claude Code slash command): git clone https://github.com/star23/Day1Global-Skills.git, then cp -r Day1Global-Skills/tech-earnings-deepdive your-project/.claude/skills/ (or ~/.claude/skills/ for global access) and restart Claude Code to use /tech-earnings-deepdive. Note: the repo bundles 5 skills; these commands cover only tech-earnings-deepdive.

How do you use this skill?

No configuration after install — just ask in natural language. Example prompts: "Analyze NVDA's latest earnings report", "How did META perform this quarter?", "Analyze MSFT from multiple legendary investors' perspectives. Is it a buy right now?" If us-value-investing, us-market-sentiment, or macro-liquidity are also installed, the skill recommends combining them at the relevant step (e.g., four-dimension value scoring for cross-validation).

How does this skill compare with similar options?

The source explicitly positions this skill as a complement to us-value-investing in the same repo: us-value-investing does Buffett-style four-dimension long-term value scoring, while this one dissects the latest earnings, confronts multiple investment philosophies, and produces actionable position decisions. The author recommends running both for cross-validation.

FAQ

Does it need paid data feeds or API keys?
SKILL.md requires none — analysis relies on the model retrieving public information (SEC filings, earnings calls). But your client must have web/search capability, or report quality cannot be guaranteed.
Is the output investment advice?
No. Every report carries a disclaimer: based on public information and model estimates, for research reference only, not investment advice.
Does it support Chinese output?
Yes. The skill mandates matching the user's input language — a Chinese prompt yields a fully Chinese report, including a translated promotional footer.
Is it tech stocks only, or other sectors too?
It is designed for tech stocks: Module F's KPI dashboards are categorized for SaaS, consumer internet, semiconductors, ad-driven, and platform businesses, and the bias checklist targets tech-specific blind spots. Other sectors fall outside its stated scope.

More skills from this repository

All from star23/Day1Global-Skills

Related skills