# Calmar ratio

> The Calmar ratio sets your annualised return against your worst peak-to-trough fall. It asks what the return was worth relative to the deepest hole you had to sit through.

> These figures are derived from a return series that the app currently flags as being reworked.
> Values and holdings are unaffected. See [about these numbers](/metrics/about-these-numbers).

The Calmar ratio divides your annualised return by your worst fall. Where
[Sharpe](/metrics/sharpe-ratio) and [Sortino](/metrics/sortino-ratio) measure
return against the average spread of daily moves, Calmar measures it against
the single worst episode.

It is the closest of the ratios to how people actually judge an investment
after the fact: was the return worth the drop I had to sit through?

## How it is calculated

```
Calmar ratio = annualised return / |maximum drawdown|
```

**Annualised return** is the compound growth rate implied by the first and last
values in your window, scaled to a year.

**Maximum drawdown** is the largest peak-to-trough fall in that window, taken
as a positive number for the division. See
[maximum drawdown](/metrics/max-drawdown) for how it is found.

A portfolio that returned 12% a year and fell 24% at its worst has a Calmar
ratio of 0.5.

## How to read it

Higher means more annual return for each unit of worst-case pain.

Two properties are worth holding in mind, because they make Calmar behave
differently from the other ratios.

**It is driven by a single event.** The denominator is one episode, not a
distribution. One unusually bad month can halve the ratio for years afterwards,
even if nothing else about the portfolio changed.

**It is unstable when the drawdown is small.** A portfolio that has barely
fallen produces a tiny denominator and therefore a very large ratio. A figure
that looks spectacular usually means the window contained no real stress,
rather than that the portfolio handles stress well.

A portfolio that never fell at all has no drawdown to divide by, and Gylder
shows zero rather than an infinity.

As with the other ratios, Gylder does not set a threshold for what counts as
good.

## Window sensitivity

Calmar is more window sensitive than any other measure in the grid, because
both halves of it change with the date range and they change for different
reasons. Extending the window gives the drawdown more chances to deepen while
smoothing the annualised return. Comparing a one year Calmar against a five
year one tells you very little.