What the geometric mean is, and when to reach for it
The geometric mean is the right average when the values multiply together rather than add together. Growth rates compound. Investment returns compound. Population growth, bacterial growth, compound interest — all of it stacks multiplicatively, not additively. And the arithmetic mean (the regular "add them up and divide" average you learned in school) systematically overstates the true compound rate when you apply it to numbers like these.
The geometric mean fixes that. Multiply all the values together, then take the nth root, where n is how many values you have. The result is the rate that — applied repeatedly to the starting value — produces the same final value as the actual sequence did.
If you're computing an average percent change, an average growth rate, an average return on investment, an average ratio of any kind, this is the average you want. Using the arithmetic mean for these will quietly give you a wrong answer that's hard to spot until someone checks the math.
How to use the Geometric Mean Calculator
The calculator takes a list of positive numbers and returns the geometric mean.
- Paste or type your numbers, separated by commas, spaces, or new lines. For return rates expressed as percentages, convert each to a multiplier first: 10% becomes 1.10, −5% becomes 0.95, 100% becomes 2.00. (Some versions of the calculator do this conversion for you — check the input field.)
- Read the geometric mean. It's a single number. If your inputs were multipliers, the result is the effective compound multiplier per period. Subtract 1 and multiply by 100 to convert back to a percentage.
The calculator only accepts positive numbers. The geometric mean of any list that contains zero is zero (because the product is zero), and the geometric mean of any list with a negative value is undefined for real numbers (because you'd be taking an even root of a negative). That's a mathematical fact, not a calculator limitation.
Everything stays in your browser. Nothing is logged or uploaded.
The formula
For a list of n positive numbers x₁, x₂, ..., xₙ:
Geometric mean = ⁿ√(x₁ · x₂ · ... · xₙ)
In words: multiply all the values together, then take the nth root of the product. For two numbers, this is the square root of their product. For three numbers, the cube root. For 12 monthly return rates, the twelfth root of their product.
For larger datasets, the formula is often computed in log-space to avoid overflow: take the average of the logs of the values, then exponentiate. Mathematically it's the same answer, but computationally it's more stable. The calculator does this internally — you don't need to worry about it.
The investment example that shows why this matters
Say an investment grows 10% in year one, 20% in year two, and loses 5% in year three. What was the average annual return?
The tempting answer is the arithmetic mean: (10 + 20 + (−5)) / 3 = 25/3 ≈ 8.33%. That's the answer most people would give. It's also wrong.
Let's check it. If you start with $1,000 and earn 8.33% per year for three years, you finish with $1,000 × (1.0833)³ = $1,271.21.
Now let's compute what actually happened. Year 1: $1,000 × 1.10 = $1,100. Year 2: $1,100 × 1.20 = $1,320. Year 3: $1,320 × 0.95 = $1,254.
The actual final value is $1,254, not $1,271. The arithmetic mean overstated the compound rate by enough to inflate the three-year result by $17. Over decades, that gap grows fast.
The geometric mean gives the right answer:
- Multipliers: 1.10, 1.20, 0.95
- Product: 1.10 × 1.20 × 0.95 = 1.254
- Cube root: ³√1.254 ≈ 1.0789
- Convert to percent: 1.0789 − 1 = 0.0789 = 7.89%
The effective compound annual growth rate (CAGR) was 7.89%, not 8.33%. Check it: $1,000 × (1.0789)³ = $1,254 — matches the actual final value exactly.
The general rule: the geometric mean is always less than or equal to the arithmetic mean, with equality only when all values are identical. The more the values vary, the bigger the gap. This is the AM-GM inequality, one of the cleanest results in elementary mathematics.
Where you'll actually use this
Anywhere percentages or ratios compound over time:
| Use case | What you're averaging | Why arithmetic mean is wrong |
|---|---|---|
| Investment returns (CAGR) | Annual return rates | Returns compound; arithmetic mean overstates final wealth |
| Mutual fund performance reports | Year-by-year returns | Industry-standard CAGR is geometric, not arithmetic |
| Population growth | Year-over-year growth rates | Population compounds; arithmetic mean inflates projections |
| Inflation-adjusted comparisons | Price ratios over years | Price changes compound |
| Index numbers (CPI, stock indices) | Index ratios across periods | Ratios chain multiplicatively |
| Bacterial / cell culture growth | Multiplicative growth factor per hour | Cell counts compound |
| Audio (decibels) averages | Sound intensity ratios | Decibels are logarithmic; geometric mean is the right average |
| Image processing (aspect ratios) | Width and height ratios | Multiplicative scaling |
If you're averaging anything that "doubles" or "halves" or grows "by X%" over time, you're in geometric-mean territory. If you're averaging things that just add up (heights, weights, test scores, distances), stick with the arithmetic mean.
The intuition: why the nth root
The geometric mean is the value that — repeated n times in a multiplication — produces the same total as the actual sequence. That's literally what the nth root of the product means.
Take the investment example again. Three years of returns produced a final multiplier of 1.254 starting from $1. The geometric mean is the constant rate that, applied three times, gets you from $1 to $1.254. That number is ³√1.254 ≈ 1.0789. Three times 7.89% growth compounded gives you the same end result as 10%, 20%, −5% did. That's the definition.
The arithmetic mean answers a different question: what's the average of the rates, ignoring how they interact? For a list of compounding rates, that's rarely what you want, because the rates don't actually exist in isolation — they multiply on top of each other.
The geometric mean and percentages: a gotcha
One common mistake: people compute the geometric mean of percentage values directly. So if returns were 10%, 20%, −5%, they try to compute ³√(10 × 20 × −5). This is wrong for two reasons. First, you can't take a real cube root of a negative number through the standard formula. Second, even ignoring the sign, the math is meaningless — multiplying percent values together doesn't represent anything physical.
