hajirazin/learnfinance-models-sac

LearnFinance SAC Model - v2026-01-23_4049af7e

Soft Actor-Critic (SAC) portfolio allocation agent using dual forecasts (LSTM + PatchTST) as features.

Model Details

• Version: v2026-01-23_4049af7e
• Model Type: SAC (Soft Actor-Critic) with dual forecasts
• Training Window: 2011-01-01 to 2026-01-23
• Symbols: 15 stocks

Components

• actor.pt - Gaussian policy network
• critic.pt - Twin Q-value networks
• critic_target.pt - Target Q-value networks
• log_alpha.pt - Entropy temperature coefficient
• scaler.pkl - PortfolioScaler for state normalization
• symbol_order.json - Ordered list of portfolio symbols

Metrics

• Actor Loss: 0.3213101402535821
• Critic Loss: 0.0
• Avg Episode Return: 0.2853802386733016
• Avg Episode Sharpe: 0.22613404050736566
• Eval Sharpe: 0.8972165609782691
• Eval CAGR: 0.19267435023534785
• Eval Max Drawdown: 0.13759327587214085

Usage

from brain_api.storage.sac import SACHuggingFaceModelStorage

storage = SACHuggingFaceModelStorage(repo_id="hajirazin/learnfinance-models-sac")
artifacts = storage.download_model(version="v2026-01-23_4049af7e")