Value at Risk (95%) — Return

5th percentile of daily returns on the Return curve (low-to-low). ~1 in 20 days the loss will exceed this value.

Computed from: Equity curve
Scope: Single report
Range: ≤ 0%
Direction: Higher is better

Basis

Computed on the low-to-low Return Series (money-weighted) — measured day-worst-point to day-worst-point, not close-to-close.

Value at Risk (95%) is the daily loss you'd exceed only 1 day in 20. In plain words: on 19 days out of 20, you lose no more than this. (Technically it's the 5th percentile of your daily results — the loss level only your worst 5% of days drop below.) It's a negative number, so a VaR of −2% reads as "1 day in 20 is worse than a 2% loss."

Key takeaways

What it is — the 5th percentile of your daily returns: the loss level that only your worst 5% of days drop below. A negative percentage.
Read it as — on 95% of days you lose no more than this. Closer to zero (less negative) is better — a smaller routine worst-case. Higher is better.
Why it matters — it puts a single number on your routine bad day, the one that shows up about once every 20 trading days. It's a threshold, not a forecast of the worst possible day.
The headline caveat — VaR says nothing about how bad the 5% beyond it gets. A −2% VaR means 1 day in 20 is worse than −2% — but the worst of those could be −2.1% or −40%. The number that does measure how deep the bad tail runs is Conditional VaR (also called expected shortfall — the average of the days past the VaR line), its natural companion.
In EquityTruth — the historical 5th percentile of each day's return (counted from the real days that happened — no tidy bell-curve assumed), on the Return, P&L, and TWR curves.

How it's calculated

Take all your daily returns, line them up worst to best, and read off the value at the 5% mark. Five percent of your days are worse than it; ninety-five percent are better. That cutoff is your VaR — the same 5th-percentile (p5) figure the tail ratio uses for its left tail.

VaR (95%) = 5th percentile of daily returns

5th percentile: the return only your worst 5% of days fall below — a negative number
95%: the cutoff: 5% of days breach the threshold (1 day in 20), 95% do not

In this product: EquityTruth uses historical VaR — the actual empirical percentile of observed daily returns (np.percentile(r_min, 5)), not a parametric mean − 1.65σ formula. That makes no normality assumption: a parametric VaR assumes a bell curve and badly understates fat-tailed risk, while the empirical percentile reads the real distribution. Computed on the daily low-to-low returns — one value per trading day, each day's lowest equity to the next day's lowest — the same p5 the [tail ratio](/glossary/tail-ratio) reads. Reported as a (negative) percentage on the Return, P&L, and TWR curves.

On the Return and TWR curves VaR is the 5th percentile of percentage daily returns. Return is money-weighted (deposits and withdrawals move it); TWR strips cashflows out for the cleanest read of the strategy's own routine worst-day.

What it tells you

These are losses, so a smaller loss — a number closer to 0 — is better: −1% beats −6%. The bands are calibrated for leveraged retail forex daily VaR — on a 1:100 account a −3% day is ordinary, so a figure that looks alarming on an equity-fund scale can be routine here. Read it relative to leverage.

Value	Reading	Notes
> −1%	Very contained	Routine bad days stay shallow — but check it isn’t simply barely trading.
−1% to −3%	Moderate	A typical band for an active forex account.
−3% to −6%	Large daily risk	Routine bad days bite hard — read with leverage and max drawdown.
−6% to −12%	Extreme	Aggressive sizing — one bad day in twenty is a serious dent. Pair with CVaR and max drawdown.
< −12%	Blow-up profile	Typical of grid / martingale just before failure. The tail past this line (CVaR) is where accounts die.

VaR is the line, not the shaded tail to its left. Two accounts can share this exact VaR while their worst-5% days run anywhere from −2.1% to −40% — the part past the line is what Conditional VaR measures and VaR cannot see.

Worked example

An account's daily returns have a 5th percentile of −2%. That's its VaR (95%): about 1 trading day in 20 is worse than a 2% loss, and on the other 19 it loses less than that — or gains. But −2% isn't your worst day — it's the doorway. Beyond it, 1 day in 20, things get worse, maybe much worse: those breach days could average −2.5% or −12%, and VaR alone won't tell you which.

Pitfalls

Pitfalls & caveats

It says nothing about the 5% beyond it. This is the headline limitation. VaR is the threshold where the bad tail begins, not how deep it runs. A −2% VaR is silent about whether the breach days average −2.1% or −40%. The metric that measures the severity past the line is Conditional VaR — always read them together.
It breeds false confidence. A reassuring VaR is reassuring about ordinary days only — and it goes quiet exactly when markets break. It's been called "the airbag that works except when you crash." A clean VaR number has lulled more blow-ups than it has prevented; treat it as one input, never as your risk ceiling.
A threshold, not a guarantee. VaR is not a maximum. You will have roughly 1 day in 20 worse than it — that's the definition, not a failure. It bounds the routine bad day, never the worst one.
Historical VaR only sees the bad days that already happened — the calm is the danger. The percentile is read off observed returns, so a young, quiet, or winning-streak record shows a tame VaR not because it's safe but because its crisis day hasn't printed yet. A six-month winning run and a strategy that simply hasn't met its shock look identical to historical VaR. Needs multiple years of daily data before the 5th percentile is stable.
Daily, not per-trade. Measured day-to-day on the equity curve (overnight gaps and intraday path included), not per individual trade.
The 95% cutoff is a choice. 5% = 1 day in 20. A 99% VaR (1 day in 100) would sit deeper in the tail and read more negative — so compare like with like, always the same cutoff.

VaR vs CVaR

VaR and Conditional VaR are two readings of the same tail. VaR is the threshold — the loss at the 5% cutoff, where the bad days begin. CVaR is the average loss beyond it — how bad the bad 5% actually are. CVaR is precisely what VaR can't see: take all the days worse than the VaR line and average them. A −2% VaR with a −2.4% CVaR is a tame, well-behaved tail; the same −2% VaR with a −9% CVaR hides a fat-tailed bomb. Always read them as a pair.

This isn't just our preference. After 2008, the Basel market-risk framework replaced 99% VaR with 97.5% Expected Shortfall (CVaR) as the bank capital standard — for exactly the reason above: VaR could not see the tail it was supposed to guard against. The regulators fired VaR and promoted CVaR.

Conditional VaR measures the severity past the VaR line, Tail Ratio reads the same 5th percentile against the 95th, Volatility is the average swing VaR's tail sits inside, and Max drawdown is the worst peak-to-valley fall the far tail can build into.

Compare withConditional VaR (95%)