Coefficient of Variation Calculator
Reviewed by CalcMulti Editorial Team·Last updated: ·← Statistics Hub
The coefficient of variation (CV) expresses standard deviation as a percentage of the mean, giving a unit-free measure of relative variability. Unlike standard deviation, CV lets you compare the spread of datasets with different units or different scales — for example, comparing the variability of stock prices versus bond yields.
This calculator computes population CV (using σ) and sample CV (using s), along with mean, standard deviation, and an interpretation of the result.
Formula
CV = (σ / μ) × 100% or CV = (s / x̄) × 100%
- σ
- population standard deviation
- μ
- population mean
- s
- sample standard deviation (divides by n−1)
- x̄
- sample mean
- CV
- coefficient of variation — expressed as a percentage
CV Benchmarks by Field
| Field | Acceptable CV Range | Notes |
|---|---|---|
| Analytical chemistry (lab assays) | < 2% | High precision required; > 10% warrants investigation |
| Clinical / biomedical | < 15% | Biological variation adds inherent noise |
| Finance — stock returns | 20–100%+ | High CV = high volatility; lower is more stable |
| Manufacturing / quality control | < 5% | Tighter tolerances → lower CV expected |
| Agricultural research | 10–30% | Environmental variability accepted |
| Social science surveys | 20–50% | Human behaviour is inherently variable |
| Sports performance | 5–20% | Depends on sport and metric measured |
CV vs Std Dev vs IQR — When CV Is the Right Choice
| Situation | Std Dev | CV | IQR |
|---|---|---|---|
| Comparing groups with different means | — | ✓ Normalises by mean | — |
| Comparing datasets in different units | — | ✓ Dimensionless | — |
| Investment / risk-per-return comparison | — | ✓ | — |
| Mean is near zero (CV undefined) | ✓ | Not valid | ✓ |
| Data has outliers / is skewed | — | — | ✓ |
| Same scale, same magnitude mean | ✓ Preferred | OK | OK |
Common Mistakes with CV
Using CV when the mean is near zero
CV = SD / mean. If the mean is close to zero, CV explodes toward infinity — or becomes negative with a negative mean. CV is only meaningful when the mean is clearly positive and non-trivial. Use SD or IQR instead for zero-anchored data.
Comparing CV across fundamentally different distributions
A CV of 40% for annual rainfall and 40% for stock returns are both "40%" but represent very different realities. CV comparisons are most meaningful within the same field or same type of measurement. Always contextualise against domain benchmarks.
Reporting CV without SD
CV alone doesn't tell you whether the dataset is precise (small SD, small mean) or highly variable (large SD, large mean) in absolute terms. Always report CV alongside the mean and SD so the reader can reconstruct the actual spread.
Case Study: Comparing Fund Volatility — Why CV Beats Absolute Std Dev
An investment analyst was comparing two equity funds for a client. Fund A had a mean annual return of 8% with SD = 4%. Fund B returned a mean of 15% per year with SD = 9%. On absolute volatility alone, Fund A looked safer (4% < 9%).
But computing CV: Fund A CV = 4/8 = 50%. Fund B CV = 9/15 = 60%. Fund B delivered significantly higher returns, and its risk per unit of return was only marginally higher than Fund A's — not the dramatic difference the raw SDs suggested.
The client chose Fund B. The CV comparison made it clear that Fund A's lower absolute volatility was largely a function of its lower return level — not that it was inherently more stable or better managed. Comparing standard deviations without normalising by the mean had been giving a misleading picture of relative risk.
Disclaimer
CV is only meaningful when the mean is positive and the data is on a ratio scale. Results should be interpreted in the context of your specific field and dataset.