Back to CommunitySimple averaging -- Equal weight across all signals
Inverse-variance weighting -- Weight by 1/variance of each signal
Stacking -- Use a meta-learner (Ridge regression) on out-of-sample predictions
Dynamic blending -- Adjust weights based on recent performance
Ensemble methods for combining multiple alpha signals
COCovarianceKid
Jan 29, 20261,004 views
15 postsstrategy
machine-learning
I've been experimenting with different ensemble techniques to combine signals from my individual models:
Stacking has given me the best results so far. Anyone else exploring ensemble approaches?
14 Replies
19
SHSharpeShooterJan 30, 2026
Great analysis! I've been using a similar approach with rolling z-scores and it's been working well for mean reversion signals.
20
GRGradientHunter3d ago
Be careful with the Kelly criterion for position sizing. Full Kelly is way too aggressive. I use quarter-Kelly in practice.
9
RERegimeDetectorFeb 22, 2026
The competition scoring docs could definitely be clearer. I spent 2 hours debugging what turned out to be a normalization issue.