How CAT Quant Negative Marking Affects Your Raw Score
Most aspirants can recite the marking scheme within a week of starting. Almost none of them have run the arithmetic on what it does to a real paper, and the gap between knowing the rule and having worked out its consequences is where a large number of marks quietly disappear.
The rule itself is small. What it produces is not: a penalty structure that makes a blank free, makes a guess on one question type free and on another expensive, and makes accuracy worth more than volume in a way that is measurable rather than motivational. This piece does the arithmetic properly and shows you what it implies for how you attempt the section.
Accuracy is built by type, not by exhortation. Our percentages practice sets are where that work happens.
- Negative marking applies only to incorrect MCQs, never to Type In The Answer questions.
- An unattempted question scores zero, so a blank costs nothing at all.
- A wrong MCQ is a four mark swing away from the correct answer, not a one mark loss.
- Attempting more without accuracy can lower a raw score, and the arithmetic shows it exactly.
- There is no attempt count that produces a percentile, and anyone quoting one is guessing.
The Rule, and What It Hides
The scheme is short. A correct answer is plus three. An incorrect MCQ is minus one. An unattempted question is zero. Type In The Answer questions, where you type the answer rather than choose it, carry plus three for a correct response and no penalty for an incorrect one.
So negative marking applies only to incorrect MCQs. That is the whole rule, and everything below follows from it.
What the rule hides is the size of the gap between outcomes. The distance between a correct MCQ and an incorrect one is four marks, not one, because you both fail to gain three and lose one. Candidates think in terms of the penalty and therefore underestimate the swing by a factor of four.
Describing the penalty as "only minus one, so it is worth a try." The comparison that matters is not between minus one and zero. It is between the three marks you would have gained by being right and the one you lose by being wrong, and that is the number driving the decision.
The Arithmetic on a Real Paper
Abstract rules do not change behaviour. Worked numbers do, so here is the calculation on a paper of 66 questions and 198 marks, which is roughly the shape recent papers have taken.
Take a candidate who attempts 63 of the 66 questions at 41 percent accuracy, which is around 26 correct. Those 26 correct answers are worth 78 marks gross. The actual net score on that paper was 54.
The 37 attempts that went wrong cost 24 marks. Had only the 26 genuinely known questions been attempted and everything else left blank, the same paper returns 78 instead of 54. Attempting nearly the entire paper made the score materially worse.
That example is a worked case rather than anyone's story, and the low accuracy is unrepresentative because the paper was deliberately attempted almost in full. The arithmetic is the transferable part and it does not depend on the accuracy figure. A blank is free at 41 percent and a blank is free at 80 percent, while a wrong MCQ costs at both.
The Break Even, Stated Honestly
Candidates want a threshold, so here is the honest version of one. On an MCQ where you have genuinely eliminated nothing, a random choice among four options gains three marks a quarter of the time and loses one the other three quarters, which averages to zero. It is not a disaster and it is not an edge.
Eliminate one option and the arithmetic turns positive. Eliminate two and it is clearly worth doing. The practical rule is that a guess is worth making once you have removed at least one option on real grounds, and is not worth making when you have removed nothing.
Treat that as a way of thinking rather than a formula to apply under pressure, because in the hall you will not be computing expectations. What you will be doing is deciding whether your elimination was real or wishful, and that is the judgement to train.
The Type In The Answer Asymmetry
On Type In The Answer questions the entire calculation disappears, because there is no penalty. If you have worked a question and arrived at a value you are unsure about, entering it costs nothing and leaving it blank guarantees zero.
This produces the single cheapest habit available in the section: never leave a worked Type In The Answer question empty. Candidates lose marks here through caution that the marking scheme does not reward, and it is entirely avoidable once the asymmetry is explicit.
| Situation | What it scores | What to do |
|---|---|---|
| MCQ, no elimination | Averages to zero over many attempts | Usually leave it, and spend the time elsewhere |
| MCQ, one option eliminated | Turns positive | Worth attempting |
| MCQ, two eliminated | Clearly positive | Attempt it |
| TITA, any worked answer | No penalty for a wrong entry | Always enter something |
| Unattempted | Zero | Free, and that is the point |
What This Does to Attempt Strategy
The marking scheme is not a detail bolted onto the paper. It is the reason selection and accuracy are the two levers that move a score and volume is not.
- A blank is free. Leaving a question costs nothing, which makes declining a legitimate move rather than an admission of defeat.
- Accuracy compounds. Raising accuracy improves the score twice over, by adding correct answers and removing penalties at the same time.
- Volume alone can hurt. More attempts at low accuracy can reduce a raw score, as the worked example shows.
- The two question types need different endings. Uncertainty on an MCQ is a reason to consider leaving it; uncertainty on a TITA never is.
Why Nobody Can Give You an Attempt Count
The most common question here is how many questions to attempt for a given percentile, and it has no answer. Not a withheld answer, an absent one.
A raw score comes from your correct and incorrect responses. A percentile is your position relative to everyone else who sat the exam, and it moves with paper difficulty and with normalisation across the three slots. There is no fixed mapping from marks to percentile, so any number quoted as "attempt this many for that percentile" is a guess dressed up as guidance.
