ML ALPHA HUB PREDICTIVE

MODELS TRAINED
Epoch: 14,204

Top Alpha Signals

NVDA Momentum Reversal 🟢 ACTIVE
Model: LSTM-Transformer Hybrid
LONG
Confidence: 87.4% Exp. Ret: +4.2%
TSLA Volatility Breakout 🟡 RETRAINING
Model: GARCH-NeuralNet
SHORT
Confidence: 92.1% Exp. Ret: -6.8%
AMZN Supply Chain Sentiment 🟢 ACTIVE
Model: NLP Extractor (FinBERT)
LONG
Confidence: 76.5% Exp. Ret: +2.1%

Drawdown Profile

Feature Importance (SHAP)

Model Prediction vs Actual (OOS)

RMSE
0.024
Hit Rate
58.2%
Sharpe (OOS)
2.15
Max DD
-4.2%

Model Strategy Allocation

Stat Arb (Pairs)40%
NLP Sentiment25%
Vol Harvesting35%
Allocation Optimizer Note: Current model ensemble heavily weights Statistical Arbitrage strategies due to sustained historical outperformance in low-volatility regimes. NLP Sentiment allocation was dynamically reduced by 5% yesterday following a detected surge in conflicting news sentiment noise related to macro tariffs.