Governance System

The governance system manages strategy selection, portfolio allocation, and risk controls based on market conditions and performance.

Components

Governor

Central orchestrator that:

• Detects market regimes
• Selects appropriate strategies
• Manages portfolio allocation
• Enforces risk controls
• Persists state across restarts

Regime Classifier

Identifies current market conditions:

• Bull: Strong uptrend, high momentum
• Bear: Strong downtrend, high fear
• Ranging: Low volatility, mean-reverting
• Volatile: High volatility, uncertain direction
• Funding Extreme: Unusual funding rates

Uses LLM-based classification with signal inputs.

Scorekeeper

Tracks strategy performance:

• Win rate and profit factor
• Sharpe ratio and max drawdown
• Recent performance (7d, 30d)
• Trade count and average hold time
• Risk-adjusted returns

Tripwires

Automatic risk controls:

• Max Drawdown: Pause strategy if DD exceeds threshold
• Daily Loss Limit: Stop trading after daily loss limit
• Sharpe Ratio: Disable underperforming strategies
• Correlation: Reduce allocation to correlated strategies

Strategy Selection

Plan Cards

Each strategy has metadata:

---
regime: ["bull", "ranging"]
risk_level: medium
max_allocation: 0.3
min_sharpe: 0.5
---

Selection Algorithm

1. Classify current regime
2. Filter strategies by regime compatibility
3. Check tripwire conditions
4. Score by recent performance
5. Allocate capital proportionally
6. Rebalance portfolio

Portfolio Management

Allocation Rules

• Minimum allocation: 10% per strategy
• Maximum allocation: 50% per strategy
• Maximum correlation: 0.7 between strategies
• Target portfolio volatility: 15%

Rebalancing

Periodic rebalancing based on:

• Performance drift
• Regime changes
• Risk metric violations
• New strategy additions

Configuration

[governance]
enabled = true
regime_check_interval_seconds = 3600
min_strategy_allocation = 0.1
max_strategy_allocation = 0.5

[governance.tripwires]
max_drawdown_pct = 15.0
max_daily_loss_pct = 5.0
min_sharpe_ratio = 0.5
lookback_days = 30

[portfolio]
rebalance_interval_seconds = 86400
target_volatility = 0.15
max_correlation = 0.7

State Persistence

Governor state is saved to state/governor.json:

• Current regime
• Active strategies and allocations
• Performance history
• Tripwire status

This allows seamless restarts without losing context.

Monitoring

Key metrics to track:

• Regime classification accuracy
• Strategy selection effectiveness
• Portfolio Sharpe ratio
• Tripwire activation frequency
• Rebalancing impact on performance