Three numbers answer three different questions.
A raw score measures performance against configured marks. A rank orders eligible results. A percentile locates a score inside a defined comparison group. Mixing them creates avoidable disputes: a learner can earn the same percentage in two mocks but receive different percentiles because the cohorts and score distributions differ. Start with the test rule, then calculate the score, then apply the comparison policy.
Correct, incorrect, skipped, and partially correct responses must use the configured marks for their question or section.
The chosen attempt, eligible learners, exclusions, and ties determine the reported position.
Equal-to-or-below and below-only methods can produce different figures, especially where scores tie.
Important: this page uses a uniform score example and an inclusive NTA-style percentile example. It does not claim that every examination—or Edugenie's production calculation—uses the same method.
Mock-test score calculator with negative marking
Enter a simple uniform marking rule. The calculator runs in your browser and does not submit these values. Use a separate calculation for each section when marks vary by question or when partial credit applies.
Calculator boundary: skipped responses are zero and every question has the same positive and negative marks in this example. It does not model partial credit, section-specific rules, cancelled questions, bonus marks, or exam-specific normalization.
Write the marking rule before deciding whether a score is correct
The simplest uniform formula is score = correct × positive marks − incorrect × deduction. Enter the deduction as a positive magnitude and subtract it once. Do not enter “−1” and subtract it again, because that would incorrectly add a mark.
48 positive marks minus 5 deducted marks. 3 skipped responses add zero under this example rule.
| Response outcome | Configured effect | What to verify |
|---|---|---|
| Correct | +4 | The accepted answer and question version are final. |
| Incorrect | −1 | The deduction matches the published mock rule. |
| Skipped | 0 | No penalty applies to an unanswered question under this rule. |
Do not assume one rule fits every test
Some examinations deduct a fixed mark, some use a fraction of the positive mark, and some have no negative marking. Obtain the current official bulletin for the exact examination pattern being simulated.
Keep partial credit separate
A multiple-answer question may have a distinct partial-credit rule. Test the complete set, a valid subset, an invalid selection, no response, and any disqualifying choice rather than forcing it into the simple formula.
Permit negative totals only intentionally
A raw score can fall below zero when deductions exceed positive marks. Decide whether the platform and published policy retain that value or apply a floor; never add a floor after seeing learner results.
Rank is an ordering policy, not a property of marks alone
To report rank, first define the eligible comparison set. Then choose which attempt represents each learner, order the evaluated scores, and apply the published tie rule. A score of 74 cannot reveal its rank until the other eligible scores and the policy are known.
| Candidate | Score | Competition rank | Reason |
|---|---|---|---|
| A | 82 | 1 | No higher score |
| B | 74 | 2 | One higher score; tied on 74 |
| C | 74 | 2 | One higher score; tied on 74 |
| D | 61 | 4 | Three candidates ahead |
The sequence is 1, 2, 2, 4 under competition ranking. Dense ranking would produce 1, 2, 2, 3. A separate tie-break policy could order B and C using an approved criterion. All are different policies; none should be selected after the scores are visible.
Who counts?
Only learners and attempts that meet the written eligibility rule. Decide how to treat absences, unsubmitted attempts, invalidated attempts, test accounts, and late completions.
Which result counts?
First, latest, or highest attempt can produce different rankings. State the rule before a repeatable mock opens and apply it to everyone.
What happens at equal score?
Shared rank is transparent. If a tie-break is necessary, publish the ordered criteria and test them with known examples before release.
Percentile is not percentage: try an inclusive cohort example
The current NTA JEE (Main) 2026 result notice describes an NTA Score using candidates in a shift with a raw score equal to or less than the candidate, divided by candidates who appeared in that shift, multiplied by 100. Other statistical tools use strict-lower or midpoint conventions. The calculator makes all three visible so the method cannot hide behind the word “percentile”; none is presented as Edugenie's verified production formula or a national-rank predictor.
| Measure | Question answered | Data required | Can it change when the cohort changes? |
|---|---|---|---|
| Raw score | What marks did the responses earn? | Responses, final key, and marking rules | No, unless the key or scoring rule changes |
| Percentage | What share of maximum marks was earned? | Raw score and maximum marks | No, unless those values change |
| Rank | What is the ordered position? | Eligible results, attempt rule, and tie policy | Yes |
| Percentile | Where is the score relative to a cohort? | Eligible cohort, score distribution, method, and rounding | Yes |
Mock-test boundary: a coaching cohort is not the official examination population. The same learner can have different mock and official percentiles even with similar raw performance because the questions, attendance, score distribution, shifts, and normalization method differ.
Run these 8 checks before publishing comparative results
A defensible result can be reproduced from the same inputs and rules. Record the policy before launch, validate it with controlled attempts, and do not silently change the cohort or tie logic after learners have seen a rank.
- 01Fix the eligible cohort
State which learners and completed attempts are included before calculating any comparison.
