Kurtosis Calculator – Measure How Heavy-Tailed Your Data Is
The Kurtosis Calculatormeasures the "tailedness" of a dataset's distribution — how much of its variance comes from infrequent, extreme deviations compared to a normal distribution. Statisticians, finance professionals analyzing tail risk in asset returns, quality-control engineers, and students studying descriptive statistics use kurtosis to quickly understand whether a dataset is prone to outliers before applying tests that assume normality.
What Kurtosis Tells You
Kurtosis is a single number that describes the shape of a distribution's tails and peak relative to a normal (bell-curve) distribution:
- Leptokurtic (positive excess kurtosis): heavier tails and a sharper peak than normal, meaning a higher probability of extreme outliers.
- Mesokurtic (excess kurtosis near zero): tail behavior similar to a normal distribution.
- Platykurtic (negative excess kurtosis): lighter tails and a flatter peak than normal, meaning fewer extreme outliers.
Four Calculation Methods
There is more than one accepted formula for kurtosis, and this calculator supports the four most common ones:
| Method | Formula | When to Use |
|---|---|---|
| Population | m4 / σ⁴ | Your data represents the entire population (reads 3 for normal) |
| Sample | m4 / s⁴ | Your data is a sample drawn from a larger population (reads 3 for normal) |
| Excess | (m4 / s⁴) − 3 | Most common choice; a normal distribution reads 0 |
| Adjusted (Bias-Corrected) | [n(n+1) / ((n−1)(n−2)(n−3))] × [Σ(x−x̄)⁴ / s⁴] − [3(n−1)² / ((n−2)(n−3))] | Small samples; matches Excel's KURT() |
How the Calculation Works
- Compute the mean of the dataset:
x̄ = Σx / n - Find each value's deviation from the mean, then square and raise it to the fourth power:
(x − x̄)²and(x − x̄)⁴ - Average the fourth-power deviations to get the fourth central moment:
m4 = Σ(x − x̄)⁴ / n - Divide by the square of the variance (population
σor samples) depending on the selected method - For excess kurtosis, subtract 3 so a normal distribution reads 0; for adjusted excess kurtosis, apply the small-sample correction factor instead
Shape Classification
Alongside the numeric coefficient, the calculator classifies the result (converted to an excess-kurtosis basis) into a plain-language interpretation band:
> 1: Strongly leptokurtic0 to 1: Mildly leptokurtic≈ 0: Mesokurtic (approximately normal)−1 to 0: Mildly platykurtic< −1: Strongly platykurtic
Raw Values vs. Grouped/Frequency Data
Use raw values mode when you have access to every individual observation — this gives the most accurate result. Use grouped/frequency mode when your data has many repeated values, such as histogram bins or survey responses, and you want to enter each unique value once alongside its count instead of typing every repetition.
Practical Applications
Kurtosis is widely used in financial risk analysis to measure tail risk in asset return distributions — fat-tailed (leptokurtic) returns imply a higher probability of extreme gains or losses. It is also used to check whether a dataset deviates from normality before applying parametric statistical tests, for quality-control process monitoring, and in exploratory data analysis alongside mean, variance, standard deviation, and skewness.
Tips for Best Results
- Use excess kurtosis for most real-world analysis — it is the most widely reported form and reads 0 for a normal distribution.
- Use adjusted (bias-corrected) excess kurtosiswhen working with small samples or when you need results that match Excel's
KURT()function exactly. - Sample and excess methods require at least 3 data points, adjusted excess kurtosis requires at least 4, and population kurtosis requires at least 2.
- Toggle the step-by-step breakdown to see the mean, fourth moment, and per-value deviations that produced the final coefficient.
- Pair this tool with the Skewness Calculatorfor a complete picture of a distribution's shape — skewness describes asymmetry, kurtosis describes tail weight.