Math calculator

Geometric Mean Calculator

The average for anything that multiplies rather than adds.

The average that compounds

Rates or plain values.

Separated by commas, spaces or new lines. Enter −10 for a 10% fall.

2 periods

−13.3975% per period

The arithmetic mean of the same rates is 0.0000%, which nothing actually earned.

Geometric mean

0.866025

as a multiplier

Arithmetic mean

1

always at least as large

Total change

0.75

over the whole period

Check

reproduces

applied 2 times

The check, carried out

0.866025^2 = 0.75, and the actual total change was 0.75. They match, which is the whole point of this average.

What the arithmetic mean would have claimed

The arithmetic mean of these rates is 0%. Applied 2 times that would give 1× — against the true 0.75×. It overstates, always, and the gap grows with how varied the rates are.

How it is computed

The n-th root of a product is computed through logarithms — sum ln of each value, divide by 2, exponentiate — so a long series cannot overflow before the root is taken.

  • The geometric mean is the number that, multiplied by itself 2 times, gives the same total as the whole series. That is what makes it the right average for anything that compounds.
  • The arithmetic mean is always at least as large, and equal only when every rate is identical. So averaging returns arithmetically always flatters them — which is why fund literature quotes the geometric one, usually calling it the compound annual growth rate.
  • +50% then −50% is the standard demonstration: the arithmetic mean says 0% and you are actually down 25%, which is −13.4% a year compounded. Nothing about the arithmetic mean is wrong; it is simply answering a different question.
  • The geometric mean is never larger than the arithmetic mean, for any positive data. That inequality is one of the oldest in mathematics and the gap between them is a measure of how spread out the values are.

The answer applied over the whole period reproduces the actual total change.

What this tool shows

+50% then −50% averages to 0% arithmetically and leaves you down 25%. The truth is−13.4% a year, and only the geometric mean finds it — because only it respects that the second year multiplies the first.

  • The geometric mean of values or growth rates
  • The answer as a compound rate per period
  • The arithmetic mean, and what it would have claimed
  • The check: applied n times, it reproduces the total
  • Why a −100% period has no meaningful average after it
  • Why the answer is never above the arithmetic mean
Rates or plain values The check performed, not described Arithmetic mean shown beside Computed in logs, so no overflow

Exact; the compounding check is carried out.

Updated 7 September 2026 · Works in any browser, no installation

Multiply the values together and take the nth root. For growth rates, turn each into a multiplier first: +50% is 1.5, −50% is 0.5. Their product is 0.75, and the square root of that is 0.866 — a compound rate of −13.4% a year. Averaging the rates arithmetically gives 0%, which is a number nothing actually earned.

At a glance

Formula shown
GM = (x\u2081 \u00d7 x\u2082 \u00d7 \u2026 \u00d7 x\u2099)^(1/n), computed through logarithms so a long series cannot overflow. For growth rates, convert each to a multiplier (1 + r/100) first and convert the answer back.
Scenario support
Averaging annual returns; averaging population or price growth; averaging ratios and index numbers where doubling and halving should cancel.
Educational estimate
Planning support from the values you enter — not professional advice.

The +50% then −50% case

Put 100 in. It grows 50% to 150. It falls 50% to 75. You are down a quarter.

The arithmetic mean of +50 and −50 is 0%, which would imply you finished where you started. It is a correct average of the two NUMBERS and a wrong answer about the money, because the second year multiplies the first rather than adding to it.

The geometric mean of 1.5 and 0.5 is √0.75 = 0.866, or −13.4% a year. Apply that twice: 100 × 0.866 × 0.866 = 75. It reproduces the actual result, which is the entire test of whether an average of rates is the right one.

Why the arithmetic mean flatters

The arithmetic mean of a set of growth multipliers is never smaller than the geometric mean, and is equal only when every value is identical. So averaging returns arithmetically ALWAYS overstates them, never understates.

The gap grows with volatility. Two steady 10% years have no gap at all; a +50 and a −50 have thirteen points of it. That is why a volatile fund can advertise a good arithmetic average while its investors are down.

It is also why regulators require compound figures in fund literature. The arithmetic mean is not a lie about the arithmetic; it is an answer to a question nobody asked.

