How gohnjalt/bubble-methodology works
Bubble Methodology turns market prices and inflation data into rolling bubble signals, then publishes a static dashboard for comparison across markets and trend windows. Its main workflow is data retrieval and caching, real-price construction, rolling-trend scoring, and site-data generation; a separate macro overlay adds context without changing the bubble signal. The repository also includes window-length evaluation and causal-filter studies. The README documents the dashboard capabilities; sampled excerpts do not show the full market registry or UI implementation, so those links are limited to documented behavior.
Main entry points: data.py Market loader · filters.py Causal filter study · evaluate.py Window evaluation · bubble.py Bubble model
Components 11
Data ingestion
- Market data feeds
- Market loader
data.py - Parquet cache
data.py
Signal models
- Bubble model
bubble.py - Macro overlay
overlay.py
Dashboard
- Site data builder
build.py - Static dashboard
site/index.html
Signal research
- Window evaluation
evaluate.py - Causal filter study
filters.py - Study tables
results
Other components
- Market viewer
Connections 14
- Market data feeds → Market loader: supplies series
- Market loader → Parquet cache: reads and refreshes
- Market loader → Bubble model: scores real prices
- Market loader → Macro overlay: builds macro score
- Site data builder → Market loader: loads markets
- Site data builder → Static dashboard: writes dashboard data
- Market viewer → Static dashboard: views signals
- Window evaluation → Market loader: loads market data
- Window evaluation → Bubble model: tests windows
- Window evaluation → Study tables: writes study tables
- Causal filter study → Market loader: loads market data
- Causal filter study → Bubble model: uses baseline model
- Causal filter study → Window evaluation: uses return metric
- Causal filter study → Study tables: writes filter tables