Snowflake-Labs/sfguide-liquidity-risk-agent

Snowflake-Labs/sfguide-liquidity-risk-agent

Language: Python

License: Apache-2.0

Stars: 2

Forks: 2

Open issues: 0

Created: 2026-02-04T17:27:32Z

Pushed: 2026-02-06T20:09:54Z

Default branch: main

Fork: no

Archived: no

README:

Liquidity Risk Agent - Snowflake Guide

Overview

This Snowflake Guide demonstrates how to build a Liquidity Coverage Ratio (LCR) analysis solution using Snowflake. It includes:

• LCR Dashboard: Real-time metrics and trend visualization
• What-If Scenarios: Run pre-defined scenarios to analyze impact on LCR
• AI-Powered Agent: Natural language queries via Cortex Analyst semantic view

Step-By-Step Guide

For prerequisites, environment setup, and instructions, refer to the QuickStart Guide.

Manual Deployment Steps

Step 1: Run Setup Script (Sections 1-9)

Open scripts/setup.sql in Snowflake and run Sections 1 through 9:

• Section 1: Role Setup
• Section 2: Warehouse Setup
• Section 3: Database & Schemas Setup
• Section 4: Stages Setup
• Section 5: Raw Schema Tables
• Section 6: Load Reference Data
• Section 7: Generate Sample Data
• Section 8: RAW_SANDBOX Schema (Sandbox Tables)
• Section 9: Presentation Schema Tables

Step 2: Upload Files to Stages

Upload the local files to their respective Snowflake stages:

Notebooks → LIQUIDITY_RISK_DB.NOTEBOOKS.LIQUIDITY_NOTEBOOK_STAGE:

• notebooks/LIQUIDITY_FORECAST.ipynb
• notebooks/LIQUIDITY_WHAT_IF_FORECAST_SANDBOX.ipynb
• notebooks/prod_calculations.py
• notebooks/utils.py
• notebooks/environment.yml

Streamlit → LIQUIDITY_RISK_DB.STREAMLIT.LIQUIDITY_STREAMLIT_STAGE:

• streamlit/app.py
• streamlit/environment.yml

You can use SnowSQL, Snowsight's data upload, or the PUT commands in Section 10 of setup.sql.

Step 3: Complete Setup (Sections 11-13)

Run Sections 11 through 13 of setup.sql:

• Section 11: Create Notebooks from Stage
• Section 12: Create Streamlit App from Stage
• Section 13: Semantic View & Agent Setup

Step 4: Run the Demo

1. Open the Streamlit App (LIQUIDITY_STREAMLIT) in Snowsight 2. Generate LCR Data: Navigate to the "Generate LCR Data" page and run the LIQUIDITY_FORECAST notebook 3. Run What-If Scenarios: Use the "What-if Scenarios" page to analyze different scenarios 4. Ask the Agent: Use the "Ask the Agent" page in Streamlit, or go to Snowflake Intelligence to chat with the agent directly

Example Questions for the Agent

Try asking the agent these questions to explore the liquidity data:

LCR Analysis

• "What is the current LCR and how does it trend over 30 days?"
• "On which days does the LCR fall below 100%?"
• "Show me the LCR forecast for the next 90 days"

HQLA (High Quality Liquid Assets)

• "What is the total HQLA value by asset classification?"
• "Which business unit has the highest HQLA holdings?"
• "Break down HQLA by Level 1, Level 2A, and Level 2B assets"

Cash Flows

• "What are the largest cash outflows by counterparty?"
• "Compare total inflows vs outflows over the forecast period"
• "Which business units have the highest net cash outflows?"

What-If Scenarios

• "Compare the baseline LCR to what-if scenario results"
• "How does the stress scenario impact LCR on day 30?"
• "Show me all what-if scenario definitions"

Positions & Holdings

• "What positions do we hold in government bonds?"
• "List the top 10 positions by USD value"
• "What is our exposure by security type?"

Structure

liquidity-risk-agent/
├── LICENSE
├── README.md
├── notebooks/
│ ├── LIQUIDITY_FORECAST.ipynb
│ ├── LIQUIDITY_WHAT_IF_FORECAST_SANDBOX.ipynb
│ ├── prod_calculations.py
│ ├── utils.py
│ └── environment.yml
├── streamlit/
│ ├── app.py
│ └── environment.yml
└── scripts/
└── setup.sql

Variables

| Variable | Value | |----------|-------| | PREFIX | LIQUIDITY_RISK | | DATABASE | LIQUIDITY_RISK_DB | | WAREHOUSE | LIQUIDITY_RISK_WH | | ROLE | LIQUIDITY_RISK_ROLE |