qualitative-filtering
SkillDev toolsQualitative stock analysis, management operating plan MOP assessment, key performance indicator KPI identification, catalyst identification and classification, earnings call transcript analysis, qualitative evidence gathering for investment thesis
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the qualitative-filtering skill
What this skill tells your AI
The instructions your AI receives, as published by agentii-ai/agentii-investment-intelligence in plugins/vertical-plugins/idea-generation/skills/agentii/qualitative-filtering/SKILL.md and read by ahel’s review.
Methodology fused from professional trading and investment frameworks; all text is an original paraphrase.
Defaults
| Parameter | Default Value | Rationale |
|---|---|---|
| catalyst_window_days | 20-60 | Trading horizon for active positions |
| kpi_trend_min_quarters | 8 | Minimum quarters of KPI history for trend analysis |
| mgmt_track_record_years | 3 | Management credibility requires 3+ years of guidance vs actuals |
| catalyst_min_impact | 5% | Minimum expected price impact to justify catalyst-driven trade |
Preflight
Run canonical pre-flight per contracts/preflight.md. Propagate X-Agentii-Trace per contracts/x-agentii-trace-header.md.
Data Source Priority
- Qualitative methodology —
references/qual-methodology.md(bundled MOP-KPI-Catalyst framework) - Company disclosures — SEC filings (Business Description, Risk Factors, MD&A) via agentii MCP
- Earnings transcripts —
search_documents(ticker={T}, form_type="earnings_call_transcript")→read_source_outline→read_source_pages(citation prefixect<N>; pages carry section_type in session_title and guidance/forward_looking/analyst_questions in labels) - Strategy and case knowledge —
search_investment_strategies+search_investment_cases+search_by_analogue
Methodology
Retrieval Scope
unstructured_document_search (earnings call transcripts + SEC disclosures)
Retrieval Strategy
Ownership & insider signals: search_institutional_holdings (top-10 holders + whale portfolios, direction=accumulating|reducing|new|exited) and search_insider_trades (Form-4 transactions with SEC URLs) are available as signal inputs.
Three-layer protocol from contracts/retrieval.md: the qualitative framework is bundled in references/qual-methodology.md. Earnings transcripts via search_documents(form_type="earnings_call_transcript") → read_source_pages (Layer 1→3). Strategy frameworks and historical analogues via MCP knowledge tools. Detailed methodology and catalyst classification in references/qual-methodology.md.
Temporal Scope
See frontmatter temporal_scope block.
Tool Allowlist
See frontmatter allowed_tools.
Protocol
This skill implements qualitative investment analysis through a three-stage framework: MOP → KPI → Results. Companies do not publish an "MOP" — the analyst infers it by identifying KPIs first, then reverse-engineering the strategic plan. The analyst verifies the chain is intact and credible. A broken chain is the highest-quality short signal.
Detailed methodology: management assessment framework (Alpha/Beta/Delta), board quality checklist, mosaic theory triangulation, catalyst taxonomy, sector-specific KPI templates, and the 20+ pattern red flag catalog are in references/qual-methodology.md.
Critical distinction: Good company ≠ good stock. For the 20-60 day horizon, catalysts are required.
Steps
-
KPI Identification: Identify industry-specific KPIs per the reference templates (SaaS, retail, manufacturing, financial services, healthcare). Determine leading vs. lagging. Assess consistency (changing KPIs = red flag), auditability, relevance. Map trends over 8+ quarters. KPI divergence from sector norms often explains quantitative outlier signals.
-
MOP Analysis (5-dimension scorecard, each 0-10, composite < 25 = high risk): Extract from earnings calls, presentations, MD&A. Infer the MOP: forward-looking statements → recurring themes → strategic narrative → test consistency/credibility. Score on: Track Record (30%, 3yr+ guidance vs. actuals), Consistency (20%), Realism (20%), Alignment (15%, insider ownership + compensation structure), Disclosure Quality (15%). Red flags: transformational M&A without plan, repeated guidance misses, high SBC with low hurdles, C-suite turnover within 18 months.
-
Management Team (Alpha/Beta/Delta): Alpha (CEO) — track record, capital allocation, communication style. Primary Betas (CFO, CPO, CTO, Corp Dev, CMO) — depth and tenure. Red flags: cluster departures, CFO departure near guidance, cluster insider selling.
