Beyond VaR: Better Risk Metrics to Use Alongside VaR

Why Value at Risk Alone Is Not Enough: Better Risk Metrics

Value at Risk is useful, but it has a critical flaw: it tells you the threshold of loss but nothing about losses beyond that threshold. If your 95% VaR is $2,000, you know you lose $2,000 on a bad day—95% of the time. But what about that 5% of days when losses are worse? VaR doesn't tell you. A market crash could hit you for $5,000 or $10,000; VaR is silent on tail severity. This chapter explores risk metrics beyond VaR that fill this gap: Conditional Value at Risk (CVaR), maximum drawdown, and stress testing. These metrics work alongside VaR to give you a complete picture of portfolio risk.

Professional investors and fund managers have known for years that risk metrics beyond VaR are essential. The 2008 financial crisis exposed VaR's weakness: countless portfolios that looked safe under VaR blew up because tail losses were far worse than the model predicted. Today, sophisticated traders use VaR as a screening tool—a first gate—but pair it with CVaR, drawdown analysis, and scenario testing to truly understand downside. This chapter shows you how.

Quick definition: Risk metrics beyond VaR include Conditional Value at Risk (the average loss beyond your VaR), maximum drawdown (the largest peak-to-trough decline), and stress tests (portfolio loss under specific market conditions). Together, these paint a complete risk picture that VaR alone cannot.

Key takeaways

• VaR is a floor, not a ceiling: it tells you the worst "normal" loss but nothing about abnormal (tail) losses.
• Conditional VaR (CVaR) / Expected Shortfall (ES) captures average tail loss; if VaR is $2,000, CVaR might be $3,500.
• Maximum drawdown is the largest peak-to-trough decline; it's easier to understand and more relevant for long-term investors than daily VaR.
• Scenario analysis tests portfolio loss under specific events (Fed rate hike, crash, volatility spike) and reveals vulnerabilities VaR misses.
• Stress testing goes further: it systematically asks "how bad could it get?" and builds in correlations that disappear in crashes.
• The holy trinity of retail risk management: use VaR (daily limit), maximum drawdown (longer-term tolerance), and one stress test (worst case).
• Combining metrics is non-negotiable: traders who use only VaR are exposed to catastrophic tail risk.

Conditional Value at Risk (CVaR): The Severity of Tail Losses

CVaR, also called Expected Shortfall (ES), answers the question: "If I exceed my VaR on a bad day, how bad on average?" It's the average of all losses worse than your VaR threshold.

Mathematical intuition:

• VaR at 95%: the 5th percentile of returns
• CVaR at 95%: the average of the worst 5% of returns

Practical example: Over 250 trading days, your portfolio's worst 5% outcomes (13 days) were:

Day 1:  -4.2%
Day 2:  -3.8%
Day 3:  -3.5%
Day 4:  -3.1%
...
Day 13: -2.1%  <-- This is VaR (5th percentile)

Average of these 13 days: (-4.2 - 3.8 - 3.5 - ... - 2.1) / 13 = -3.2%

Your 95% CVaR is -3.2% of your portfolio. On a $100,000 account, that's $3,200. Compare this to your 95% VaR of $2,100 (the 2.1% worst day). CVaR tells you that on your truly bad days (the 5% tail), you lose an average of $3,200—50% worse than VaR predicts.

Why CVaR matters: VaR says "one day per month, lose up to $2,100." CVaR says "but on those bad months, expect $3,200." That extra $1,100 makes the difference between sleep and panic.

Calculating CVaR

Using Excel:

CVaR = AVERAGE(IF(returns < PERCENTILE(returns, 0.05), returns, ""))

(Entered as an array formula: Ctrl+Shift+Enter on Windows.)

Using Python:

var_threshold = np.percentile(returns, 5)
tail_returns = returns[returns < var_threshold]
cvar = np.mean(tail_returns) * portfolio_value

Interpretation: If VaR is your "expected" worst-case daily loss, CVaR is your "actual" worst-case average loss. Many professional investors target CVaR instead of VaR because it's more realistic.

Maximum Drawdown: The Longest Pain

Maximum drawdown is the largest peak-to-trough decline your portfolio experiences. If your portfolio hits $125,000 (peak) then falls to $95,000 (trough), your maximum drawdown is 24% ($30,000). This metric captures sustained losses over weeks or months—not just daily volatility.

