Update to Bigdata.com

{
  "name": "bigdata-earnings-preview",
  "description": "Create a forward-looking earnings preview for a public company ahead of its next earnings call, using Bigdata.com MCP data (estimates, tearsheet financials, news, filings, transcripts, events calendar). Produces an EPIC driver table, earnings quality screen with forward watch-fors, structured sentiment and positioning data, what's priced in plus valuation cross-check, FaVeS variant perception, bull and bear cases, bull/base/bear scenarios with probability-weighted expected value, and key metrics to watch — fully cited. Triggers: \"earnings preview for X\", \"preview X earnings\", \"pre-earnings analysis\", \"Q3 preview\", \"what to expect before X reports\", \"what should I watch when X reports\", \"set up for X earnings\", \"bull and bear case into the print\".",
  "included_files": [
    {
      "relative_path": ".DS_Store",
      "size_in_bytes": 10244
    },
    {
      "relative_path": "README.md",
      "size_in_bytes": 2834
    },
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 368
    },
    {
      "relative_path": "assets/bigdata-icon-black.svg",
      "size_in_bytes": 3278
    },
    {
      "relative_path": "assets/report-template.md",
      "size_in_bytes": 6120
    },
    {
      "relative_path": "references/epic-framework.md",
      "size_in_bytes": 7472
    },
    {
      "relative_path": "references/faves-framework.md",
      "size_in_bytes": 6061
    },
    {
      "relative_path": "references/quality-of-earnings.md",
      "size_in_bytes": 8126
    },
    {
      "relative_path": "references/reverse-dcf.md",
      "size_in_bytes": 7752
    },
    {
      "relative_path": "references/thesis-construction.md",
      "size_in_bytes": 9072
    },
    {
      "relative_path": "scripts/scenario_probability.py",
      "size_in_bytes": 18087
    }
  ],
  "skill_md_contents": "---\nname: bigdata-earnings-preview\ndescription: >\n  Create a forward-looking earnings preview for a public company ahead of its next earnings\n  call, using Bigdata.com MCP data (estimates, tearsheet financials, news, filings, transcripts,\n  events calendar). Produces an EPIC driver table, earnings quality screen with forward\n  watch-fors, structured sentiment and positioning data, what's priced in plus valuation\n  cross-check, FaVeS variant perception, bull and bear cases, bull/base/bear scenarios with\n  probability-weighted expected value, and key metrics to watch — fully cited. Triggers:\n  \"earnings preview for X\", \"preview X earnings\", \"pre-earnings analysis\", \"Q3 preview\",\n  \"what to expect before X reports\", \"what should I watch when X reports\", \"set up for X\n  earnings\", \"bull and bear case into the print\".\n---\n\n# Bigdata Earnings Preview\n\nForward-looking, pre-earnings research note on a public company. Use Bigdata.com plugin tools for every fact; apply the pre-synthesis filter below before writing.\n\n**Use this skill when** the user wants analysis *before* a company reports. Not this skill when:\n\n| Request | Use instead |\n|---------|-------------|\n| Analysis of results already reported | Earnings digest / earnings reaction |\n| Retrospective summary of recent news | Company brief |\n| \"What is it worth\" with no earnings event | Valuation snapshot |\n| Comprehensive risk mapping | Risk assessment |\n\n## Data foundation (MCP tools)\n\n| Tool | Purpose | Prerequisite |\n|------|---------|--------------|\n| `find_securities` | Resolve company name → RavenPack `entity_id` | None |\n| `bigdata_company_tearsheet` | Financials, estimates, margins, sentiment/positioning fields | `find_securities` |\n| `bigdata_events_calendar` | Next earnings call date | `find_securities` |\n| `bigdata_search` | News, filings, transcripts, analyst and regulatory coverage | None |\n\nIf the company name is ambiguous after `find_securities`, ask:\n\n> \"I found multiple companies named [X]. Did you mean [Company A] in [Industry] or [Company B] in [Industry]?\"\n\n## Before you synthesize — quality over quantity\n\nDo this **after** gathering data and **before** writing the draft:\n\n1. List candidate drivers from tearsheet + search.\n2. Rank them and keep the top **2–3** as primary drivers. Do not give 20 findings equal weight.