How nickr145/worldmarket-screener works
WorldMarket is a local market-attractiveness screener. Its main user workflow is to enter a company description, review and confirm the parsed profile, then view ranked country candidates with labels, estimates, intervals, and caveats. The repository also contains a data-refresh and model-evaluation pipeline, plus dashboard detail, explanation, and model-card pages. Some subsystem links below rely on README descriptions or the source index; unsampled modules are represented at their exact file paths without assuming unshown calls.
Main entry points: run_panel.py Panel refresh orchestration · app.py Streamlit dashboard · services.py Dashboard services · score_cells.py Shipped cell scoring
Components 23
User interfaces
- Streamlit dashboard
screener/app/app.py - Screen page
screener/app/views/screen.py - Market detail
screener/app/views/detail.py - Label explainer
screener/app/views/explainer.py - Model card page
screener/app/views/modelcard.py - Command-line screener
screener/cli.py
Profile and screening
- Dashboard services
screener/app/services.py - Description parser
screener/parse/parser.py - Profile schema
screener/parse/schema.py - Candidate selection
screener/score/candidates.py - Labels and display rules
screener/score/labels.py - Plain-language narrator
screener/parse/narrate.py
Data preparation
- Market data ingestion
screener/ingest - Financial data cleaning
screener/clean - Panel construction
screener/features/build_panel.py - Panel refresh orchestration
screener/features/run_panel.py
Scoring and evaluation
- Growth baselines
screener/model/baselines.py - Boosted models
screener/model/boosted.py - Risk models
screener/model/risk_models.py - Shipped cell scoring
screener/score/score_cells.py - Verdict evaluation
screener/score/verdict_v2_eval.py
Other components
- Company screener user
- Anthropic API
Connections 35
- Company screener user → Streamlit dashboard: opens
- Streamlit dashboard → Screen page: routes to
screener/app/app.py - Streamlit dashboard → Market detail: routes to
screener/app/app.py - Streamlit dashboard → Label explainer: routes to
screener/app/app.py - Streamlit dashboard → Model card page: routes to
screener/app/app.py - Company screener user → Command-line screener: runs
- Screen page → Description parser: parses description
screener/app/views/screen.py - Description parser → Profile schema: validates profile
screener/parse/parser.py - Description parser → Anthropic API: sends prompt
screener/parse/parser.py - Screen page → Dashboard services: requests scoring
screener/app/views/screen.py - Screen page → Candidate selection: selects markets
screener/app/views/screen.py - Command-line screener → Description parser: parses prompt
screener/cli.py - Command-line screener → Candidate selection: selects markets
screener/cli.py - Dashboard services → Panel refresh orchestration: loads panel logic
screener/app/services.py - Dashboard services → Growth baselines: loads model helpers
screener/app/services.py - Market detail → Dashboard services: reads market data
screener/app/views/detail.py - Market detail → Plain-language narrator: requests narration
screener/app/views/detail.py - Market detail → Labels and display rules: formats labels
screener/app/views/detail.py - Label explainer → Labels and display rules: uses definitions
screener/app/views/explainer.py - Label explainer → Plain-language narrator: requests narration
screener/app/views/explainer.py - Model card page → Dashboard services: reads model data
screener/app/views/modelcard.py - Model card page → Labels and display rules: uses definitions
screener/app/views/modelcard.py - Plain-language narrator → Anthropic API: generates text
screener/parse/narrate.py - Panel refresh orchestration → Market data ingestion: loads source data
screener/features/run_panel.py - Panel refresh orchestration → Financial data cleaning: uses currency checks
screener/features/run_panel.py - Financial data cleaning → Panel construction: supports panel building
screener/clean/currency_check.py - Panel refresh orchestration → Panel construction: builds panel
screener/features/run_panel.py - Growth baselines → Panel refresh orchestration: loads panel data
screener/model/baselines.py - Risk models → Growth baselines: reuses predictors
screener/model/risk_models.py - Risk models → Boosted models: uses model definitions
screener/model/risk_models.py - Shipped cell scoring → Panel refresh orchestration: loads panel data
screener/score/score_cells.py - Shipped cell scoring → Growth baselines: uses growth models
screener/score/score_cells.py - Shipped cell scoring → Risk models: uses risk models
screener/score/score_cells.py - Shipped cell scoring → Labels and display rules: applies display rules
screener/score/score_cells.py - Verdict evaluation → Risk models: evaluates verdict signals
screener/score/verdict_v2_eval.py