Why it matters:

• VaR is a daily metric; it misses multi-week losing streaks.
• Drawdown is psychological; it tells you the longest period of pain you'll endure.
• Many retail traders are OK with $2,000 daily losses, but panic if they're down 15% on their account—drawdown captures this reality.

Calculating Maximum Drawdown

Conceptually:

1. Calculate cumulative portfolio value over time (daily closing balances).
2. Find the running maximum (the highest point reached so far).
3. Calculate the decline from that running maximum: Drawdown = (Current Value - Running Max) / Running Max.
4. Find the maximum of all drawdowns.

Excel formula:

Running Max: =MAX($B$1:B1)  (drag down)
Drawdown: =(B2 - Running Max) / Running Max  (drag down)
Max Drawdown: =MIN(Drawdown column)

Python:

cumulative_value = portfolio_values.cumprod()
running_max = cumulative_value.expanding().max()
drawdown = (cumulative_value - running_max) / running_max
max_drawdown = drawdown.min()
print(f'Maximum Drawdown: {max_drawdown*100:.2f}%')

Real example: A trader's $100,000 account over 2 years:

• Jan 2023: Peaks at $135,000 (35% gain)
• Mar 2023: Falls to $108,000 (20% drawdown from peak)
• May 2023: Recovers to $150,000 (new peak)
• Sep 2023: Crashes to $105,000 (30% drawdown from peak) ← This is the max drawdown: -30%
• Dec 2024: Back to $125,000 (still below May peak)

Max Drawdown = -30%. The account lost 30 cents on every dollar at its worst point.

Drawdown vs. VaR: Different Lenses

Metric	Timeframe	Focus	Calculation
VaR	Daily	Threshold loss	Percentile of daily returns
CVaR	Daily	Tail loss severity	Average of losses beyond VaR
Max Drawdown	Weeks to months	Sustained pain	Peak-to-trough decline

When to use each:

• VaR: Position sizing, daily risk limits, margin requirements.
• CVaR: Understanding true tail risk, stress-testing portfolios.
• Max Drawdown: Long-term account sustainability, psychological tolerance, drawdown recovery time.

Scenario Analysis: What-If Testing

Scenario analysis asks: "If X happens, how much do I lose?" It's more specific than VaR. VaR assumes returns follow historical patterns; scenarios test behavior outside history.

Scenario Example 1: Fed Rate Hike Shock

Scenario: Federal Reserve raises rates 0.50% in a single meeting (an unexpected hawkish surprise).

Historical precedent: March 2022, May 2023. Fed surprises triggered broad selloffs.

Your portfolio: 60% stocks, 30% bonds, 10% cash.

Estimated losses:

• Stocks down 2% (near-term flight to safety)
• Bonds down 3% (duration risk; longer-maturity bonds hurt more)
• Cash unchanged

Portfolio loss:

0.60 × -2% + 0.30 × -3% + 0.10 × 0% = -2.1%

On a $100,000 account, a 2.1% loss = $2,100. Your daily VaR might be $1,500, so this scenario loss exceeds VaR—revealing hidden bond duration risk.

Scenario Example 2: Equity Market Crash

Scenario: S&P 500 drops 15% in one week (like March 2020, August 2011, or September 2008).

Historical precedent: COVID crash, 34% in 4 weeks; 2008, 57% over the year.

Your portfolio: 50% S&P 500 index, 30% international stocks (correlated 0.75 to SPY), 20% bonds.

Estimated losses:

• S&P 500: -15% (given)
• International: -15% × 0.75 = -11.25% (correlated)
• Bonds: +2% (flight to safety; inverse correlation to stocks)

Portfolio loss:

0.50 × -15% + 0.30 × -11.25% + 0.20 × 2% = -8.575%

On $100,000, that's $8,575 loss. Compared to daily VaR of $1,500, a crash is 5.7× worse.

Scenario Example 3: Volatility Spike (No Direction Change)

Scenario: VIX jumps from 15 to 40 (volatility quadruples) but stock prices barely move.

Why it matters: Your portfolio's Greeks show you delta (directional) risk, but vega (volatility) risk is hidden. A long gamma position benefits from vol spikes; a short gamma position suffers.

Portfolio: Long call options on tech stocks (long gamma, short vega).

Loss: Vol spike hurts short vega more than long gamma helps. Estimate: -3% on the portfolio for a VIX jump of 15 to 40.

