← Projects·Quantitative Researcher
Crypto Funding-Carry
A novel-data carry study: perpetual-swap funding rates are a real, observable, regularly-paid signal — and the obvious trade is to fade them. The interesting question is whether the carry persists OOS post-2023.
+11.9%
Annualized funding
BTC perp, ~30-day centered mean
1.11
Fade IS Sharpe
−0.05
Fade OOS Sharpe
decay post-2023
The hypothesis
Perpetual-swap funding rates are paid by the leveraged long to the short (or vice versa) every 8 hours. The textbook trade is to fade the funding — collect the carry and pray the basis doesn’t blow up.
What the project does
- Pulls Binance BTC perpetual funding rates.
- Computes the centered rolling average as a “carry-as-signal” feature.
- Runs the fade (short when funding > 0, long when funding < 0) with locked IS / OOS windows.
- Pre-registers the OOS window before touching data.
The result
- The funding rate averages +11.9% annualized — a real, structural premium.
- IS Sharpe (fade): 1.11 — looks like an edge.
- OOS Sharpe: −0.05 — the carry structurally decayed post-2023 as the participant mix shifted.
What’s transferable
Carry signals are observable, regularly-paid, and visible to anyone with the data — which means they crowd fast. This is a clean example of alpha decay detected before overfit, in production reporting cadence.