-
Board Assessment: Independence ≥ 75%, expertise present, financial expert on audit committee. "Political incest" check: management/board overlap. Red flags: classified board, supermajority voting, tenure > 15yr, CEO as Chair, related-party transactions.
-
Industry Analysis: Five Forces and SWOT as thinking prompts, not rigid boxes. Read competitor 10-Ks to cross-check management narrative. Do not trust management pronouncements on competition — verify independently. "Explain to 10-year-old" test: describe in 1-2 sentences. "3-5 factors" rule: identify drivers that matter; more than 5 = spread too thin. For formal peer-set construction and relative benchmarking, defer to the
peer-benchskill rather than rebuilding it here. -
Consensus Reconstruction (run after steps 1-5, never before — reading consensus early anchors the analysis to the expectation it is meant to test): Establish the published sell-side average and its dispersion, then triangulate the effective buy-side expectation, which typically moves ahead of the published figure. Weight recent revisions over the stale average. Output a range with a direction, never a point estimate. Wide dispersion = no consensus exists, so the disconnect framing does not apply; tight dispersion with stale revisions is the highest-value setup. If the buy-side bar cannot be triangulated, mark the disconnect unquantified and flag a coverage gap rather than substituting the published number. State the variant view as: market expects X, evidence indicates Y, because [KPI/MOP finding], closing when [catalyst] by [date].
-
Catalyst Identification: Identify all catalysts within 20-60 days. Classify: Earnings / Corporate Action / Regulatory / Management / Industry / Macro. Assess: specificity (dateable?), magnitude (≥ 15% high, 5-15% standard, < 5% insufficient), probability, binary vs. spectrum (binary → reduce size). Tumbleweed test: < 1 non-earnings press release/month = avoid. Catalyst stacking: multiple = higher conviction; zero = investment, not trade.
-
Red Flag Scan: Scan against catalog (see reference). 3+ flags = hard stop for longs. Key flags: non-recurring charges in 3+ of 4 quarters, SBC > 10% revenue, GAAP losses + non-GAAP profits, trade data contradicts management, competitor filings describe different dynamics.
-
MCP Integration:
search_investment_strategies(kind=qualitative)→search_investment_cases(domain=catalyst_driven)→search_by_analogue(event_type, company_situation). Handoff: conviction score (1-10), quantified consensus disconnect (orunquantified), ranked catalyst calendar, KPI summary, MOP score, management/board flags, red flag count.
Output File
{ticker}/{YYYY-MM-DD_HHMM}_qualitative-filtering_{affix}.md
Output Structure
- Executive Summary — Qualitative conviction level, key catalyst, MOP credibility rating
- KPI Analysis — Industry-specific KPIs, trend assessment, leading vs. lagging classification
- MOP Assessment — Management strategy, credibility evaluation, red flags, track record
- Business Quality — Competitive position, industry dynamics, product/service assessment
- Consensus Disconnect — Published sell-side average and dispersion, triangulated buy-side range with direction, the quantified gap (or
unquantified), and the variant view in one structured sentence - Catalyst Calendar — All identified catalysts with type, date, expected impact, probability
- Earnings Call Analysis — Key takeaways from recent transcripts, management tone, analyst sentiment
- Qualitative Red Flags — Governance concerns, strategy pivots, disclosure quality issues
- Knowledge Integration — Matched strategies and historical analogues with /v/ citations
- Handoff Summary — Conviction score, priority catalyst, recommended next step (template/proceed/watch)
- Coverage Gaps — Data limitations, degraded-mode annotations
Error Handling
| Error | Fallback |
|---|---|
search_documents(form_type="earnings_call_transcript") returns empty | Use SEC filings only; flag transcript gap |
| No catalyst within 60-day window | Flag as watchlist item; do not force a catalyst |
search_by_analogue returns empty | Note "no relevant analogues found"; do not fabricate |
Memory Load
See contracts/memory-load.md.
Snapshot
See contracts/snapshot-synthesis.md.
Final Summary (TUI)
Include ### Key Citations block with 0-10 clickable /v/ URLs.
References
references/qual-methodology.mdcontracts/citation-and-memory.mdcontracts/output-frontmatter-schema.mdcontracts/memory-load.mdcontracts/snapshot-synthesis.mdcontracts/preflight.mdcontracts/retrieval.md
Signals
- GitHub stars
- 204
- Forks
- 16
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
qualitative-filtering- Source
- github.com/agentii-ai/agentii-investment-intelligence