Stress Testing: Systematic Extreme Scenarios

Stress testing goes deeper than single-scenario analysis. It runs 5-20 scenarios covering different market regimes (recession, credit crisis, geopolitical shock, commodity crunch) and examines portfolio behavior across all of them.

Real stress test framework (simplified):

Scenario 1: Recession (stocks -20%, bonds +5%, volatility +50%)
Scenario 2: Inflation Shock (stocks -15%, bonds -8%, commodities +25%)
Scenario 3: Credit Crisis (all risky assets -25%, safe assets +10%)
Scenario 4: Geopolitical (oil +20%, stocks -10%, safe havens +5%)
Scenario 5: Currency Crisis (dollar -15%, EM stocks -20%, gold +15%)
...and so on.

For each scenario, calculate your portfolio's loss. The average loss across scenarios is your "stress loss."

Example (5 scenarios, simplified):

Scenario	Stocks	Bonds	Commodities	Portfolio Loss
Recession	-20%	+5%	-10%	-11.5%
Inflation	-15%	-8%	+25%	+0.5% (bonds hedge stocks poorly)
Credit Crisis	-25%	-10%	-15%	-17.5%
Geopolitical	-10%	+3%	+20%	+3.5%
Currency Crisis	-8%	-2%	+18%	+1.2%

Average stress loss: -4.8%. If your account is $100,000, expect $4,800 loss under typical stress scenarios. Compare this to daily VaR of $1,500; stress loss is 3.2× worse.

Decision Tree: Which Risk Metric to Use When

Real-World Examples: VaR + Beyond

Example 1: Swing Trader, Tech-Heavy

Portfolio: $75,000 account, 10 stocks (all tech sector, Nasdaq correlation ~0.85)

VaR: 95% one-day VaR = $1,875 (2.5% daily volatility)

CVaR: Average of worst 5% days = -3.5% = $2,625

Max Drawdown: Historical max drawdown = -28% (during March 2020 pandemic crash)

Scenario: Tech Sector Crash (-20% in one week)

• Portfolio loss: -20% × 0.85 (correlation) = -17% = $12,750

Interpretation: Daily losses up to $1,875 feel manageable. But tail days average $2,625 losses. Worst case from history: 28% drawdown. If tech crashes 20%, expect $12,750 loss. The trader should:

1. Reduce position size if daily losses exceed $1,875
2. Prepare for $2,625+ losses on tail days (hold more cash buffer)
3. Build in stop-losses to cap 28% drawdown scenario
4. Consider hedging (put options) if concentrated tech exposure is core strategy

Example 2: Long-Term Investor, Diversified

Portfolio: $500,000 account, 60/30/10 (stocks/bonds/cash)

VaR: 95% one-day VaR = $3,250 (0.65% daily volatility)

CVaR: Average tail loss = -0.95% = $4,750

Max Drawdown: Historical = -22% (2008 crisis) = $110,000

Scenario: 2008-like crisis (stocks -50%, bonds flat to +5%)

• Portfolio loss: 0.60 × -50% + 0.30 × 0% + 0.10 × 0% = -30% = $150,000

Interpretation: Daily VaR is small ($3,250) because volatility is low. But max drawdown of 22% ($110,000) shows extended pain during crises. A true crisis (2008-like) could hit 30% ($150,000)—even worse than historical max. The investor should:

1. Accept that long-term bonds provide drawdown cushion (not protection in all crises)
2. Be prepared for 6-12 month recovery after major crashes
3. Consider increasing cash to 15-20% if 30% loss feels unacceptable
4. Rebalance regularly (sell winners, buy losers) to reduce drawdown duration

Example 3: Day Trader, Forex + Commodities

Portfolio: $50,000 account, leverage 5:1, uncorrelated assets (eur/usd, crude oil, gold)

VaR: 95% one-day VaR = $1,000 (2% daily volatility, leveraged)

CVaR: Tail loss = -3% = $1,500

Max Drawdown: Historical = -18% (over 3 weeks)

Scenario: Black Swan (all correlations spike to 0.95, all assets down 5%)

• Portfolio loss: -5% × all positions = -5% = $2,500 (on leveraged account, true loss is higher: $12,500 on $50K notional)

Interpretation: Leverage amplifies all metrics. Daily VaR of $1,000 seems manageable, but tail loss of $1,500+ is common on volatile days. In a black swan, the day trader could lose $12,500+ (the entire account). Solutions:

1. Reduce leverage from 5:1 to 2-3:1
2. Set account stop-loss: if daily loss exceeds $1,500, close all positions
3. Stress test black swans monthly; if outcome is account-wipeout, leverage is too high
4. Use options (long puts) to hedge tail risk on leveraged positions

Common Mistakes with Advanced Risk Metrics

Mistake 1: Over-Complicating Scenarios You build a 50-scenario stress test, each with hand-tweaked parameters. Half of them are never updated. A simpler 5-scenario test that you update quarterly is better. Complexity kills discipline.

