Spaces:
Running
Running
Khanna, Videh Rakesh Rakesh
Fix empty universe + full-market top picks; drop dead NSE price APIs
b0d9ebf Download ml_predictor/__init__.py from V1deh/PaperTrade: direct link, hf CLI and curl.
- Browser
- Download file 796 Bytes
-
https://huggingface.co/spaces/V1deh/PaperTrade/resolve/main/ml_predictor/__init__.py
- Command line
-
hf download hf://spaces/V1deh/PaperTrade/ml_predictor/__init__.py
-
curl -L -o __init__.py https://huggingface.co/spaces/V1deh/PaperTrade/resolve/main/ml_predictor/__init__.py
796 Bytes
| """ml_predictor β standalone supervised ML price-prediction model for NSE equities. | |
| A quantile-regression predictor (sklearn HistGradientBoosting) that outputs, per | |
| timeframe (INTRADAY / 1D / 3D): a price estimate, buy-price suggestion, stop-loss, | |
| and trend (BULLISH / BEARISH / NEUTRAL). Independent of the LLM debate pipeline. | |
| Modules: | |
| features.py β shared point-in-time numeric feature builder (all indicators + ML sub-features) | |
| dataset.py β build training_data.csv from the ohlcv_cache.db (offline) | |
| train.py β fit the 15 HistGBM estimators + manifest.json | |
| infer.py β MLPredictor: predict(ticker, tf) / predict_all_tf(ticker) | |
| See CLAUDE.md and the plan for the full design. | |
| """ | |
| from .features import FEATURE_COLUMNS, TIMEFRAMES, compute_features # noqa: F401 | |