> Sharpe ratio from your own per-period returns — annualised under a stated convention, with the confidence interval every other calculator omits.

- Canonical: https://hadalinstruments.com/tools/sharpe-ratio-calculator/

---

Tool

# Sharpe ratio calculator

A mean over a standard deviation — the division is the easy half. A Sharpe ratio is an **estimate**, computed from however many periods you actually have, and it carries error bars that shrink only with more periods. This computes the ratio from your own returns, annualises it under a convention it states rather than assumes, and prints the interval your sample supports — the part every ranking calculator omits, and the part that decides whether the number means anything.

Annualised Sharpe —

Annualised Sharpe ratio —

95% interval your sample supports —

Per-period Sharpe —

Mean and standard deviation per period —

Periods counted —

## Why the interval is the answer

Two records can both print a Sharpe of 1.2 and mean entirely different things: from three years of monthly returns the estimate has begun to settle, from eleven months it is consistent with brilliance and with nothing. The standard error shrinks with the square root of the sample, so halving the uncertainty costs four times the history — there is no shortcut through it, and a Sharpe quoted without its sample size has told you a numerator and hidden the part that decides its meaning. This is the same [effective sample size](https://hadalinstruments.com/glossary/effective-sample-size/) discipline the glossary applies to win rates, pointed at the industry's favourite statistic.

## The stated conventions, and the stated caveat

Annualisation multiplies by the square root of periods per year — the panel names the multiplier it used, because monthly-times-root-twelve and daily-times-root-252 are elections, not laws, and a Sharpe compared across pages that chose differently is a comparison of conventions. The interval assumes returns are independent draws (the Lo 2002 iid standard error): real return series are autocorrelated, and autocorrelation makes this interval too *narrow*, never too wide — so read it as the optimistic bound on your uncertainty, not the pessimistic one.

## What this does not do

It does not adjust for selection. A Sharpe computed on the best of many tried variants is not an estimate of anything — the more configurations were tried, the higher the best one's Sharpe by luck alone, and the correction for that lives at [deflated Sharpe ratio](https://hadalinstruments.com/glossary/deflated-sharpe-ratio/) and in the [Overfit Auditor](https://hadalinstruments.com/instruments/overfit-auditor/). It does not annualise the interval's assumptions away, and it does not accept a mean and a standard deviation typed directly — the series is required, because the sample size is not an optional garnish on this number. It is the number.

The term itself is defined at [Sharpe ratio](https://hadalinstruments.com/glossary/sharpe-ratio/), and the drawdown the ratio's denominator smooths over is measured by the [maximum drawdown calculator](https://hadalinstruments.com/tools/maximum-drawdown-calculator/) on the same paste-a-series pattern. Whether a win rate can carry a sizing is the [risk of ruin calculator](https://hadalinstruments.com/tools/risk-of-ruin-calculator/), and the full list is under [calculators](https://hadalinstruments.com/tools/). And when the inputs cannot support an answer, the panel above refuses and says why rather than rendering a zero — [why an instrument refuses to answer](https://hadalinstruments.com/docs/insufficient-n/) is that design, written down.