Value at Risk

What is Value at Risk?

Value at Risk (VaR) answers the question: "How bad was a bad candle for this strategy?" Specifically, it finds the return at the 5th percentile -- the point where 95% of candles did better and only 5% did worse.

Think of it like a grade curve. If you sort every candle's return from worst to best, VaR is the grade that separates the bottom 5% from everyone else. It tells you where the "bad" zone starts, but it does not tell you the details of what happened inside that zone.

For example, a VaR of -3.00% means that 95% of candles had a return of -3% or better. The remaining 5% of candles lost more than 3%.

How It's Calculated

Simula uses the historical method. It collects every candle's return during the backtest, sorts them from worst to best, and finds the value at the 5th percentile. No assumptions are made about the shape of the return distribution -- VaR is read directly from the actual returns.

In plain English: line up all your candle returns from worst to best, then find the point where only 5% of returns fall below it. That point is VaR.

Worked Example

Reading Your Results

VaR is displayed as a percentage. More negative values indicate larger losses at the 5th percentile boundary.

Things to Watch Out For

VaR is a boundary, not a worst case

VaR tells you where the bottom 5% starts, but not what happened inside that bottom 5%. Two strategies can both have a VaR of -3%, but the worst candle in one might be -4% while the worst candle in the other is -30%. VaR treats those identically. If you want to know the average severity of the worst candles, look at Conditional Value at Risk (CVaR).

VaR is based on historical returns

Simula calculates VaR from the actual returns in the backtest. If the backtest period did not include a market crash or a period of extreme volatility, VaR will reflect only the conditions that were present. A different backtest period covering more extreme conditions could produce a more negative VaR.

Small sample sizes make VaR less stable

With very few candles, the 5th percentile is estimated from a handful of returns. Adding or removing a single bad candle can shift VaR significantly. Longer backtests with more candles produce more stable VaR estimates. See Statistical Significance & Trade Count and Timeframes & Data for more on how date range and candle interval affect metric stability.

Leverage amplifies VaR

If your strategy uses leverage, each candle's return is magnified. A 2% loss becomes a 4% loss at 2x leverage, pushing VaR proportionally more negative. Simula reports VaR for each leverage level separately.

Related Metrics

Conditional Value at Risk (CVaR) -- averages the returns that fell in the bottom 5%. Where VaR marks the boundary, CVaR measures how severe the losses were beyond it.

Max Drawdown -- measures the largest peak-to-trough decline in the equity curve. VaR looks at individual candle returns in isolation, while max drawdown captures the cumulative effect of consecutive losing candles.

Volatility -- measures the overall variation in returns, both up and down. VaR focuses on the downside only, looking at a specific percentile of losses rather than the average size of all movements.

Monte Carlo: Interpreting Results -- includes a VaR metric computed across thousands of reshuffled trade sequences, showing the 5th-percentile final return when the same trades arrive in different orders.