Premium reversion

Fades short-horizon dislocations between perp premium, spot-adjusted fair value, and local funding pressure.

Paper status

+18.6%

Out-of-sample Sharpe

1.94

Max drawdown

-8.9%

API route preview

/v1/features/crypto/premium-reversion

Function docs Example notebooks

How it is built

Universe limited to liquid crypto perpetuals with stable spot reference pricing.
Premium is smoothed over a short window before ranking to suppress microstructure noise.
Portfolio is cross-sectional and dollar-neutral, with caps per instrument and venue.
Live paper track is updated after the session close and compared against the historical reference layer.

Where it can fail

Underperforms during sustained one-way momentum squeezes where premium persists.
Sensitive to venue-specific outages and sharp liquidity fragmentation.
Capacity is lower than slower carry signals because turnover rises in stressed tape.

How to call it in code

universe = al.universes.crypto_perps_top_20()
signal = al.features.premium_reversion(universe)

Read the library docs

Planned API route

GET /v1/features/crypto/premium-reversion
  ?as_of=2026-04-30
  &universe=crypto-perps-top20
  &format=panel

Related signals

Funding basis carryCorr 0.28

Realized vol regimeCorr -0.19

Liquidity imbalanceCorr 0.34

Example notebooks

Single-feature premium reversion

Shows one signal becoming one strategy, one verified backtest, and one paper-tracked run.

Python notebook / Verified strategy walkthrough

Market-neutral premium basket

Combines signals into a market-neutral strategy, verifies the backtest, then starts paper tracking.

Python notebook / Construction workflow

Universe-change safe pipeline

Shows how Alphora's universe tools preserve point-in-time correctness before anything reaches a verified backtest.

Python notebook / Universe workflow

AI agent library composition reference

A short reference for getting an LLM or agent to compose the library, submit verified backtests, and inspect paper results safely.

Python + JavaScript snippets / Agent workflow

Browse all examples

Next places to look

AI agent guide

Give your agent one page of rules so it uses this signal correctly and checks the right caveats first.

Pricing

See which tier includes full library access, paper tracking, and reproducible example notebooks.

Signal launch

The first release ships when each signal page can stand on its own.

Each launch signal page needs a clear research summary, backtest context, code example, and paper-trading status. Join the waitlist if you want the first catalogue drop when access opens.

Join launch waitlist

One launch email. No funnel.