XAT runs two separate penalty systems in Part 1, and they punish different behaviour. Getting a question wrong costs 0.25 marks, the same flat rate most aspirants expect from a CAT-style exam. But skipping too many questions carries its own, separate cost: the first eight unattempted questions in Part 1 are free, and every one after that costs 0.10 marks - even if you never wrote a wrong answer.
The rule, stated precisely
Per XLRI's own published exam pattern, XAT's Part 1 (Verbal & Logical Ability, Decision Making, and Quantitative Ability & Data Interpretation, taken together across 170 minutes) allows up to 8 unattempted questions with no penalty. From the 9th unattempted question onward, each additional skip deducts 0.10 marks. This allowance is for Part 1 as a whole - it is not eight free skips per section, it is eight total across all three sections combined. General Knowledge (Part 2) does not carry this rule; it has no negative marking of any kind, for wrong answers or skips.
Why this is a genuinely different decision from wrong-answer negative marking
On most exams with only wrong-answer penalties, a cautious candidate can, in principle, skip every question they are unsure of and take zero risk - the cost of caution is only the marks they would have earned, never a direct deduction. XAT removes that safety net past a threshold: excessive skipping is itself penalised, which means a strategy of "only answer what I'm fully confident about" has a real, quantifiable cost once it pushes your unattempted count past eight.
Working the arithmetic requires knowing how many options a XAT MCQ has - five, with one correct, per XLRI's own official exam-day instructions. That number matters more than it looks: a -0.25 penalty against five options lands exactly on the break-even point. The general formula for a random guess's expected value is (1/n) x 1 - ((n-1)/n) x p, where n is the option count and p is the penalty. For XAT: (1/5)(1) - (4/5)(0.25) = 0.2 - 0.2 = exactly 0 - the same design principle behind GATE's 1/3 penalty on four options (also exactly 1/(n-1), also exactly break-even). This is very unlikely to be a coincidence across two different Indian exams using the same 1/(n-1) ratio.
| Behaviour | Expected value |
|---|---|
| Blind guess, no options eliminated (5 options) | 0 - exactly break-even |
| Guess, 1 option eliminated (4 remain) | +0.0625 |
| Guess, 2 options eliminated (3 remain) | +0.167 |
| Guess, 3 options eliminated (2 remain) | +0.375 |
| Skip, within your first 8 unattempted (Part 1 total) | 0 - identical to a blind guess |
| Skip, the 9th unattempted and beyond | -0.10, guaranteed |
Read the practical rule directly off that table: within your first eight skips, a blind guess and a skip are mathematically identical (both zero) - there is no edge either way, so skipping there costs nothing extra and is a legitimate choice for a question you genuinely cannot start. Past the eighth skip, that equivalence breaks: a blind guess still nets zero, but a skip now costs a guaranteed -0.10 - so a guess, even with zero information, strictly dominates skipping once the free allowance is used up. And the moment you can eliminate even one of the five options, guessing beats skipping outright, free-allowance or not.
What this means in practice
The eight-skip allowance functions as a buffer for questions you cannot even begin - not a target to use up by choice. A candidate managing their attempts well should be aiming to stay near or under that threshold through genuine time management across Part 1's three sections, and treating the 9th-and-beyond skip cost as a real deterrent against a "skip anything mildly uncertain" habit that might be safe on a different exam but is not free here.
Where this is uncertain
- Whether the eight-question allowance is measured per section or across all of Part 1 combined is the detail most likely to be misreported - this post states it as whole-Part-1, cross-checked against XLRI's own pattern page and a secondary discussion source, but if a future cycle changes this specific mechanic, verify against that cycle's own official bulletin before relying on it.
- Numbers above are cited for the XAT 2026/2027 cycle pattern; XAT's core format has been stable since the 2025 removal of the essay section, but marking-rule specifics should be re-checked against each new cycle's official material.
Sources
- xatonline.in - The Complete XAT Exam Pattern and Syllabus Guide: xatonline.in/blog/xat-2026-syllabus-exam-pattern - official XLRI domain, marking scheme.
- Expected-value arithmetic in this post is original, worked from the published marking scheme above - not quoted from XLRI.
Tracking how many questions you are leaving unattempted per mock - not just your accuracy on the ones you answer - is the number this specific rule asks you to watch. Karma Yogi's XAT tracking logs attempted and correct counts per section on every full mock.
End of essay
- Anish Guruvelli