The Grade Curve Calculator, Explained
After reading this you will know what each curving method does to a set of raw scores, how to compute it by hand, and how to check whether a curve helped or hurt the class before you commit to it.
What a curve actually does
A curve is a rule that maps each raw score to a new score. The rule is the same for everyone, so the ranking of students never changes: whoever was fifth before is still fifth after. What changes is the spacing and the level of the scores.
Here is the hook. Suppose a hard exam produces a class median of 68 and one student earned a 55. Under the square-root curve the 55 becomes \sqrt{55} \times 10 \approx 74.2, a jump of 19 points, while a 91 becomes \sqrt{91} \times 10 \approx 95.4, a jump of only 4 points. The same rule moved the bottom student four times as far as the top student. That asymmetry is the whole point of choosing one method over another.
The calculator applies four families of curves (flat, scale-to-max, square-root, linear-to-mean, and bell) and shows you the before and after statistics side by side. Everything below explains the arithmetic so the numbers on the tool page are never a mystery.
When to curve, and when not to
Curve when the scores measure the class fairly but on the wrong scale. A well-written exam that turned out harder than intended is the classic case: the ranking is trustworthy, the absolute numbers are not. A curve fixes the level without discarding the ranking.
Do not curve to hide a broken question or a mis-keyed answer. Fix the key and re-score instead. Do not curve when grades are meant to be absolute (a driving test, a licensing exam, a safety checklist), because a curve makes a grade depend on who else sat the exam.
A bell curve is relative grading. If you force a target mean of 75 and a top student scored 100 on a fair exam, the bell can pull that 100 down below 100 unless you cap it. A student can then lose points for reasons that have nothing to do with their own work.
The five methods and their formulas
Let x be a raw score, \bar{x} the class mean, and s the class standard deviation.
Add flat points. Every score rises by the same amount a.
Spacing is untouched. If two students were 6 points apart, they still are. Only the level moves.
Scale best score to 100. Divide by the top score m and rescale.
This stretches the whole scale so the highest scorer lands exactly on 100. Gaps grow in proportion: a class with a top score of 91 has every score multiplied by 100/91 \approx 1.0989.
Square-root curve. Take the root and multiply by 10.
Because the square root grows fastest near zero, low scores gain the most and 100 stays fixed (\sqrt{100}\times 10 = 100). This compresses the top of the class and spreads out the bottom.
Shift to a target mean (linear). Add the difference between the target mean \mu_t and the current mean.
This is a flat add in disguise, but the amount is computed for you so the class mean lands exactly on the target. Spacing is preserved.
Bell curve to target mean and SD. Standardize each score, then rebuild it around the target mean \mu_t and target standard deviation \sigma_t.
The term (x - \bar{x})/s is the z-score: how many standard deviations the student sits from the mean. The bell curve is the only method here that changes both the level and the spread.
A worked example on the demo data
Curving the 12 default scores
The demo scores are 55, 62, 68, 71, 74, 75, 78, 80, 83, 85, 88, 91.
- Sum the scores: 55 + 62 + \dots + 91 = 910. Divide by 12 to get the mean \bar{x} = 75.83.
- Compute each squared deviation from the mean, average them, and take the root. The standard deviation (population) is s \approx 9.96.
- Add 5 points: every score rises by 5. The 55 becomes 60, the 91 becomes 96. The new mean is 75.83 + 5 = 80.83, the spread is unchanged.
- Scale best to 100: multiply by 100/91 \approx 1.0989. The 55 becomes 60.4, the 91 becomes 100.
- Square-root curve: the 55 becomes \sqrt{55}\times 10 = 74.2, the 91 becomes 95.4. New mean \approx 86.8.
- Bell to mean 75, SD 10: the 55 has z-score (55 - 75.83)/9.96 = -2.09, so it maps to 75 + (-2.09)(10) = 54.1. The 91 has z-score 1.52 and maps to 90.2.
| Method | 55 becomes | 91 becomes | Gap | New mean |
|---|---|---|---|---|
| Raw | 55 | 91 | 36 | 75.83 |
| Add 5 | 60 | 96 | 36 | 80.83 |
| Scale to 100 | 60.44 | 100 | 39.56 | 83.33 |
| Square-root | 74.16 | 95.39 | 21.23 | 86.81 |
| Bell (75, 10) | 54.09 | 90.24 | 36.15 | 75.00 |
Notice the gap column. Adding points and the bell curve keep the gap near 36. Scaling widens it to 39.56. The square-root curve shrinks it to 21.23, which is exactly why it rescues weak students at the cost of the top spread.
