# Starter Dashboards

Superset dashboard definitions for Stonks Oracle research, analysis, and trading review.

## Dashboards
- Symbol Overview — company profiles, source health, recent documents, and market snapshots
- Sentiment Heatmap — market-wide sentiment by sector and symbol, catalyst analysis, contradiction tracking
- Prediction Accuracy — predicted signals vs realized price moves, confidence calibration, per-symbol accuracy
- Paper Trading PnL — cumulative PnL, daily performance, position snapshots, order history, and scorecards

## Data Sources
These dashboards query the Trino `lakehouse` catalog over MinIO-backed analytical fact tables:
- `lakehouse.stonks.documents` — ingested document metadata
- `lakehouse.stonks.document_extractions` — AI extraction outputs
- `lakehouse.stonks.trade_signals` — aggregated trend signals
- `lakehouse.stonks.market_bars` — OHLCV bar data
- `lakehouse.stonks.prediction_vs_outcome` — prediction accuracy tracking
- `lakehouse.stonks.pnl_daily` — daily PnL records
- `lakehouse.stonks.positions_daily` — end-of-day position snapshots
- `lakehouse.stonks.trade_orders` — order submission records
- `lakehouse.stonks.trade_fills` — fill and execution records

## Setup
1. Import the dashboard JSON files into Superset via the Superset UI or CLI
2. Ensure the Trino datasource is configured: `trino://trino@trino:8080/lakehouse/stonks`
3. Create the lakehouse views from `lakehouse/views/` for additional drill-down capability

## Trino Connection
The dashboards use the default Superset Trino connection configured in `infra/superset/superset_config.py`.