Confusion Matrix i
TP — TAKE + Win
FP — TAKE + Loss
TN — AVOID + Loss saved
FN — AVOID + Win missed
Feature Importance (SHAP) i
By Direction i
Validation Stability i
Signals
Automated prediction · market conditions are read-only · sourced from Databento · generated daily at market open
—
Market Conditions read-only · Databento
▶ Show
Today's Signal
LONG ▲
—
loading…
SHORT ▼
—
loading…
Signal History & Model P&L
auto-populated from paper trading
▼
DateLongShortModel P&LLabel
Loading…
Launch Training
Configure training mode, select a strategy, and submit a new training job
Training Mode
Manual
Auto-SHAP
Auto-Search
Trains with the GFPB Approach A feature set. Select which indicators to include below.
Feature Set — GFPB Approach A
Select indicators to include in this training run:
13 / 13 selected
Strategy
Queued…
Waiting for worker…
Training History
Loading runs...
All Runs
Learning Signals — FN Trades
i
AFML Meta-labeling (Ch. 3 — López de Prado): The base strategy is the primary model; the ML classifier is the secondary meta-labeler. FN trades are where the primary was correct but the secondary filtered it out — highest-value data points for retraining.
Select a strategy to view details
Playback
Simulate strategy execution on historical days using tick data from S3
Setup
Strategy
Date Range
→
i
TP
FP
TN saved
FN missed
Market Data
Bar files shared across all strategies · sourced from Databento · universal input for training
Glossary
Term definitions from the platform