MHT CET Percentile Calculation: The CET Cell’s Own Formula, Explained

MHT CET runs in dozens of sessions, each with its own question paper, so the CET Cell does not publish marks. It publishes a percentile worked out inside each session. The method is set out in a seven-page “Document on Normalization” that has not changed since 2024: the 2025 and 2026 files have the same MD5 checksum, and the 2024 file’s text matches apart from whitespace (our check, 26 September 2026). This guide walks through the formula, what it does and does not do, and why it produces so many “100 percentile” results.

CET Cell “Document on Normalization”; CAP brochure 2026-27.

The document defines the percentile as the percentage of candidates in the session who scored equal to or below a candidate. In the CET Cell’s own form:

A worked example with the formula: in a session of 30,000 candidates, a candidate with 29,700 people at or below their raw score (including themselves) gets 100 × 29,700 ÷ 30,000 = 99.0000000. Nothing about marks survives into that number — two candidates with the same raw marks in different sessions can have different percentiles, and the CET Cell treats that as the point: it assumes candidates are spread randomly across roughly equal sessions, so a candidate’s standing in their own session is a fair measure.

Because every session has a top scorer, and the top scorer of every session gets exactly 100 — as does anyone tied with them. More sessions therefore mean more 100s, whatever the papers were like. The CET Cell’s own counts show it.

In 2026 the PCB second attempt had four sessions and exactly four 100-percentile candidates; the PCB first attempt had eleven and eleven. The PCM numbers run above the session count because of ties at the top. None of this says anything about how hard a paper was — it is the formula doing what it says.

The document states its aim plainly: candidates should be “neither benefitted nor disadvantaged due to the difficulty level” of the session they happened to sit. It gets there by assumption rather than by adjusting marks — it takes candidates to be allocated randomly across sessions of roughly equal size, so that every session holds a similar spread of ability. If that holds, being in the top 1% of your session means the same thing as being in the top 1% of any other, whatever each paper was like.

Two consequences are worth knowing. First, the bottom of the scale depends on session size: the lowest scorer in a session of 30,000 gets a much smaller percentile than the lowest scorer in a session of 3,000, because far fewer people are at or below them in proportion. The document notes this itself. Second, a hard paper does not hurt you by itself — if your session found it hard, your rivals did too, and your percentile reflects your rank among them, not the marks. What does hurt is losing easy marks your session-mates took, which is why accuracy on the straightforward Class 12 questions matters more than chasing the hardest ones. And with no negative marking, a guess on an unanswered question can only move you up.

The seven decimals do most of the work — ties on the total percentile are rare — but when two candidates share a total percentile, the brochure’s “marks or score” in each subject decides who is ahead, starting with Mathematics. That makes the Mathematics section, already worth 100 of the 200 marks, doubly important for PCM candidates.

The Document on Normalization ends with worked tables. They are illustrations, and they do not all follow the formula: the same percentile, 99.9119718, is shown for sessions of 2,249 and 2,216 candidates, which the formula cannot produce. Use the rule on pages 1 and 4, not the example numbers.