\n3. Run the full **EPIC** filter on each primary driver:\n\n| Test | Question |\n|------|----------|\n| **E**ffect (material) | Would getting this wrong move value or the debate on the name meaningfully? |\n| **P**redictability | Can *you* form a view with evidence, not speculation? |\n| **I**ndependence | Does consensus or price **systematically** under- or over-weight this factor? |\n| **C**onsensus gap | Does **your** view differ from consensus in a specific, falsifiable way? |\n\n4. Map each primary driver to an implication (bullish / bearish / neutral) with specific metrics.\n5. Deprioritize template sections that are immaterial this period — say so in one line rather than padding.\n\nDepth: [references/epic-framework.md](./references/epic-framework.md).\n\n## Workflow\n\n### Step 1 — Identify the company\n\nCall `find_securities` with the company name to get the `entity_id`.\n\n### Step 2 — Financial baseline\n\nCall `bigdata_company_tearsheet` with the `entity_id` for:\n\n- Recent quarterly performance trends and YoY comparisons\n- Historical earnings surprises\n- Analyst estimates for the upcoming quarter\n- Key financial metrics and margins\n- **Positioning / sentiment fields when exposed** — sentiment scores, news/social metrics, ownership concentration, insider summary, options or short interest. Capture whatever the tool returns; never substitute analyst headlines for systematic data when numbers exist.\n\n### Step 3 — Earnings quality quick screen\n\nBefore building narratives, record a credibility table with a forward-looking **watch for** column (approximate if necessary; flag data gaps):\n\n| Check | This period / trend | Red-flag threshold | **Watch for (next print)** |\n|-------|---------------------|--------------------|----------------------------|\n| OCF / Net income | | Healthy often >0.8 sustained; <0.6 or widening gap → dig | further OCF/NI divergence; one-time boosts rolling off |\n| DSO vs revenue growth | | DSO rising faster than revenue → recognition risk | DSO days vs rev growth; channel inventory mentions |\n| GAAP vs non-GAAP EPS gap | | Large or widening gap → quality question | stock comp, restructuring, \"adjusted\" add-backs |\n\nIf the tearsheet lacks a line, search: \"[Company] operating cash flow vs net income non-GAAP reconciliation\".\n\nDepth: [references/quality-of-earnings.md](./references/quality-of-earnings.md).\n\n### Step 4 — Earnings date\n\nCall `bigdata_events_calendar` with the `entity_id` to find the next earnings call. If unknown, ask:\n\n> \"I don't have the exact earnings date yet. Shall I proceed with the preview based on recent developments and expectations?\"\n\n### Step 5 — Search: developments, legal/regulatory, positioning\n\nCast a **wide net** so material non-operational risks (court rulings, probes, tax disputes) are not missed. Use `bigdata_search` over the last **60–90 days** (extend if coverage is thin). Run **at least 8–10 targeted queries** across all three buckets, then merge redundant results.\n\n**Core company & industry**\n- \"[Company] recent developments last 90 days\"\n- \"[Company] product launches initiatives\"\n- \"[Company] guidance commentary management\"\n- \"[Company] analyst expectations earnings preview\"\n- \"[Industry] trends headwinds tailwinds\"\n\n**Regulatory, legal, policy (mandatory)**\n- \"[Company] lawsuit litigation court ruling settlement regulatory investigation last 90 days\"\n- \"[Company] SEC investigation DOJ antitrust fine penalty Europe\"\n- \"[Company] tax dispute regulatory approval compliance\"\n\n**Market positioning & flows (mandatory — fills the structured table in the output)**\n- \"[Company] insider buying selling Form 4 transactions last 90 days\"\n- \"[Company] institutional ownership 13F changes fund flows\"\n- \"[Company] short interest options put call ratio open interest\"\n- \"[Company] news sentiment score\" (or closest available)\n\nIf news is sparse, ask:\n\n> \"There's been limited news recently. Would you like me to expand the search period or focus on industry trends?\"\n\n### Step 6 — Sentiment & positioning (structured, not anecdotes)\n\nBuild the output table from data, not from a single analyst note:\n\n1. Pull every **numeric** sentiment / flow / positioning field from the tearsheet.\n2. Use Step 5 results to fill gaps (insider trades, large holder moves, options/skew, quantified sentiment).\n3. If a cell is unavailable, write **\"Not available in data\"** — the section still appears.