Value at risk (VaR)
Value at risk (VaR) is the loss that a position or account should not exceed over a given horizon with a given probability, for example a one-day 95% VaR of 1,300 dollars. It is a threshold, not a worst case: it says how often losses exceed it, not how large those losses are.
Senzoukria · Glossary · Updated September 2026
At a glance
- Inputs
- Confidence level, horizon, distribution of P&L
- Historical VaR
- Percentile of past P&L, e.g. 5th percentile for 95%
- Parametric VaR (normal)
- z × σ − mean; z = 1.645 at 95%
- Blind spot
- Size of losses beyond the threshold
Two ways to compute it
Historical VaR sorts past daily results and reads the loss at the chosen percentile. With 250 daily P&L values, the one-day 95% VaR is the loss exceeded on 5% of the days, the boundary between the 12th and 13th worst days. It needs no distribution assumption but depends entirely on which days are in the window.
Parametric VaR assumes a distribution, usually normal. If daily P&L has a mean near zero and a standard deviation of 800 dollars, the one-day 95% VaR is 1.645 × 800 ≈ 1,316 dollars. Under the additional assumption of independent days, a five-day VaR scales by √5, to about 2,943 dollars. Both assumptions are optimistic for trading results, whose tails are usually heavier than the normal distribution and whose bad days cluster.
What VaR does not tell you
- How bad the days beyond the threshold are: two strategies with the same VaR can have very different worst days. Expected shortfall answers that.
- Path risk: a daily VaR says nothing about a string of moderate losses that adds up to a large drawdown.
- Intraday excursion: a daily figure computed on closing results misses losses that recovered before the close but could have breached an intraday account rule.
- VaR is not always subadditive: the VaR of two positions combined can exceed the sum of their VaRs, which makes it awkward for aggregating risk.
In Senzoukria
The desktop does not compute VaR. The quantitative indicator catalogue excludes it as a chart study, because it is defined on periodic portfolio returns and a holding horizon, not on the bars of one contract. The risk distributions the software does provide are built from your own trades: the Gauntlet reports the median, 5th-percentile and 1st-percentile maximum drawdown over 10,000 reshuffles of the trade order, and the prop firm Drawdown zones show how often each level of a drawdown allowance is reached. A historical VaR of daily results can be computed from exported trades by grouping them by session day.
Related
In the same section
- Value migration
- Value area low
- Vanna
- Value area high
- Vanna and charm flows
- Value area
- Variance swap
- Upthrust
This page in other languages
Frequently asked questions
- Is a 95% VaR the most I can lose?
- No. It is a loss exceeded on about 5% of the periods under the model. On those periods the loss can be much larger than the VaR, and heavy-tailed results make such days more common than a normal model predicts.
- Which is better for a trader, VaR or maximum drawdown?
- They answer different questions. VaR describes a single period's loss at a probability; maximum drawdown describes the worst cumulative decline along a path. Account rules such as drawdown floors and daily loss limits map more directly onto drawdown and daily loss statistics.