← Research logStat-Arb & Pairs
PARTIAL · weak, decays
#

A market-neutral basket of correlated alts has a residual that follows an Ornstein–Uhlenbeck process; trading deviations beyond a calibrated band captures reversion with positive expectancy.

Model the de-meaned basket spread $X_t$ as:

$$ dX_t = \theta(\mu - X_t)\,dt + \sigma\,dW_t $$

Mean-reversion speed $\theta$ gives the half-life, which sets holding time:

$$ t_{1/2} = \frac{\ln 2}{\theta} $$

Optimal entry band scales with the stationary standard deviation $\sigma/\sqrt{2\theta}$.

Estimate $\theta,\mu,\sigma$ by MLE on a rolling window. Enter at $\pm1.5$ stationary-sigma, exit at $\mu$, hard time-stop at $3\times t_{1/2}$. 30 liquid USDT-M alts, path-dependent.

Half-life (median)6.5h
Gross Sharpe (in-sample)1.6
Net Sharpe (out-of-sample)0.5
Win rate58%
Edge after 4 bps + funding+0.8%/mo
PARTIAL EDGE
A small, real edge survives costs (+0.5–1%/mo) but decays fast as $\theta$ destabilises in trends. Viable only as a low-weight overlay with a strict time-stop, never standalone.
OU is the right model for a basket residual, but $\theta$ is regime-dependent. Size the trade by the estimated half-life, and respect the time-stop — reversion that does not arrive on schedule is a trend.

We publish the failures too.

This is one of 100+ documented hypotheses. Browse the full lab notebook, or see the strategies that survived into production.