---
id: agiprolabs/claude-trading-skills/risk-management
version: "3047bf3f"
license: MIT
install: manual
updated: 2026-06-24
---

# risk-management — Risk Management establishes a survival-first framework for protecting trading capital through drawdown thresholds, daily and weekly loss limits, and concentration controls across tokens and strategies. It includes automated circuit breakers triggered by consecutive losses or volatility spikes, recovery math tables, and crypto-specific safeguards for smart contract, rug pull, and MEV risks.

Publisher: agiprolabs · Stars: 248 · Updated: 2026-06-24

Install (manual): `git clone https://github.com/agiprolabs/claude-trading-skills`

## SKILL.md

# Risk Management

Portfolio-level risk controls for crypto and Solana trading. This skill provides frameworks for drawdown management, exposure limits, circuit breakers, and crypto-specific risk considerations.

## Risk Management Hierarchy

Every decision must respect this priority order:

1. **Survival** — Never risk account ruin. No single trade, day, or week should threaten your ability to continue trading.
2. **Capital preservation** — Protect what you have. Losses compound geometrically; recovery requires outsized gains.
3. **Growth** — Only after survival and preservation are secured, pursue returns.

Violating this hierarchy (chasing growth at the expense of survival) is the primary cause of account blowups.

## Portfolio-Level Controls

### 1. Maximum Drawdown Limits

Halt trading when portfolio drawdown from equity peak reaches a threshold:

| Account Type | Max Drawdown | Action |
|---|---|---|
| Conservative | -15% | Full stop, review all strategies |
| Moderate | -20% | Full stop, reduce to minimum size on recovery |
| Aggressive | -25% | Full stop, mandatory cooling period |

Recovery math makes this critical: a -20% drawdown requires +25% to recover. A -50% drawdown requires +100%. See `references/drawdown_management.md` for the full recovery table.

### 2. Daily Loss Limits

Stop opening new positions after daily P&L (realized + unrealized) hits:

- **Conservative**: -3% of account
- **Moderate**: -4% of account
- **Aggressive**: -5% of account

Reset at midnight UTC. Three consecutive days hitting the daily limit triggers a weekly halt.

### 3. Weekly Loss Limits

Reduce size or halt after weekly P&L reaches:

- **Reduce size by 50%**: -5% weekly loss
- **Minimum size only**: -7% weekly loss
- **Full halt**: -10% weekly loss

### 4. Concentration Limits

Maximum allocation to any single dimension:

| Dimension | Max Concentration |
|---|---|
| Single token (blue chip) | 10% of account |
| Single token (mid-cap) | 5% |
| Single token (small-cap) | 2% |
| Single token (PumpFun/micro) | 0.5% |
| Single sector/narrative | 30% |
| Single strategy | 40% |

### 5. Exposure Limits

Total deployed capital constraints:

- **Normal conditions**: 50–80% deployed, 20–50% cash reserve
- **Elevated risk**: 30–50% deployed
- **Drawdown >10%**: 20–30% deployed
- **Max concurrent positions**: 5–10 depending on account size

### 6. Correlation Management

Crypto assets correlate >0.7 during sell-offs. Effective diversification requires:

- Treat all meme tokens as a single correlated bucket
- Limit total meme exposure to one position-size equivalent
- Diversify across *strategies* (trend, mean-reversion, scalp), not just tokens
- Monitor rolling correlation and reduce when correlations spike

See `references/exposure_limits.md` for detailed limits by token type and strategy.

## Drawdown Management

### Response Framework

| Drawdown | Status | Response |
|---|---|---|
| 0–5% | Normal | Continue trading at full size |
| 5–10% | Caution | Reduce position sizes by 25–50% |
| 10–15% | Warning | Minimum position sizes only |
| 15–20% | Critical | Halt new trades, manage existing positions only |
| >20% | Emergency | Full stop, review everything before resuming |

### Recovery Requirements

| Loss | Required Gain to Recover |
|---|---|
| -5% | +5.3% |
| -10% | +11.1% |
| -15% | +17.6% |
| -20% | +25.0% |
| -30% | +42.9% |
| -40% | +66.7% |
| -50% | +100.0% |

The asymmetry accelerates rapidly. Managing small drawdowns prevents them from becoming catastrophic. See `references/drawdown_management.md` for the full framework.

## Circuit Breakers

Automated controls that restrict trading when conditions are met:

### Time-Based
- No trading for 24 hours after hitting daily loss limit
- 48-hour cooling period after weekly loss limit
- Mandatory weekly review day (no new positions)

### Loss-Based
- 3 consecutive losses → reduce size 50%
- 5 consecutive losses → minimum size only
- 7 consecutive losses → halt 24 hours, full review

### Volatility-Based
- Portfolio volatility >2× rolling average → reduce exposure 50%
- Market-wide liquidation events → pause all new entries
- Individual token volatility spike → exit or tighten stops

### Emotional (Self-Assessed)
- Recognize tilt: anger after losses, urge to "make it back"
- FOMO: rushing entries without proper analysis
- Overconfidence: increasing size after a win streak without justification

See `references/circuit_breakers.md` for implementation details.

