Crypto_US_yields

Compare Aave lending/borrowing APY to US Treasury yields and visualize the rate spread against BTC forward returns.

What this repo does

• Pulls Aave supply/borrow APY from a Dune query.
• Pulls US Treasury yields (6m, 2y, 5y, 10y) from FRED.
• Joins the daily series, optionally forward-filling Treasury yields.
• Computes Aave minus Treasury spreads.
• Streamlit app plots the time series and a scatter of BTC forward returns vs the spread moving average.

Repo layout

• main.py fetches Aave + FRED data, builds spreads, writes data/crypto_us_yields.parquet.
• streamlit_app.py reads the parquet + BTC 1-min CSV and renders the UI.
• data/btcusd_1-min_data.csv is the BTC price input.
• btcReturnsSpec.md documents the BTC forward returns view requirements.

Setup (uv)

• uv sync
• Create a .env file (optional) or export environment variables:
  • DUNE_API_KEY
  • FRED_API_KEY

Example .env:

DUNE_API_KEY=...
FRED_API_KEY=...

Run all scripts

1. Build the dataset (writes data/crypto_us_yields.parquet):

   uv run python main.py

   Options:

   • --query-id 4280536 (override Dune query id)
   • --output data/crypto_us_yields.parquet (custom output path)
   • --no-ffill-yields (use inner-joined business days only)

2. Launch the Streamlit app:

   uv run streamlit run streamlit_app.py

How the data flows

• main.py reads Aave rates from Dune, auto-detects the date/supply/borrow columns, and converts decimals to percent when needed.
• FRED series are requested from the earliest Aave date, merged by day, then spreads are computed for each tenor.
• streamlit_app.py loads the parquet, lets you choose rate type + tenor + MA window, and overlays BTC forward returns computed from daily median price.