Percentile Calculator

Background.

The Quanta percentile calculator answers the question 'where does this value sit in my actual data?' — not in a theoretical bell curve, but in the real numbers you paste in. Give it a percentile and it finds the value that sits there; give it a value and it finds the percentile rank of that value within your dataset. Both directions use the same underlying idea: sort your data from smallest to largest, and a percentile is simply a position in that sorted list, expressed as a percentage of the way through it. The 90th percentile of a set of exam scores is the score below which 90% of the class falls; if your own score is the 90th percentile, you outperformed 90% of your classmates.

This calculator is deliberately built on your actual sample data rather than on an assumed distribution. That distinction matters and is easy to miss. Quanta's z-score and standard-deviation calculators are parametric: you tell them a population mean and standard deviation, and they compute a percentile from the theoretical normal curve using the Abramowitz-Stegun approximation to the normal cumulative distribution function. That is exactly the right tool when you know (or are willing to assume) that the underlying population is normally distributed — IQ scores, manufacturing tolerances, and many biological measurements are close enough to normal for this to work well. But it is the wrong tool whenever your data is skewed, bounded, multimodal, or you simply don't want to assume a shape at all — a set of exam scores from one particular class, a distribution of house prices, or response times in a user study rarely look like a textbook bell curve, and forcing them through a normal-distribution formula can give a misleading answer. This calculator instead uses order statistics: it sorts your actual numbers and finds positions in that real, empirical distribution, with zero assumptions about shape.

A subtlety worth stating plainly, because it is the single biggest source of disagreement between percentile calculators online: when the percentile you ask for doesn't land exactly on one of your sorted values, there is more than one accepted convention for filling in the gap. This calculator uses linear interpolation between the two closest ranks — the method statisticians call 'Type 7,' after Hyndman and Fan's 1996 survey of nine competing definitions used across different software packages. Type 7 is the default method in R's quantile() function, in NumPy and pandas, and it matches Microsoft Excel's PERCENTILE and PERCENTILE.INC functions. If you instead compare against Excel's PERCENTILE.EXC function, or a textbook using an 'exclusive' method, you may see a slightly different answer for the same input, especially with small datasets — this is not a bug in either tool, it is a genuine, well-documented difference in convention, and Hyndman and Fan's paper is the standard reference cataloguing exactly which major software uses which method.

The reverse direction — finding the percentile rank of a specific value — uses the empirical cumulative distribution function: the proportion of your dataset that falls at or below the value in question, expressed as a percentage. If 7 of your 10 data points are 85 or below, the value 85 sits at the 70th percentile of your sample. This is the calculation behind 'what percentile am I in' questions for test scores, growth measurements, and performance benchmarks, whenever the comparison population is a real dataset rather than a published reference curve.

Use this calculator whenever you have your own data — a class's worth of scores, a season's worth of measurements, a survey's worth of responses — and want to know either what value sits at a given rank, or what rank a given value sits at. For percentiles computed against a known theoretical distribution instead of raw sample data, use the z-score or normal-distribution calculators; for the specific 25th/50th/75th split used to build box plots, use the quartile calculator, which is this same math applied at three fixed, well-known percentiles.

What is percentile calculator?

A percentile is a value below which a given percentage of the observations in a dataset fall. The Pth percentile is the value at rank position P/100 of the way through the sorted data — the 50th percentile is the median, the 25th percentile is the first quartile, and so on. Percentiles belong to the general family of quantiles, and unlike the mean or the standard deviation, they make no assumption whatsoever about the shape of the underlying distribution — they are computed purely from the rank order of the observations you actually have. When the requested percentile lands exactly on one of the sorted values, there is no ambiguity. When it falls between two sorted values — which is the common case for anything other than a handful of round-number percentiles — different statistical software packages fill the gap differently. Hyndman and Fan (1996) catalogued nine distinct conventions in active use ('Type 1' through 'Type 9'); this calculator implements Type 7, linear interpolation between the two nearest order statistics, which is the default behaviour of R, NumPy, pandas, and Excel's PERCENTILE/PERCENTILE.INC functions — almost certainly the convention you have already been using if you have ever called a percentile function in a modern data-analysis tool. The reverse operation, finding the percentile rank of a specific value, uses the empirical cumulative distribution function: the fraction of the dataset at or below that value, converted to a percentage.

How to use this calculator.

1. 01Paste your dataset into the box as comma-separated numbers, in any order.
2. 02Choose a mode: 'Value at a given percentile' if you know the percentile and want the corresponding value; 'Percentile rank of a given value' if you have a specific value and want to know its rank within the dataset.
3. 03In 'Value at a given percentile' mode, enter the target percentile (0-100) — for example 90 for the 90th percentile.
4. 04In 'Percentile rank of a given value' mode, enter the value you want to look up.
5. 05Read the primary result, and check the count (n) shown against how many values you intended to enter, to catch any parsing mistakes.
6. 06Remember that other tools may report a slightly different value for the same percentile if they use a different interpolation convention (e.g. Excel's PERCENTILE.EXC). This calculator uses the widely-adopted Type 7 method — see the FAQ for the full disclosure.

The formula.

To find the value at percentile P (Type 7, Hyndman & Fan 1996), first sort the n values into ascending order statistics x₍₁₎ ≤ x₍₂₎ ≤ … ≤ x₍ₙ₎. Compute the interpolated rank position h = 1 + (P/100)·(n − 1). If h is a whole number, the answer is simply x₍ₕ₎ — no interpolation is needed. If h falls between two integers, take the value at the position just below (x₍⌊h⌋₎) and the value at the position just above (x₍⌈h⌉₎), and interpolate linearly using the fractional part of h: result = x₍⌊h⌋₎ + (h − ⌊h⌋)·(x₍⌈h⌉₎ − x₍⌊h⌋₎). For example, with 10 sorted values and P = 90, h = 1 + 0.9·9 = 9.1: the 9th and 10th order statistics are blended with weight 0.1 on the 10th value. This is exactly the 'linear interpolation between closest ranks' method that Hyndman and Fan label Type 7, and it is the default in R, NumPy, pandas, and Excel's PERCENTILE/PERCENTILE.INC. The reverse direction — the percentile rank of a given value — uses the empirical cumulative distribution function: count how many of the n values are less than or equal to the target value, divide by n, and multiply by 100. This is a direct estimate of P(X ≤ target) from the sample itself, with no distributional assumption. Internally the calculator parses the comma-separated input, coerces every token to a number (throwing a descriptive error naming any token that fails), and sorts a numeric copy of the array before applying either formula.

A worked example.

Frequently asked questions.

References& sources.

1. [1]Hyndman, R.J. & Fan, Y. (1996). 'Sample Quantiles in Statistical Packages.' The American Statistician, 50(4), 361-365 — the canonical peer-reviewed survey of the nine competing sample-quantile conventions; this calculator implements Type 7.
2. [2]OpenStax. Introductory Statistics 2e. Rice University, 2023. §2.3, 'Measures of the Location of the Data' — percentile and quartile definitions and worked examples.
3. [3]NIST Dataplot Reference Manual — Empirical CDF Plot. Defines the empirical cumulative distribution function step formula used for the percentile-rank direction of this calculator.
4. [4]Casella, George and Berger, Roger L. Statistical Inference, 2nd edition. Duxbury / Cengage, 2002 — order statistics and the empirical distribution function treated rigorously in Chapter 5.
5. [5]Khan Academy. 'Percentiles.' Statistics and probability — free explainer aligned to the same order-statistics definitions used above.