## Risk Metrics

### Value at Risk (VaR)

95th-percentile daily loss estimate using historical returns:

```python
import numpy as np

def historical_var(returns: list[float], confidence: float = 0.95) -> float:
    """Calculate historical VaR at given confidence level."""
    sorted_returns = sorted(returns)
    index = int((1 - confidence) * len(sorted_returns))
    return abs(sorted_returns[index])

# Example: 95% VaR of 3.2% means on 95% of days, loss won't exceed 3.2%
```

### Expected Shortfall (CVaR)

Average loss in the worst (1 - confidence)% of scenarios:

```python
def expected_shortfall(returns: list[float], confidence: float = 0.95) -> float:
    """Average loss beyond VaR threshold."""
    sorted_returns = sorted(returns)
    index = int((1 - confidence) * len(sorted_returns))
    tail = sorted_returns[:index]
    return abs(sum(tail) / len(tail)) if tail else 0.0
```

### Maximum Drawdown

```python
def max_drawdown(equity_curve: list[float]) -> float:
    """Peak-to-trough decline as a fraction."""
    peak = equity_curve[0]
    max_dd = 0.0
    for value in equity_curve:
        peak = max(peak, value)
        dd = (peak - value) / peak
        max_dd = max(max_dd, dd)
    return max_dd
```

### Additional Metrics

- **Win/loss streak tracking**: Detect hot/cold streaks for circuit breaker logic
- **Rolling Sharpe ratio**: 30-day rolling risk-adjusted returns
- **Calmar ratio**: Annualized return / max drawdown
- **Sortino ratio**: Return / downside deviation (penalizes only negative volatility)

## Crypto-Specific Risks

### Smart Contract Risk
- Never allocate >5% of account to a single unaudited protocol
- Diversify across audited protocols for yield strategies
- Monitor exploit databases and social channels for emerging threats

### Rug Pull Risk
- Size inversely with token age: newer tokens get smaller positions
- Verify: locked liquidity, renounced mint authority, holder distribution
- Cross-reference with `token-holder-analysis` skill for red flags

### Bridge and Custody Risk
- Don't hold >20% on any single platform or bridge
- Self-custody the majority of trading capital
- Budget for bridge fees and delays in execution planning

### MEV and Execution Risk
- Budget 1–3% for MEV/slippage on Solana DEX trades
- Use priority fees during congestion
- See `slippage-modeling` skill for detailed cost estimation

### Correlation Spikes
- In crashes, crypto correlations approach 1.0
- Your "diversified" portfolio may behave as one position
- Stress-test portfolio assuming all positions drop simultaneously

## PumpFun Risk Framework

PumpFun and similar meme token platforms require a distinct risk approach:

### Core Principle
Treat every PumpFun trade as a potential 100% loss. Size accordingly.

### Position Limits
- **Per-token maximum**: 0.1–0.5 SOL
- **Daily PumpFun budget**: Fixed allocation (e.g., 2 SOL/day)
- **Never exceed budget**: When daily allocation is gone, stop

### Tracking
- Track PumpFun P&L separately from main portfolio
- Calculate PumpFun win rate and expectancy independently
- Don't let PumpFun losses affect main portfolio risk limits

### Risk Adjustments
- No stop-losses on PumpFun (assume 100% loss at entry)
- Take profits aggressively: 2×, 3×, 5× partial exits
- Time-based exit: close within hours, not days

## Integration with Other Skills

- **`position-sizing`**: Use risk limits from this skill to constrain position sizes
- **`exit-strategies`**: Circuit breakers override exit strategies (forced exits)
- **`portfolio-analytics`**: Feed portfolio metrics back for risk assessment
- **`liquidity-analysis`**: Adjust position limits based on available liquidity
- **`slippage-modeling`**: Factor execution costs into risk calculations

## Files

### References
- `references/drawdown_management.md` — Drawdown math, response framework, causes, and remediation
- `references/exposure_limits.md` — Position limits by token type, portfolio limits, correlation management
- `references/circuit_breakers.md` — Implementation details for all circuit breaker types

### Scripts
- `scripts/risk_dashboard.py` — Portfolio risk dashboard with limit checking and color-coded status
- `scripts/drawdown_analyzer.py` — Equity curve drawdown analysis with response recommendations

## Quick Start

```bash
# Run the risk dashboard with demo data
python scripts/risk_dashboard.py --demo

# Analyze drawdowns on a demo equity curve
python scripts/drawdown_analyzer.py --demo
```


[View on SkillFed](https://skillfed.io/agiprolabs/claude-trading-skills/risk-management) · [View on GitHub](https://github.com/agiprolabs/claude-trading-skills)