Mistake 2: Ignoring Liquidity in Tail Scenarios Your stress test assumes you can exit all positions at the scenario prices. In reality, if everyone's selling (crash), bid-ask spreads widen and you can't exit. Add 1-2% slippage to tail scenario losses.

Mistake 3: Using Only Historical Worst Cases Your max drawdown is -22% from 2008. You assume that's the worst case. But the next crisis could be worse (e.g., financial system breakdown). Instead, use a 30-40% drawdown assumption for stress planning.

Mistake 4: Forgetting Leverage in CVaR and Drawdown You calculate CVaR and max drawdown assuming no leverage, but you're actually using 2:1 margin. Multiply all drawdown metrics by your leverage ratio.

Mistake 5: Testing Scenarios, Then Ignoring Results You run a stress test showing -25% loss in a recession. You nod and file it away. Six months later, a recession starts, and you're unprepared because you never acted on the stress result. Stress test outputs should trigger action: reduce position size, add hedges, or increase cash.

FAQ

How do I choose between VaR and CVaR?

Use both. VaR is your daily risk limit (for position sizing). CVaR is your true tail risk (for stress planning). A professional investor monitors both daily.

Is maximum drawdown more important than VaR for long-term traders?

Yes, for long-term traders. Long-term investors care about drawdown recovery time and psychological tolerance. A 25% drawdown over 6 months hurts more than 25 daily 1% losses. VaR and drawdown answer different questions.

How often should I recalculate CVaR and maximum drawdown?

Weekly for CVaR (tail risk changes with volatility), annually for max drawdown (historical worst case). Max drawdown is stable unless a new crash occurs; CVaR changes with market regime.

What's a "safe" maximum drawdown for retail accounts?

15-20% for risk-tolerant traders, 10% for conservative traders. Below 10%, your account is very stable (good for beginners). Above 30%, you're exposed to significant pain (acceptable only for professionals). Any trader who experiences a 50%+ drawdown should radically change their approach.

Can I use stress tests instead of VaR?

No. Stress tests are point-in-time (they assume a specific scenario happens); VaR is probabilistic (it captures all possible outcomes). Use both: VaR for daily limits, stress tests for "what-if" planning.

How do I interpret a scenario that shows bigger losses than historical max drawdown?

That's the point of stress testing. If your scenario loss (-30%) exceeds historical max drawdown (-22%), it means you're identifying a risk that history hasn't yet revealed. This is information worth acting on (reduce position size, add hedges).

What's the relationship between volatility and CVaR?

Higher volatility = higher CVaR. When volatility spikes, the left tail (losses) gets thicker, and average tail losses increase. If volatility doubles, CVaR doesn't double—it increases by roughly sqrt(2) ≈ 40%. This is why CVaR is more stable than VaR.

Should I use Monte Carlo simulation for CVaR?

Yes, if you want non-normal distributions. Historical CVaR assumes past patterns repeat. Monte Carlo allows you to model heavy tails (like crashes) explicitly. For retail traders, historical CVaR is enough.

Related concepts

• Practical VaR for a Retail Portfolio
• Using VaR to Size Positions
• VaR Calculators and Tools for Retail
• What Is Value at Risk?
• What Is a Drawdown?

Summary

VaR alone is insufficient for comprehensive risk management. It tells you the daily threshold but nothing about the severity of losses beyond that threshold. Conditional Value at Risk (CVaR) fills this gap by measuring average tail loss. Maximum drawdown captures sustained, multi-week losses that daily VaR misses. Scenario analysis and stress testing explore portfolio behavior under specific events and extreme market regimes. Together—VaR + CVaR + max drawdown + stress testing—these metrics form the foundation of professional risk management. A trader using only VaR is flying blind on tail risk. A trader using all four metrics has a complete picture and can size positions, set stops, and manage leverage with confidence.

Next

→ What Is a Drawdown?