Skip to main content
Hydra

Project Hydra

META-LEARNING ENGINE โ€” LEARNS FROM ALL STRATEGIES
OWNER ONLY โ€” SHADOW MODE
ArenaScript public contract
Hydra parle ArenaScript planned
Contrat public lisible pour builders, CUDA, MCP et TradingView. Le monstre garde ses internes privรฉs ; ArenaScript expose seulement l'interface validable.
๐Ÿ”’

Owner access only

Project Hydra is available only to the owner account.

Loading...
Live Trading โ€” Shadow Mode
Current signal
โ€”
Waiting for model
Capital
$10,000
PnL: โ€”
Position
โ€”
0 BTC
Trades
0
Win rate: โ€”
Consensus
โ€”
Buy vs Sell
Model Status
โ€”
0 ticks
ML Performance
Accuracy
โ€”
AUC-ROC
โ€”
Precision
โ€”
Recall
โ€”
Trades Analyzed
โ€”
Strategies
โ€”
Hydra Model Comparison
Hydra XGBoost
โ€”
Existing meta-learner
HydraLSTM Shadow
โ€”
Waiting for shadow inference
Training Curves
Loss (Train vs Validation)
AUC-ROC per Epoch
Feature Importance
Confusion Matrix
Pred +
Pred -
Actual +
โ€”
โ€”
Actual -
โ€”
โ€”
Strategy Reliability Ranking
# Strategy ML Score Win Rate Avg PnL Trades Best Regime
Regime Analysis

Hydra ML: 3 machine learning brains

Hydra now shows the existing XGBoost meta-learner and a calibrated HydraLSTM shadow head side-by-side. The LSTM is not active in production until its live shadow signals are validated.

๐Ÿ‰ HYDRA SELF-EVOLUTION (Karpathy Loop)

Every night, the Darwin Engine evolves Hydra's ML model weights by market regime. Which model works best in BULL, NEUTRAL, or BEAR? The system finds out automatically.

Inspired by Karpathy's autoresearch. See all results โ†’