- 02Choose the attempt
For repeatable mocks, define whether first, latest, highest, or another approved attempt enters the ranking.
- 03Freeze the answer key
Resolve accepted answers and scoring corrections before publishing final comparative results.
- 04Verify every scoring branch
Hand-check correct, incorrect, skipped, multiple-answer, partial-credit, and user-input examples that the test uses.
- 05Publish the tie rule
State whether ties share a rank and what, if anything, breaks a tie before learners begin.
- 06Define the percentile method
Record whether the count uses scores below or equal to-or-below, the cohort boundary, and the rounding precision.
- 07Separate live and final results
If the cohort is still changing, label comparative figures as provisional and state when they become final.
- 08Keep an audit record
Retain the test version, cohort rule, answer-key version, calculation check, approver, and release time.
Minimum learner-facing result note
Example wording: “This rank compares eligible submitted attempts for this mock under the published attempt and tie rules. Comparative figures remain provisional until the stated result-finalisation time. They do not predict an official examination rank.”
Adapt the wording to the actual cohort, attempt rule, tie method, and release schedule. Do not publish a generic note that contradicts the configured behavior.
Configure scoring, then verify the complete result workflow.
Edugenie currently supports positive and negative decimal marks, automatic evaluation for supported question formats, result detail, comparative rank and percentile, and leaderboards. The approved public brief does not define the exact production cohort, attempt-selection, tie-breaking, percentile, rounding, or recalculation rules. Confirm those details in the current product workflow before making a consequential promise to learners.
Positive and negative marks, supported answer evaluation, totals, selected and correct responses, and explanations when configured.
Stored rank and percentile plus leaderboards for evaluated attempts.
Advanced item analysis, AI insights, weak-topic detection, and cohort trend reporting are not presented as current capabilities.
Before release: use known sample attempts to verify score, negative marks, skipped responses, partial-credit behavior where used, rank, percentile, ties, and leaderboard order.
References and review basis
The inclusive percentile example follows the session-based formula stated in NTA's JEE (Main) 2026 result notice. SciPy's statistical documentation shows why strict, weak/inclusive, and mean conventions need explicit naming. The JEE information-bulletin page is linked as an example of checking the current exam rule, and NTA's test-development guidance supports independent answer-key review. These references neither define Edugenie's production algorithms nor endorse the product.
- NTA: JEE (Main) 2026 Session 2 result and NTA Score notice — equal-to-or-less percentile formula and the distinction from percentage of marks.
- SciPy: percentileofscore conventions — strict, weak/inclusive, rank, and mean approaches for tied scores.
- JEE (Main): official information bulletins — an example of using the current official examination instructions as the rule source.
- National Testing Agency: development of tests — blueprinting, repeated review, and independent answer-key verification principles.
Last reviewed: .
Explain every result—without promising a rank.
Straight answers about negative marking, skipped responses, percentage, percentile, rank, ties, cohorts, and Edugenie's current boundary.
What is the basic negative-marking formula?
For a uniform rule, raw score equals correct answers multiplied by marks per correct answer, minus incorrect answers multiplied by the deduction per incorrect answer. Skipped answers contribute zero only when the published test rule says so. Section-specific and partial-credit rules must be calculated separately.
Is negative marking compulsory in a mock test?
No. Use it only when it matches the assessment purpose or the official exam pattern being simulated. A diagnostic classroom test may need a different rule from a competitive-exam mock.
What is the difference between percentage and percentile?
Percentage compares a learner's score with the maximum available marks. Percentile compares that learner with a defined group of candidates under a stated calculation method. The two numbers answer different questions and need not be similar.
Can rank be calculated from percentile alone?
Not reliably. You also need the exact eligible cohort, the percentile definition, treatment of ties, attempt-selection rule, exclusions, and often the underlying score distribution. A percentile should not be converted into a promised external-exam rank.
What happens when two learners have the same score?
The result depends on the published tie policy. They may share the same competition rank, receive dense ranks, or be separated by approved tie-break criteria. The rule should be fixed before the test rather than chosen after seeing results.
Should absent or unsubmitted learners count in percentile?
Use only the cohort defined in the result policy. An NTA-style appeared-cohort example uses candidates who appeared, not everyone registered. A coaching mock may use another documented eligibility rule, but it should be applied consistently.
Does a mock-test percentile predict an official exam percentile or rank?
No. The mock cohort, question difficulty, attendance, attempt policy, tie rules, and any official normalization process differ. Treat a mock percentile as a comparison inside that stated mock cohort, not a national rank guarantee.
Does Edugenie support negative marks, rank, and percentile?
Edugenie currently supports positive and negative decimal marks, automatic evaluation for supported question formats, comparative rank and percentile, and leaderboards. Confirm the current cohort, attempt, tie, rounding, and update rules before relying on a result for a consequential decision.