Risk Manager
Position sizing, Value-at-Risk, stress testing, correlation analysis, and risk-adjusted performance metrics. Institutional-grade risk control for any portfolio.
When to Use
USE this skill when:
- “how much should I allocate to ETH” / “position size”
- “what’s my portfolio VaR” / “value at risk”
- “stress test my portfolio” / “what if BTC drops 50%”
- “correlation between my holdings”
- “Sharpe ratio” / “risk-adjusted returns”
- “am I too concentrated” / “diversification check”
- “Kelly criterion for this trade”
- “max drawdown analysis”
When NOT to Use
DON’T use this skill when:
- User wants to execute a trade — use fin-trading
- User just wants current prices — use fin-market-data
- User wants asset research or analysis — use fin-expert
- User wants to set up DCA plans — use fin-dca-strategy
- User wants news or sentiment — use fin-news-intel
Tools
Existing Tools
fin_portfolio_positions— fetch current portfolio holdingsfin_portfolio_balances— fetch account balancesfin_portfolio_history— historical portfolio snapshots for drawdown analysisfin_market_price— historical OHLCV data for VaR, correlation, and volatility calculationsfin_ticker_info— 24h metrics, volume, and market cap context
Risk-Specific Tools (Documented)
-
fin_risk_var— compute Value-at-Risk and Conditional VaR- Parameters:
portfolio_id,confidence_level(0.95 | 0.99),method(historical | parametric),horizon_days(1 | 5 | 10) - Returns: VaR amount, CVaR (Expected Shortfall), contribution by position
- Parameters:
-
fin_risk_stress_test— run stress scenarios against portfolio- Parameters:
portfolio_id,scenario(crypto_winter | flash_crash | defi_contagion | rate_hike | black_swan | custom),custom_shocks(optional asset-level overrides) - Returns: portfolio P&L under scenario, per-position impact, margin call risk flag
- Parameters:
-
fin_risk_correlation— rolling correlation matrix- Parameters:
assets[],window(30 | 60 | 90 days),method(pearson) - Returns: correlation matrix, highly-correlated pairs (>0.85), effective diversification ratio
- Parameters:
-
fin_risk_position_size— calculate optimal position size- Parameters:
method(fixed_fractional | kelly | half_kelly | mean_variance),capital,risk_per_trade(%),win_rate,avg_win,avg_loss,target_asset - Returns: recommended position size, risk amount, expected value per trade
- Parameters:
Risk Methodology
1. Position Sizing
Calculate optimal allocation using one of four methods:
- Fixed Fractional Risk: Risk a fixed percentage (1-3%) of total capital per trade. Position size = (Capital x Risk%) / (Entry - StopLoss).
- Kelly Criterion: f* = (bp - q) / b where b = avg_win/avg_loss, p = win_rate, q = 1 - p. Maximizes long-term geometric growth.
- Half-Kelly (recommended): f*/2. Sacrifices ~25% of growth for ~50% less volatility. Default recommendation for most users.
- Mean-Variance Optimization: MPT-based allocation targeting the efficient frontier. Requires return estimates and covariance matrix.
Always present all methods side-by-side and recommend Half-Kelly for most individual investors.
2. Value-at-Risk (VaR)
Three calculation approaches:
- Historical VaR: Sort portfolio returns over lookback window, find the loss at the chosen percentile. Non-parametric, captures fat tails.
- Parametric VaR: Assume normal distribution, VaR = mu - z * sigma. Fast but underestimates tail risk in crypto.
- Conditional VaR (Expected Shortfall): Average of all losses beyond the VaR threshold. Always report CVaR alongside VaR — it answers “when things go wrong, how bad?”
Default: 95% confidence, 1-day horizon. Always report in both absolute ($) and percentage terms.
3. Stress Testing
Five predefined scenarios plus custom:
| Scenario | BTC | ETH | Alts | Stables | Description |
|---|---|---|---|---|---|
| Crypto Winter | -60% | -70% | -80% | -1% | Prolonged bear market (2022 analog) |
| Flash Crash | -30% | -35% | -50% | 0% | Sudden liquidation cascade |
| DeFi Contagion | -20% | -40% | -60% | -5% | Smart contract exploit / protocol failure |
| Rate Hike Shock | -15% | -20% | -30% | 0% | Unexpected Fed tightening |
| Black Swan | -50% | -60% | -90% | -10% | Exchange collapse / systemic event |
For custom scenarios, accept per-asset percentage shocks and compute aggregate portfolio impact.
4. Risk-Adjusted Performance Metrics
Calculate and present in a summary table:
- Sharpe Ratio: (Return - Rf) / StdDev. >1 acceptable, >2 good, >3 excellent.
- Sortino Ratio: (Return - Rf) / DownsideStdDev. Penalizes only downside volatility.
- Calmar Ratio: CAGR / MaxDrawdown. Measures return per unit of drawdown risk.
- Max Drawdown: Largest peak-to-trough decline. Report magnitude, duration, and recovery time.
- Beta: Portfolio sensitivity to BTC (crypto) or S&P 500 (traditional). Beta > 1 = amplified market risk.
- Information Ratio: (Portfolio Return - Benchmark Return) / Tracking Error. Measures active management skill.
5. Correlation Analysis
- Compute rolling Pearson correlation over 30/60/90-day windows.
- Flag highly correlated pairs (r > 0.85) — these provide minimal diversification benefit.
- Calculate the effective diversification ratio: (sum of individual VaRs) / (portfolio VaR). Higher = better diversified.
- Visualize as a correlation matrix with color coding.
Response Guidelines
- Always fetch the actual portfolio before running risk analysis — never assume holdings.
- Present VaR and stress test results in clear tables with both dollar and percentage impact.
- When recommending position sizes, show the calculation step-by-step so users understand the logic.
- Compare multiple sizing methods and explain trade-offs (growth vs. safety).
- Flag concentration risk: warn if any single position exceeds 25% of portfolio.
- Flag correlation risk: warn if effective diversification ratio is below 1.5.
- For stress tests, always show the worst-case scenario’s impact on total portfolio value.
- Include actionable recommendations: “To reduce VaR by 20%, consider reducing Position X by Y%.”
Risk Disclosures
- Risk metrics are based on historical data and statistical models. Past performance and historical correlations do not guarantee future results.
- VaR does not capture the magnitude of losses beyond its threshold — always review CVaR alongside VaR.
- Stress test scenarios are illustrative. Real market events may produce outcomes outside these ranges.
- Position sizing recommendations are mathematical frameworks, not financial advice. Always consider your personal financial situation and risk tolerance.
- Crypto markets exhibit higher volatility and fatter tails than traditional markets. Parametric VaR may significantly underestimate risk.