The number that is right for you depends on your accuracy, on that paper, on that day. Anyone who hands you a count is telling you about their own paper rather than about yours.
In mock review, calculate what your score would have been if you had left every attempt you were unsure about. Comparing that figure with your actual score turns the abstract rule into a number about your own paper, and it changes behaviour far faster than reading the scheme again.
The Sectional Limit Makes It Sharper
Each section runs 40 minutes with a hard limit and no movement between them, and that interacts with the marking in a way worth naming.
Time spent on a question you eventually get wrong is charged twice. You pay the minutes, and then you pay the mark. Since those minutes come out of questions at the end of the section you will now never see, a single expensive wrong MCQ can cost far more than four marks once the opportunity cost is included.
Building the Lever That Actually Moves
If accuracy is the lever, it is worth being specific about what raises it, because "be more accurate" is not an instruction anyone can follow.
Accuracy in Quant is mostly built by type rather than in general. A candidate is not uniformly 70 percent accurate; they are reliable in some question types and unreliable in others, and the unreliable ones are identifiable from mock review. Concentrated work on those types, in the areas that recur most, is what moves the number. Arithmetic and Algebra have historically been the largest contributors to QA, with Geometry and Number System behind them, so that is the sensible order for the concentrated work.
The second contributor is knowing where your line is. A candidate who can tell the difference between "I recognise this and can finish it" and "I think I can probably work this out" makes far better attempt decisions than one who cannot, and that self-knowledge comes from reviewing attempts by confidence rather than only by outcome.
One Number Not to Read Into
While running these counterfactuals you will be looking at a percentile on your mock report, and it is worth saying what that figure is and is not.
A mock percentile is your position within that mock's test-taking cohort. Those cohorts are self-selected and small, and they are not the CAT field. A mock percentile is therefore useful as a relative signal across your own mocks from the same series, and it is not a prediction of a CAT percentile. Treat movement in it as information about your trend and treat its absolute value with suspicion.
This matters here because the counterfactual exercise produces a second number, and it is tempting to convert both into percentiles and compare. Do not. Compare raw scores, which are the thing the marking scheme actually determines, and leave the percentile out of an arithmetic exercise it cannot support.
The Summary
The rule is small and its consequences are not. Plus three for a correct answer, minus one for an incorrect MCQ, zero for a blank, and no penalty at all on Type In The Answer questions.
That makes the gap between a right and wrong MCQ four marks rather than one, makes a blank genuinely free, and makes accuracy worth more than volume in a way the worked example demonstrates: 63 attempts at 41 percent accuracy returned 54 marks where attempting only the 26 known questions would have returned 78.
So guess on an MCQ only once you have eliminated at least one option on real grounds, never leave a worked Type In The Answer question blank, and stop looking for an attempt count, because percentile has no fixed relationship to a raw score and the right number for you depends on your accuracy on that paper. Run the counterfactual on your own mocks and the rule stops being something you know and becomes something you use.
- Do you know what your last mock would have scored with every uncertain attempt left blank?
- Do you enter an answer on every Type In The Answer question you have worked?
- Can you tell the difference between real elimination and wishful elimination?
- Do you know which question types your accuracy drops in, by name?
If those questions are hard to answer from your own data, the marking scheme is still an abstraction for you rather than a tool. A CAT preparation strategy review will run the numbers on your attempts, and a personalised CAT preparation plan puts the accuracy work where the losses actually are.
Run the Counterfactual Once
Seeing what your own paper would have scored without the uncertain attempts changes behaviour faster than any rule.
Build My Weekly PlanFrequently Asked Questions About CAT Negative Marking
How does negative marking work in CAT Quant?
A correct answer is plus three, an incorrect MCQ is minus one, and an unattempted question is zero. Type In The Answer questions carry plus three for correct and no penalty for incorrect, so negative marking applies only to incorrect MCQs.
Should I guess on CAT MCQs?
With nothing eliminated, a random choice among four options averages to zero over many attempts, so it buys nothing. Once you have removed at least one option on real grounds the arithmetic turns positive, and removing two makes it clearly worth doing.
How many questions should I attempt for a high percentile?
No such number exists. Raw score comes from your correct and incorrect responses while percentile is your position relative to everyone else, and it moves with paper difficulty and normalisation, so any quoted attempt count is a guess.
Should I ever leave a TITA question blank?
Not if you have worked it and arrived at a value. There is no penalty for an incorrect Type In The Answer response, so entering an uncertain answer costs nothing while leaving it blank guarantees zero.
Practice these Quant concepts, chapter by chapter
Thousands of CAT Quant practice questions by chapter and difficulty, each with a worked solution.
More from Quant
Continue reading
XIMB Admissions and the Deadline That Actually Binds
Are CAT VARC Passages Deliberately Written to Confuse?
What High Scorers Actually Do in Their Final CAT Week
When You Cannot Understand a CAT RC Passage Opening
Put it into practice