← ML Lab · V8 History Builder · Dashboard

OmmAlpha · V8 ML Trainer

V8 Phase 2A · strategy-isolated dataset creation, chronological model comparison and saved production candidates.

Runtime & dataset readiness

MISSINGPython runtime

—Python

—XGBoost

2,721V8 BACKTEST signals

942Resolved success rows

99.8%Historical RS coverage

Strategy isolation: this trainer exports only OmmAlpha-v8 BACKTEST signals. Older V2 history is not mixed into the V8 model.

Training policy: OmmAlpha compares Logistic Regression and XGBoost for each objective. The champion is chosen on the validation period, then reported on an untouched test period. It does not automatically declare XGBoost the winner.

Recent training runs

Run	Status	Rows	Python	Started	Completed
v8_20260927_190350	COMPLETED	2,721	3.13.15	2026-09-27 09:33:52	2026-09-27 09:34:01
20260926_174731	COMPLETED	4,662	3.13.15	2026-09-26 08:17:33	2026-09-26 08:17:46

What this phase does not do yet

Phase 2A trains, validates and saves the models. It does not yet place ML probabilities on the live dashboard. That is Phase 2B, after we inspect the first server-trained metrics.