Seeing the shape change
Statistics summarize, but the histogram tells you what the curve did to the shape of the class. The chart below counts how many of the 12 students fall in each 10-point band, raw versus square-root.
Reading the before and after statistics
Three numbers tell you most of the story: the mean (the level), the standard deviation (the spread), and the count above your pass line.
- Mean
- The class average. Adding 5 points raises the demo mean by exactly 5, to 80.83. The bell curve sets it to whatever you type.
- Standard deviation
- The typical distance from the mean. The raw value is 9.96. Add and linear leave it unchanged. Square-root shrinks it. Bell sets it to your target.
- Pass count
- How many students clear the threshold. If passing is 70, the raw data has 8 of 12 passing. The square-root curve lifts the 68 to 82.5 and the 62 to 78.7, so all 12 pass.
Decide your target before you look at the methods. If you want a mean of 78 with no student above 100, the linear shift does it cleanly and keeps every gap intact. Reach for the bell only when you genuinely want to control the spread.
Common mistakes
Capping silently changes the mean. Turn the cap on and any score the curve pushed past 100 gets clipped back to 100. That lowers the achieved mean below your target. In the bell example, if three scores mapped above 100 and got capped, the realized mean drops below 75 even though you asked for 75.
Confusing scale-to-max with a bell. Scaling to 100 only guarantees the top student hits 100. It does not fix the mean or the spread. If the top student had an outlier-high score, everyone else barely moves.
Curving twice. If you add 5 points and later apply a square-root curve to the already-curved scores, the second curve operates on inflated numbers and the result is meaningless. Always curve the raw scores once.
Treating the square-root as fair for strong classes. On an easy exam where the mean is already 88, the square-root curve barely helps low scorers and still compresses the top. It is a hard-exam tool.
Related tools
Once grades are set, a few neighboring calculators finish the job. Use the Final Grade Calculator to find what a student needs on the final to reach a target course grade. Convert letter grades and credits into a grade point average with the GPA Calculator. And if you want to reason about guessing on a multiple-choice section rather than curving it after the fact, the Exam Pass Probability Calculator handles the binomial math.
Frequently asked questions
Does curving change who ranks first in the class?
No. Every method here is a monotonic function, so it preserves order. The student with the highest raw score keeps the highest curved score. Only the spacing between students can change.
Why does the square-root curve multiply by 10?
Raw scores run 0 to 100, but the plain square root of 100 is only 10. Multiplying by 10 rescales the output back to a 0 to 100 range, so \sqrt{100}\times 10 = 100 and the top stays put.
Which method should a teacher pick after a hard exam?
If the ranking is fair and you want to help the weakest students most, the square-root curve does that. If you want to lift everyone equally, a flat add or a shift to a target mean keeps the spacing intact and is easier to explain.
Why did my bell-curved mean not equal the target I typed?
The cap. With the cap on, any curved score above 100 is clipped to 100, which pulls the realized mean below your target. Turn the cap off to hit the target exactly, or accept that the mean will sit a little under it.
Can a curve ever lower a student's score?
Yes. A bell curve can push a top scorer below their raw mark if their raw score sat well above the target mean plus one target SD. Scaling to 100 and adding points never lower a score, but the bell can.
Related tools
Final Grade Calculator
What do you need on the final exam to reach your target grade? Instant answer for any weight.
GPA Calculator
Compute your GPA from course grades and credits — or ask what term GPA you need to hit a cumulative goal.
Exam Pass Probability Calculator
How likely are you to pass a multiple-choice exam by guessing the rest? Supports ruling out options and per-section pass rules.
Math Worksheet Generator
Printable worksheets: arithmetic, fractions, algebra, order of ops, decimals, mixed numbers and percents.
Spaced Repetition Simulator
SM-2, FSRS, HLR, Leitner and Pimsleur side by side — workload, retention and knowledge per study hour.
Flashcard Workload Planner
How many daily reviews will your Anki deck really cost — vacation strategies, easy days, exam pacing, SM-2 vs FSRS.