← Research logStat-Arb & Pairs
KILLED · cost > tail edge
#

A copula captures non-linear, tail-dependent co-movement that linear cointegration misses; trading conditional-probability extremes of the copula beats z-score pairs.

$$ h(u\mid v) = \frac{\partial C(u,v)}{\partial v},\quad \text{trade when } h\to 0 \text{ or } 1 $$

Fit Student-t / Clayton copulas to pair return ranks, trade mispricing index extremes, costs applied.

Tail-dependence capturedyes
Extra signals vs z-scorefew
Net after fees
KILLED
The copula does model tail dependence the z-score misses, but the incremental signals are too few to cover the extra turnover, and fit instability adds noise. Sophistication without net edge. Killed.
A more flexible dependence model is only worth it if the extra signals it finds pay for themselves. Elegance is not edge.

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.