The right approach is to convert each percent to its multiplier first:
- +10% → 1.10 (the value grew to 110% of what it was, a multiplier of 1.10)
- +20% → 1.20
- −5% → 0.95
Then take the geometric mean of the multipliers (in this case, 1.0789), and convert back to a percent (7.89%). This is one of those small but consequential conventions that trip up undergraduates and finance professionals alike.
Geometric mean and the AM-GM inequality
For any list of positive numbers, the arithmetic mean is always greater than or equal to the geometric mean, with equality if and only if all the values are identical. Mathematicians call this AM ≥ GM, or the arithmetic-geometric mean inequality.
The intuition: the arithmetic mean treats every value the same way regardless of size. The geometric mean is dragged down by small values more aggressively than it's pushed up by big ones (because multiplying by a small number does more "damage" than multiplying by a big number does "good"). When values are spread out, the geometric mean drops below the arithmetic mean. When values are all identical, the two are equal.
This is why mutual fund prospectuses are required by the SEC to report compound annual growth rate (CAGR, a geometric mean) rather than just the average of yearly returns (the arithmetic mean). Reporting the arithmetic mean would systematically overstate fund performance, especially for funds with volatile returns. The geometric mean is the honest number.
Limitations
Three cases where the geometric mean doesn't apply, or where you have to be careful:
- Zero values. Any list with a zero has a geometric mean of zero, because the product is zero. If one period had a 100% loss (multiplier of zero), the geometric mean correctly reports "the average compound rate was nothing, the investment was wiped out" — which is mathematically right, but rarely the question you wanted to answer.
- Negative values. The geometric mean is undefined for negative numbers (you can't take an even root of a negative real number). For returns expressed as multipliers (where 1.0 = no change), this isn't usually a problem — a multiplier of −0.05 doesn't make sense anyway. But if you accidentally tried to compute the geometric mean of raw percent values that include negatives, the formula breaks.
- Not for additive quantities. Don't use the geometric mean for heights, weights, ages, test scores, distances — quantities that combine additively. The geometric mean of these doesn't have a useful physical interpretation. Use the arithmetic mean for additive data.
Related tools
The geometric mean is one of several "means" that show up in statistics and finance. Companions you might want:
- Average Calculator — the arithmetic mean (plus median, mode, range) for additive data.
- Percent Error Calculator — for checking measurements against a reference value.
- Exponent Calculator — for the underlying compound-interest math, when you want to see how a geometric mean rate compounds out over many periods.
- Root Calculator — the inverse of an exponent; the nth root is what the geometric mean formula uses.
Frequently asked questions
What is the geometric mean?
The geometric mean of n positive numbers is the nth root of their product. It's the value that, multiplied by itself n times, equals the product of the original list. For two numbers, it's the square root of their product. For three, the cube root. It's the right average for growth rates, return rates, ratios, and anything else that compounds multiplicatively.
When should I use the geometric mean instead of the arithmetic mean?
Whenever the values you're averaging combine multiplicatively rather than additively. Investment returns, growth rates, population changes, inflation rates, decibel sound levels, index numbers — all geometric. Heights, weights, test scores, monthly expenses — arithmetic. If using the average to project forward by multiplication, use the geometric mean. If using it to estimate a total by addition, use the arithmetic mean.
Why does the arithmetic mean overstate compound growth?
Because compounding amplifies losses relative to gains in a way that the arithmetic mean doesn't capture. A 50% loss followed by a 50% gain doesn't return you to where you started — you end up at 75% of the original value. The arithmetic mean of −50% and +50% is 0%, suggesting no change. The geometric mean correctly reports a loss. This effect compounds (literally) the more volatile the returns are.
Can the geometric mean be negative?
No, not for real-number inputs. The geometric mean of positive numbers is positive. For lists with negative values, the geometric mean is undefined (you'd need to take an even root of a negative number, which has no real result). Convert percent values to multipliers first (1.10 instead of +10%, 0.95 instead of −5%) — multipliers are always positive for normal cases.
What's the geometric mean of two numbers?
The square root of their product. Geometric mean of 4 and 9 is √(4 × 9) = √36 = 6. Note that 6 sits between 4 and 9, but it's not their arithmetic mean (which would be 6.5). For two numbers, the geometric mean is always less than or equal to the arithmetic mean — they're equal only when the two numbers are identical.
How do I compute compound annual growth rate (CAGR)?
CAGR is the geometric mean of year-over-year growth multipliers, minus 1, expressed as a percent. For an investment that grew from $1,000 to $1,500 over 5 years, CAGR = (1500/1000)^(1/5) − 1 = 0.0845 = 8.45% per year. You can also compute it as the geometric mean of each year's individual return multipliers if you have them broken out — same answer.
What if one of my values is zero?
The geometric mean becomes zero, because the product of the list is zero. For investment returns, a multiplier of zero means a 100% loss in that period — the investment was wiped out. The geometric mean correctly reports that, on average, you ended up at zero. If you have a zero you weren't expecting (a data-entry error, a missing value), check your inputs before trusting the result.
Is the geometric mean used anywhere outside of finance?
Yes — in audio engineering (averaging decibel levels), biology (averaging cell counts that grow multiplicatively), demographics (population growth rates), image processing (averaging aspect ratios), epidemiology (averaging reproduction numbers like R₀), and chemistry (averaging dilution ratios). Anywhere ratios or growth rates appear, the geometric mean tends to be the appropriate average.