CAGR is this, renamed

Compound annual growth rate is the geometric mean of the annual multipliers, minus one. Same calculation, different name, and finance uses the name.

It answers: what single steady rate would have produced this end result from this start? By construction it reproduces the total, which is what makes it comparable between investments.

It also hides the path. Two investments with the same CAGR can have had entirely different years, and one of them may have been unholdable. CAGR is the right summary and it is only a summary.

When to use it on plain values

Outside growth rates, the geometric mean is the right average whenever the quantities are RATIOS or multiply together.

Aspect ratios, price indices, dilution factors, and anything where doubling and halving ought to cancel. The geometric mean of 2 and 8 is 4, which sits proportionally between them — twice the first and half the second. The arithmetic mean, 5, does not.

A useful test: if multiplying every value by 10 should multiply the answer by 10, and halving some values should cancel doubling others, you want the geometric mean.

Why it needs positive numbers

A zero makes the product zero, and the nth root of zero is zero — whatever the other values were. That is technically correct and not an average of anything.

Negatives are worse: an even root of a negative product has no real value at all, so the answer would exist or not depending on how many values you happened to have.

In the rate reading it makes sense. A −100% period leaves nothing, and no subsequent growth rate can recover from nothing, so there is no meaningful average afterwards. This page says that rather than returning a number.

The inequality, and what the gap measures

For any positive data: harmonic ≤ geometric ≤ arithmetic ≤ root mean square. Equality happens only when every value is identical.

So the gap between two of these means is itself a measure of spread. Two averages that agree tell you the data barely varies; a wide gap tells you it varies a lot, before you compute any variance.

The arithmetic-geometric inequality is one of the oldest results in mathematics and has a proof short enough to fit in a paragraph. This site checks it computationally on every generated series rather than quoting it.

Sources and methodology

The geometric mean is standard; these set out where it is required rather than merely preferred.

Method. Values are converted to multipliers where rates were given, then the mean is computed as exp(mean of the logs) so a long series cannot overflow before the root is taken. The answer is then raised to the power n and compared with the actual product; the page reports that comparison rather than assuming it. The arithmetic mean is computed alongside and its implied total is shown for contrast. That engine is verified on every change against 61 hand-written assertions, including that the geometric mean raised to n reproduces the product across two thousand generated series, and that it never exceeds the arithmetic mean. The count and the per-case breakdown are published on the formula verification page.

Related calculators

Where this goes next:

Harmonic MeanSixty there and forty back averages 48, not 50. The page simulates the journey from its legs as well as computing the formula, so the answer arrives with its own proof.
Weighted AverageEach value carries the weight you give it, with every item's share of the total shown as a percentage so you can see what is actually driving the answer.
Root Mean SquareFor values whose sign must not cancel. Mains voltage averages zero and still boils a kettle; squaring, averaging and rooting gives the number that describes the work.
Doubling TimeHow long a quantity takes to double at a constant rate, or from two measurements — with the rule of 70 and rule of 72 measured against the exact answer.
Compound InterestSee how savings grow as interest earns interest, with adjustable contributions and compounding frequency.
ROISimple, date-based, and net ROI with annualised ROI (CAGR), a reverse target solver, and a two-investment comparison.

More in Math, or browse all calculators.

Read the guide

For rates over a fixed distance or quantity — speeds, prices per unit — the Harmonic Mean Calculator is the right one. For the forward direction, turning a rate into a final amount, the Compound Interest Calculator does that.

Educational use disclaimer

This is an educational tool and not investment advice. The arithmetic is exact; a past average growth rate says nothing about a future one.

How we calculate · Found an error? email us

Authorship & verification

Written and maintained by , a business operator who builds spreadsheet-based calculators.

What's changed (3 updates)

Published 7 September 2026

  1. Published the geometric mean page built around the +50% then -50% case, because the arithmetic mean says you broke even and the money says otherwise.
  2. Growth rates may be entered as percentages or as multipliers; the page converts and shows the compounded total so the mean can be checked against it.
  3. A zero or a negative value is refused with the reason rather than silently dropped, since the product would be meaningless either way.

Add this calculator to your site

Responsive embed — and private: nothing your visitors type leaves their browser.