NEPSE Predator

v13 Triple-barrier · floorsheet features

Ranked probabilistic signals from your trained bundle. Use cache for speed; rebuild when raw CSVs change.

Target session

Leave empty or tap Latest for the newest date in your pivot cache. Press Enter to run.

Scoring runs in the background. Scroll, read system health, or review the table below until new rows replace the previous run (often ~10-30s).

System health

Run options

Include HOLD Show neutral rows in the table (noisier). Backtest path Slower; simulates barriers for diagnostics. Use cache pickle Recommended. Turn off only if rebuilding in-session. Past date outcomes If the session is before the newest day in cache, run triple-barrier (stop / profit / time) and show exit + return %.

Environment variables

NEPSE_PREDATOR_FLOORSHEET_DIR · NEPSE_PREDATOR_MODEL_PATH · NEPSE_PREDATOR_CACHE_PATH · NEPSE_PREDATOR_INDEX_CSV

Ready when you are

Run signals to load the model, score the universe for your session, and render ranked rows below.

Scoring your first session

The page stays open. When the run finishes, the signals table appears below.

#	Symbol	Signal	Score	Price	Target%	Stop%	DD20	Vol Z	Ret 3d	Exit / Return	Days	Fwd %	✓ TP
No symbols match your search.
													✅ ❌ —

System health

Run options

Ready when you are

Signals