Count the marks.
Understand the score.

No. Raw marks come from the answer ledger. Multi-session papers use a separate statistics-based transformation to obtain normalized marks. The score formula then uses raw marks for a single-session paper or normalized marks for a multi-session paper.

No. The 2027 formula is published, but its actual paper and session statistics are not available before the exam is evaluated. This page offers explicitly declared assumptions and historical 2026 parameters, not a calibrated future rank, percentile or admission prediction.

The displayed algebraic calculation is 350 + 550 × (M − Mq) ÷ (Mt − Mq). Mq is the General qualifying anchor for the paper, even when another category threshold is lower. Mt is the paper-wide top 0.1% or top 10 mean, whichever group is larger.

It is an unrounded algebraic value using the shown parameters. Historical report values are rounded published statistics. This tool does not reconstruct undisclosed precision, official score rounding or the reported score ceiling. Values above 1000 are labelled algebraic, and negative values are preserved. Use the actual official result for applications.

Ordinary wrong MSQ and NAT responses earn zero without a deduction. Wrong MCQs deduct one-third of their mark value. MSQ receives no partial credit: only the complete correct combination counts as correct. This aggregate ledger does not judge your answer combinations.

Under the 2027 pattern, a complete paper has 65 questions and 100 available marks. The ledger checks both, across six type-and-mark buckets. A subset has a separately labelled subtotal and cannot be transferred as a complete paper’s raw mark. Correct, wrong and unanswered counts are mutually exclusive.

The ledger supports ordinary scoring for valid questions only. Apply the final official answer key, including accepted NAT ranges, when deciding correct counts. Special awards, withdrawals or revised-key adjustments are outside these ordinary buckets; use officially adjusted marks in the score input instead.

The official September 2026 statistical report supplies36 records, including the separate GG branches and XH subjects. CE and CS used normalized marks in 2026; other included records use raw marks. Category cutoffs are copied as published, rather than recalculated from rounded General values.

The separate statistics panel calculates General = max(25, min(40, mean + standard deviation)), then the indicated category ratios. These are unrounded arithmetic ratios, not an issued cutoff table. Custom score parameters use the General anchor; historical comparisons use published 2026 category marks.

No. In normalization, each mean-plus-standard-deviation sum is a distribution statistic without the 25 floor or 40 cap. The score formula’s General qualifying anchor is a separate quantity. The normalization top group is top 0.1%, while the score anchor uses top 0.1% or top 10, whichever is larger.

It algebraically solves M = Mq + (target −350) × (Mt − Mq) ÷550 for the selected parameters. For a multi-session paper this is a normalized-mark target, not a raw-answer count. An out-of-range raw target is flagged.

You can compare up to eight numerical marks scenarios under one clearly labelled parameter set and download a CSV showing provenance and marks basis. Entered counts and labels stay in page memory; the tool does not request private response-sheet links or credentials. Reloading clears the inputs.