Models
Model families (“Series”) and their Marks. Families give us clear intent; marks give us audited evolution. We optimize for reliability first: calibrated predictions, regime-aware evaluation, and explicit constraints so results survive frictions.
EQ Core Series
Daily tilts designed to be held, not chased. Core aims for interpretable, low-churn edges that survive across regimes. Rolling-window training, budgeted turnover, sector & symbol caps, and reliability checks (ECE) keep behavior honest.
EQ Sentinel Series
Risk sentries and overlays. Sentinel watches for volatility spikes, liquidity stress, and structural breaks. Outputs are hedges, halts, and confidence dials rather than raw bets. The goal is to cut left-tail and time-to-recover, not maximize headline return.
EQ Experimental Series
Incubator for new ideas. Experimental hosts research marks under live shadowing. We monitor calibration drift, operational health, and regime stability. Candidates that pass promote to Core or become Sentinel modules; the rest are documented and retired—no curve-fitting by omission.
