Upload README.md with huggingface_hub

README.md

@@ -8,26 +8,20 @@ tags:
 license: mit
 ---
 
-# PutStrike iTransformer — Universal Stock Forecasting Model
+# PutStrike iTransformer — Per-Stock Forecasting Models
 
-**iTransformer** (ICLR 2024) trained on 90 stocks for 60-day price forecasting.
+**iTransformer** (ICLR 2024) — individual models trained per stock for 60-day price forecasting.
 
-## Model Details
+## Per-Stock Models
 
-- **Architecture**: iTransformer with RevIN (427,641 parameters)
+**89 individual per-stock models**, each under `per_stock/{SYMBOL}.onnx`.
+
+- **Architecture**: iTransformer with RevIN (427,641 parameters each)
 - **Input**: 60 days x 83 features (OHLCV technicals + macro)
 - **Output**: 60-day forward return forecast
-- **Training**: 208,970 samples, walk-forward validation (70/15/15)
-- **Loss**: HuberLoss(delta=0.02)
-
-## Test Metrics
-
-| Horizon | Directional Accuracy |
-|---------|---------------------|
-| 7-day   | 66.1% |
-| 14-day  | 70.1% |
-| 30-day  | 73.0% |
-| 60-day  | 74.0% |
+- **Training**: Walk-forward validation (70/15/15), HuberLoss(delta=0.02)
+- **Average 7-day directional accuracy**: 50.9%
+- **Average 30-day directional accuracy**: 55.8%
 
 ## Usage
 
@@ -35,7 +29,9 @@ license: mit
 import onnxruntime as ort
 import numpy as np
 
-session = ort.InferenceSession("itransformer.onnx")
+# Load per-stock model
+session = ort.InferenceSession("per_stock/AAPL.onnx")
+
 # features shape: (1, 60, 83)
 output = session.run(None, {"features": features})[0]
 # output shape: (1, 60) — predicted returns