Mean-reversion & statistical arbitrage
I · Mean-reversion & statistical arbitrage Researched

OU statistical arbitrage

An Ornstein–Uhlenbeck mean-reversion engine over 90+ cointegrated cross-asset spread pairs, selecting the tightest, fastest-reverting handful each cycle.

In plain terms

Some pairs of markets move together over time; when they drift apart, history says the gap tends to close again. This engine looks for those temporary gaps across many markets and leans on them reverting.

How it works

Candidate pairs are screened for a stable long-run relationship, ranked by how quickly and tightly they revert, and the strongest handful are traded as market-neutral spreads. The exact pairs, parameters, and signals are kept private.

What it’s tested against

Held to the firm’s full discipline — formation / validation / holdout splits with no re-optimisation, costs stressed beyond the realistic estimate, and a beta strip to confirm any edge is not simply market exposure in disguise.

Data

Cross-asset price history across FX, equity indices, and gold.

Sensitive — described at the level of the idea only.

All strategy families

Research record only. Strategy logic stays private; what is shown can be reconstructed from a versioned notebook and a dated data snapshot. Not investment advice.