RAD

RAD gives small hedge funds real-time risk visibility by pulling live data from custodians and markets, replacing slow Excel reports with instant, unified exposure and alerts.

Comment

RAD: Real-Time Aggregated Data for Small Hedge Funds

Inspiration

In our research and conversations with people working in finance, we kept hearing the same story:

“I spend HOURS every morning downloading files from multiple prime brokers, merging them in Excel, and hoping nothing broke… and even then, our risk numbers are already outdated.”

Small hedge funds don’t have massive quant teams or expensive risk systems like the big firms.
They rely on Excel + email + end-of-day batch reports to understand risk.
That’s insanely dangerous in fast markets.

We realized one key truth:
If a small fund blows up, it’s not because they didn't know finance…
it’s because they didn’t have real-time visibility.

So we asked:
What if we build the risk platform small hedge funds WISH they had? Fast. Automated. Real-time.

That's how RAD was born.

## What We Built RAD (Real-time Aggregated Data)** is a mini real-time risk engine that:

  • Ingests position files from multiple custodians (CSV/Excel)
  • Uses AI/rules to standardize messy formats
  • Stores positions in Snowflake (clean, scalable, queryable)
  • Streams live market prices from Alpaca WebSocket
  • Calculates real-time P/L and % change: [ \text{PctChange} = \frac{\text{LivePrice} - \text{SODPrice}}{\text{SODPrice}} ]
  • Flags positions when losses exceed a threshold
  • Pushes updates to a live dashboard via WebSockets

It’s basically “Bloomberg Risk” for small funds… without the Bloomberg price tag.

How We Built It

 Backend:
  • FastAPI + Uvicorn (REST APIs & live WebSockets)
  • Snowflake (data warehouse + real-time SQL risk logic)
  • Alpaca API (live market data)
  • In-memory + SQL MERGE for position consolidation
  • AI placeholder for custodian normalization (LLM-ready design)

Frontend:

  • Clean upload interface
  • Real-time position and risk flag dashboard

What We Learned

  • Custodian files are never consistent, every broker has its own format
  • Finance workflows are still surprisingly manual
  • Snowflake is powerful enough to handle real-time calculations with SQL
  • WebSockets make risk feel ALIVE
  • Building for small funds requires simplicity AND speed

Challenges

  • Designing a standard schema that could handle any custodian
  • Getting WebSocket reconnections stable
  • Snowflake allows only one statement per execute (forced us to split some logic)
  • Handling timezone and timestamp accuracy
  • Balancing real-time processing with performance
  • Building something “institutional-grade” in 24 hours

Final Result

We turned a broken, manual, risky workflow into:

  • Fully automated
  • Real-time
  • Scalable
  • Hedge-fund ready (even small ones)

Small funds deserve institutional tools.
RAD gives them superpowers.

Real-time Aggregated Data.
Risk in real time. Finally.

Built With

Share this project:

Updates