Update README_SPACE.md

README_SPACE.md

@@ -1,12 +1,28 @@
-# Deploying as a Hugging Face Space
-
-1. Create a new **Gradio** Space.
-2. Upload the **contents of `space/`** to the Space root.
-3. Add Space Secrets:
-   - `HF_TOKEN`
-   - For BigQuery: `GCP_SERVICE_ACCOUNT_JSON`, `GCP_PROJECT`
-   - For MotherDuck: `MOTHERDUCK_TOKEN`, `MOTHERDUCK_DB`
-   - Optional tracing: `LANGFUSE_PUBLIC_KEY`, `LANGFUSE_SECRET_KEY`, `LANGFUSE_HOST`
-4. Set `SQL_BACKEND` to `bigquery` or `motherduck`.
-5. Set `HF_MODEL_REPO` to your private model repo id.
-6. (Optional) Set `ORCHESTRATOR_MODEL` for the tiny CPU LLM.
+# ALCO Liquidity & IR Dashboard (MotherDuck) — Hugging Face Space
+
+
+This Space provides an offline-first dashboard for bank ALCO/ALM:
+- Liquidity: T+1 gap and maturity ladder
+- Interest-Rate Risk: duration (modified), DV01, and ±100bp parallel shock
+- Explainability: top T+1 contributors by tenor (months), segment, and currency
+- Exports: Excel (multi-sheet) and a lightweight PDF summary
+
+
+## Deploy
+1) Create a new **Hugging Face Space** (SDK: **Python**).
+2) Add the files from this repo: `app.py`, `requirements.txt`, `README.md`.
+3) In the Space **Settings → Secrets**, add `MOTHERDUCK_TOKEN`.
+4) Your source table: `my_db.main.masterdataset_v` with columns
+- `as_of_date` (DATE like `2025-09-15`)
+- `product`, `months`, `segments`, `currency`, `Portfolio_value`, `Interest_rate`, `days_to_maturity`
+5) On first run, the app creates a normalized view `my_db.main.positions_v` (Assets vs SoF bucket).
+
+
+## Usage
+- Press **Refresh** to query latest `as_of_date` and render KPIs, charts, and tables.
+- Use **Export Excel** / **Export PDF** to download artifacts for ALCO.
+
+
+## Notes
+- Sign convention: Assets=inflow (+), SoF=outflow (–). Net gap T+1 = Assets_T1 − SoF_T1.
+- Duration/convexity/DV01 use transparent approximations with tenor T in years and yield y from `Interest_rate`.