Value at Risk (VaR) is a statistical risk measure that answers one specific question: what is the maximum dollar amount you can expect to lose over a given time horizon at a given confidence level? A 1-day 95% VaR of $1,200 means there is a 5% chance — roughly one trading day per month — that your portfolio loses more than $1,200 in a single session. JPMorgan popularized VaR in 1994 through its RiskMetrics publication, and it has since become the dominant risk metric in institutional finance.
Key Takeaways
- A 1-day 95% VaR of $500 on a $50,000 account means losses should stay under 1% of capital on 19 out of 20 trading days — use this as a daily position-sizing constraint, not just a reporting metric.
- Portfolio VaR is not the sum of individual position VaRs — correlation between positions can reduce total risk to as little as 85% of the naive sum, or amplify it if positions are highly correlated.
- VaR’s critical blind spot is that it tells you nothing about losses beyond the threshold; pair it with Conditional VaR (CVaR/Expected Shortfall) to understand worst-case tail exposure.
How to Calculate Value at Risk (VaR)
Three calculation methods exist, each with different assumptions and failure modes.
Historical VaR uses actual past returns with no distributional assumptions. Collect 252 trading days of daily P&L, sort returns from worst to best, and take the 13th worst result — that is the 1-day 95% threshold (5% of 252 = ~13 observations). This method captures real fat-tail events but is limited by the historical window and may miss new market regimes.
Parametric (Variance-Covariance) VaR assumes returns follow a normal distribution and uses the formula:
VaR = Position Value × Daily Volatility × Z-Score
For a 95% confidence level, Z = 1.645. SPY has a historical daily volatility of roughly 1%, so a $30,000 SPY position has a parametric 1-day 95% VaR of approximately $494 ($30,000 × 0.01 × 1.645). This method is fast but systematically underestimates tail risk — the 2008 crisis saw major banks record VaR breaches on more than 20 consecutive days, a near-statistical impossibility under normal distribution assumptions.
Monte Carlo VaR simulates thousands of return scenarios using modeled volatility and correlations. It is the most accurate method for portfolios containing options or non-linear instruments, but requires significantly more computation and input assumptions.
Quick Reference
| Aspect | Detail |
|---|---|
| Formula (Parametric) | Position Value × Daily Volatility × Z-Score (1.645 for 95%, 2.326 for 99%) |
| Common Horizons | 1-day (trading), 10-day (Basel regulatory) |
| Confidence Levels | 95% (retail/internal), 99% (regulatory Basel II/III) |
| Regulatory Standard | Basel II: 10-day 99% VaR for bank market risk capital |
| Key Limitation | Silent on losses beyond the threshold — use CVaR alongside it |
Practical Example
A trader holds a $50,000 account with three positions: 100 shares of AAPL at $175 ($17,500), one ES futures contract (margin $12,000, but notional ~$250,000), and $20,500 in SPY.
After collecting 252 days of historical daily P&L across all three positions, they sort the combined daily returns and find the 13th worst day shows a portfolio loss of $1,850 — a 1-day 95% VaR of 3.7% of account equity. Their internal target is 1% ($500).
Breaking down contributions: the ES contract alone accounts for roughly $1,200 of that VaR, given S&P 500 futures daily range typically runs 25-35 points ($1,250-$1,750 per standard contract) under normal conditions. The fix is to replace the full ES contract with a Micro ES contract (1/10th the notional), cutting the futures VaR contribution from ~$1,200 to ~$120. Combined portfolio VaR falls to approximately $780 — not yet at the $500 target, but a 58% reduction by adjusting a single position.
Value at Risk, or VaR, tells you the maximum loss your portfolio should suffer on a typical bad day. A 95% one-day VaR of one thousand dollars means only one trading day per month should exceed that loss. It helps traders size positions to stay within a daily risk budget.
Common Mistakes
-
Treating VaR as a worst-case loss. VaR is a threshold, not a ceiling. A 95% VaR of $800 means 5% of days will exceed $800 — sometimes by $5,000 or more. Conditional VaR measures the average of those tail losses and should always accompany a VaR figure.
-
Summing position VaRs without correlation adjustments. Adding two $500 VaR positions does not produce a $1,000 portfolio VaR unless both positions move together perfectly. Correlation adjustments — even a simple 0.7 correlation between two S&P-linked positions — can reduce combined VaR to roughly $850 instead of $1,000.
-
Using too short a lookback window. A 30-day historical VaR window misses the volatility regime captured by a full 252-day window. Low-volatility periods produce artificially low VaR estimates that fail during sudden market stress.
-
Calculating VaR once and ignoring it. Volatility regimes shift. A VaR calculated during a calm market will understate risk in a trending or crisis environment. Recalculate at minimum weekly if you carry overnight positions.
How JournalPlus Tracks Value at Risk
JournalPlus logs every trade’s daily P&L, giving traders the raw data needed to calculate Historical VaR directly from their actual account performance. The max drawdown and risk-per-trade metrics in the analytics dashboard provide natural companion inputs — a trader targeting 1% daily VaR can cross-reference their average daily exposure against their Sharpe ratio to confirm whether the risk taken is generating proportional returns.