\n\n### Step 7 — What's priced in + valuation cross-check\n\n**Before** writing bull/bear narratives, establish what the current price embeds for this quarter and the near-term trajectory. Use tearsheet multiples, consensus, and reverse-DCF-style reasoning (conceptual is fine — [references/reverse-dcf.md](./references/reverse-dcf.md)).\n\n| Lens | Implied by market | Consensus | Your assessment |\n|------|-------------------|-----------|-----------------|\n| Growth (revenue / key volume) | | | |\n| Margin level or expansion | | | |\n| Beat magnitude / \"whisper\" vs published consensus | | | |\n\n**Multiples sanity check:** current EV/EBITDA, P/E, FCF yield (or sector-standard multiples) vs ~5-year range or peer median where data allows. State whether valuation implies **optimism**, **consensus**, or **pessimism** relative to the setup.\n\n### Step 8 — Variant perception (FaVeS) + scenarios\n\n**FaVeS — mandatory structure in the output:**\n- **Fundamentals** — the 2–3 KPIs that drive the quarter; where consensus could be wrong (link to bull/bear).\n- **Valuation** — tie to the *What's priced in* table and valuation cross-check; cross-reference rather than repeat prose.\n- **Sentiment** — tie to the *Sentiment & positioning* table; separate what is priced in **behaviorally** from **fundamentally**.\n\nDepth: [references/faves-framework.md](./references/faves-framework.md).\n\n**Scenario analysis — mandatory:** build Bull / Base / Bear with\n\n- **Probability weights** summing to ~100% (e.g. 30/50/20), briefly justified\n- **Key assumptions** per scenario (growth, margin, one-timers, legal outcomes)\n- **Price level or range** per scenario (spot, consensus PT band, or a rough DCF/multiple bridge — show the assumptions)\n- **Probability-weighted expected value** with the arithmetic shown (EV = Σ p×P; state expected upside/downside % vs spot)\n\nMethodology: [references/thesis-construction.md](./references/thesis-construction.md). Compute in prose/table by default; run [scripts/scenario_probability.py](./scripts/scenario_probability.py) only if the user explicitly asks for scripted math.\n\n### Step 9 — Synthesize\n\nLead with the 2–3 primary drivers and their EPIC documentation. Cover, in order of materiality only:\n\n- **Recent developments** — launches, partnerships or M&A, operational shifts, geographic/share changes, plus material legal and regulatory items\n- **Industry trends** — macro drivers, competitive landscape, supply chain and cost pressure, policy\n- **Bull case** — each point specific, measurable, evidence-backed and resolvable over a sensible horizon; tie to consensus line items where possible (\"consensus models X% growth in segment Y; channel evidence suggests Z%, ~$Nm revenue upside\"); cite a source per claim\n- **Bear case** — same discipline; quantify downside (margin bps, revenue %, one-time vs recurring)\n- **Key metrics to watch** — the KPIs that matter *this* quarter, those that move the stock given what's priced in, and where surprise volatility is highest vs whisper/consensus\n\n## Output\n\nFollow [assets/report-template.md](./assets/report-template.md) exactly — section order, mandatory tables, sources, and footer.\n\n- Add inline citations as superscript-style numbers `[1]`, `[2]` immediately after claims, hyperlinked to the document URL.\n- Every deliverable ends with the **Powered by Bigdata.com** line and the **Disclaimer**, verbatim.\n- Default format is Markdown; offer a Word (.docx) or presentation version at the end if useful.\n\n## Quality bar\n\nPass the PM test before delivering: **What's different?** **What matters (2–3 drivers)?** **What should I do about it?** (net assessment, key risk, next catalyst — no position sizing). Would it survive a short, skeptical morning meeting without reading as a data dump?\n\nNon-negotiables in every preview:\n\n- EPIC table for each elevated driver — filled, not placeholder\n- Scenario table with probabilities, prices, and **EV math shown**\n- Sentiment & positioning as structured data — tearsheet first, then search\n- Regulatory/legal queries run, and material items surfaced in developments and the bear case\n- **Watch for** column on the quality screen — forward monitoring, not only backward checks\n- *What's priced in* built before bull/bear, so both cases are relative to embedded expectations\n- Facts separated from analysis and implications\n"
}