LiquidEye — Liquidity Risk Monitor

A serverless AWS system that monitors a bank's Liquidity Coverage Ratio (LCR) in real time, applying Basel III regulatory calculations and alerting risk teams when thresholds are breached.

What it does

When a treasury system drops a daily cash flow position file into S3, LiquidEye automatically:

1. Parses the position data (HQLA levels, cash outflows, cash inflows)
2. Calculates the Basel III LCR with proper haircuts (Level 1: 0%, Level 2A: 15%, Level 2B: 50%) and the 75% inflow cap
3. Stores the timestamped result in DynamoDB for historical tracking
4. Publishes custom CloudWatch metrics for dashboard visualisation
5. Alerts via email (SNS) if the LCR drops below the 100% regulatory threshold

Architecture

Treasury System → S3 Bucket → Lambda (LCR Calculation)
                                  ├── DynamoDB (Result History)
                                  ├── CloudWatch (Dashboard Metrics)
                                  └── SNS → Email (Breach Alerts)

Services used: S3, Lambda (Python 3.12), DynamoDB, SNS, CloudWatch

The LCR Formula

LCR = Effective HQLA / Net Cash Outflows (30-day horizon) × 100

Where:

• Effective HQLA = Level 1 assets (100%) + Level 2A assets (85%) + Level 2B assets (50%)
• Net Cash Outflows = Total Outflows − min(Total Inflows, 75% × Total Outflows)
• Regulatory minimum: LCR must be ≥ 100% under Basel III

Data Schema

Each position file is a JSON document representing a single day's liquidity snapshot:

{
  "bank_id": "BANK_001",
  "report_date": "2026-03-14",
  "currency": "GBP",
  "hqla": {
    "level_1": 250000000,
    "level_2a": 60000000,
    "level_2b": 15000000
  },
  "cash_outflows_30d": {
    "retail_deposits": 95000000,
    "unsecured_wholesale": 230000000,
    "secured_funding": 70000000,
    "other_outflows": 55000000
  },
  "cash_inflows_30d": {
    "secured_lending": 45000000,
    "retail_inflows": 20000000,
    "other_inflows": 12000000
  }
}

Sample Data

The sample-data/ directory contains 6 daily position files simulating a liquidity stress event:

Date	Effective HQLA	Net Outflows	LCR	Status
2026-03-11	£617M	£260M	237.3%	Healthy
2026-03-12	£578M	£274M	210.8%	Healthy
2026-03-13	£398M	£320M	124.4%	Healthy
2026-03-14	£309M	£373M	82.7%	BREACH
2026-03-15	£239M	£425M	56.2%	BREACH
2026-03-16	£558M	£264M	211.2%	Recovery

Breach Alert Example

When LCR drops below 100%, the risk team receives an email like this:

=====================================
  LIQUIDITY COVERAGE RATIO BREACH
=====================================

Bank:            BANK_001
Report Date:     2026-03-14
Currency:        GBP

LCR:             82.71%
Threshold:       100.0%
Shortfall:       17.29 percentage points

Effective HQLA:  GBP 308,500,000.00
Net Outflows:    GBP 373,000,000.00

ACTION REQUIRED: Immediate review by Treasury/Risk team.
=====================================

Setup

Prerequisites

• AWS account with free tier access
• AWS CLI configured (optional, for deployment)

Deployment (Console)

1. S3: Create bucket liquideye-cashflow-positions in your preferred region
2. DynamoDB: Create table liquideye-lcr-results with partition key bank_id (String) and sort key report_date (String)
3. SNS: Create topic liquideye-lcr-alerts, subscribe your email, confirm the subscription
4. Lambda: Create function liquideye-lcr-calculator (Python 3.12), paste lambda_function.py, update the SNS_TOPIC_ARN variable
5. IAM: Attach AmazonS3ReadOnlyAccess, AmazonDynamoDBFullAccess, AmazonSNSFullAccess, and CloudWatchFullAccess to the Lambda execution role
6. S3 Trigger: Add an S3 event notification on the bucket for "All object create events" with .json suffix, pointing to the Lambda
7. CloudWatch: Create dashboard LiquidEye-LCR-Monitor with LCR_Percent line graph and Basel III threshold annotation at 100%

Test

Upload any of the sample JSON files to the S3 bucket. Check:

• CloudWatch Logs for the Lambda execution and LCR result
• DynamoDB table for the stored result
• Your email for breach alerts (files with LCR < 100%)

Future Enhancements

• Net Stable Funding Ratio (NSFR) as a second Basel III metric
• Granular run-off rates on outflow categories (retail: 5-10%, wholesale: 25-100%)
• API Gateway endpoint for real-time position queries from the DynamoDB table
• CloudWatch Alarms with automatic escalation (e.g. PagerDuty integration)
• Multi-bank support leveraging the DynamoDB partition key design
• Terraform/IaC deployment for reproducible infrastructure

Tech Stack

• Compute: AWS Lambda (Python 3.12)
• Storage: Amazon S3, Amazon DynamoDB
• Messaging: Amazon SNS
• Monitoring: Amazon CloudWatch (custom metrics + dashboard)
• Security: IAM least-privilege roles

License

MIT