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

Decompose the alt return covariance into principal components; the top eigenportfolios are systematic risk, the residual (idiosyncratic) returns mean-revert. Trade the residual, hedged against the top factors.

Eigendecompose the correlation matrix $C = V\Lambda V^\top$, keep top-$k$ factors $F$, regress each asset on them:

$$ r_{i,t} = \sum_{j=1}^{k}\beta_{ij}F_{j,t} + \tilde r_{i,t} $$

Trade the cumulative idiosyncratic residual as an OU process (cf. Avellaneda & Lee 2010):

$$ s_{i,t}=\sum_{\tau\le t}\tilde r_{i,\tau},\qquad \text{score}_i = -\frac{s_{i,t}-\bar s_i}{\sigma_{s_i}} $$

Top-5 eigenportfolios as factors on 40 alts, daily refit, dollar-neutral residual book, 1h replay with full two-sided fees + funding + borrow.

Gross Sharpe (residual book)1.9
Turnover~340%/week
Net Sharpe after fees0.2
Eigenvector stability (top-5)low — rotates weekly
KILLED
Gross signal is real (Sharpe ~1.9) but turnover kills it: ~340%/week at crypto taker fees leaves Sharpe ~0.2 net. Eigenvectors also rotate, so the factor hedge is stale. Works for an equities desk at 5 bps, not at crypto cost structure.
A high gross Sharpe with high turnover is a fee-rebate strategy, not an alpha strategy. The cost structure decides which textbook strategies are even reachable.

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.