Most candidates think about negative marking as a punishment to be avoided. It is more useful to treat it as a price. Every attempt is a bet with a known payoff, and the whole question is whether the bet is worth taking. Once you think about it that way, the answer becomes arithmetic rather than nerve.
The arithmetic
Take a four-option question worth one mark, with a penalty of one quarter for a wrong answer, which is the banking standard.
If you guess blindly, you have a one in four chance of gaining one mark and a three in four chance of losing a quarter.
Expected value = (1/4 × 1) − (3/4 × 1/4) = 0.25 − 0.1875 = +0.0625
Marginally positive, which surprises people. But that is the mathematically pure case, and it assumes your guess is genuinely random. In practice it is not: exam setters build distractors specifically to attract the plausible wrong reasoning, so a candidate who has half-solved a question is more likely than chance to land on a designed trap. Treat blind guessing in a banking paper as break-even at best.
If you eliminate one option, you are choosing between three.
Expected value = (1/3 × 1) − (2/3 × 1/4) = 0.333 − 0.167 = +0.167
If you eliminate two options, you are choosing between two.
Expected value = (1/2 × 1) − (1/2 × 1/4) = 0.5 − 0.125 = +0.375
Now the same question in a railway paper, where the penalty is one third:
| Situation | Banking, penalty 1/4 | Railways, penalty 1/3 |
|---|---|---|
| Blind guess of four | +0.06 | 0.00 |
| One option eliminated | +0.17 | +0.11 |
| Two options eliminated | +0.38 | +0.33 |
The practical rule that falls out of this table:
- Eliminate two options: always attempt. In both exams this is clearly profitable.
- Eliminate one option: attempt. Positive in both, though thinner in railways.
- Eliminate nothing: leave it. The expected gain is zero or near zero and the variance is all downside, because your guess is not truly random.
Why the real cost is time, not marks
Here is what the expected-value table hides. In a sixty-minute prelims paper with a twenty-minute sectional clock, the scarce resource is not marks, it is seconds. A hard puzzle set you eventually get wrong has cost you a quarter mark and eight minutes. The eight minutes is by far the more expensive loss, because it was enough time for six questions you would have got right.
This is why the post-attempt analysis on this site shows time per question against a benchmark, and flags your slowest questions. A wrong answer that took four minutes is a completely different failure from a wrong answer that took thirty seconds, and they need different fixes.
The selection habit
Selection is a skill and it does not appear under pressure unless you have practised it. The method that works:
Take two passes through each section.
On the first pass, answer only what you can see through immediately. Do not fight anything. If a question has not resolved in about fifteen seconds, mark it for review and move on. In reasoning, read all the puzzle sets first and start with the one whose constraints look most determinate, not with the first one printed.
On the second pass, return to the marked questions with the time that remains, in order of how close you were.
Set a hard per-question cap and obey it. Something like forty-five seconds in prelims quant, ninety seconds for a full data interpretation set, and a firm limit on any single puzzle set. The cap matters far more than its exact value, because the failure mode is almost always the same: sinking four minutes into one question out of sheer unwillingness to abandon the investment already made.
That unwillingness has a name, the sunk cost fallacy, and recognising it in the moment is most of the cure. The three minutes are gone whether or not you spend a fourth.
Sectional cutoffs change the objective
In IBPS prelims and mains there are sectional cutoffs. That means your goal is not to maximise the total. It is to clear every section and then maximise.
Concretely: if you are strong in English and weak in quant, do not spend your English section pushing from twenty-five to twenty-eight correct. Clear it comfortably, then take the saved minutes into quant, where the marginal mark is the one that decides whether you qualify at all. A candidate who scores brilliantly in two sections and misses one sectional cutoff by half a mark has failed the paper.
RRB NTPC is different. There is no sectional cutoff and no sectional timing in either computer based test, so the whole ninety minutes can go wherever it earns most. That freedom is a genuine advantage and most candidates underuse it.
The one number worth tracking
Not your score. Track this instead, after every mock test:
Of the questions you attempted and got wrong, how many did you have no business attempting?
That number is your selection error rate, and it is the most improvable thing in your preparation. Knowledge takes months to build. Selection discipline can improve measurably in a week, and it is worth several marks in every paper you sit.
Frequently asked questions
Is it better to attempt more questions or fewer with higher accuracy?
Neither in the abstract. What maximises your score is attempting every question where your expected value is positive and leaving every question where it is negative. That usually produces high accuracy as a side effect, but accuracy is not the goal; it is a symptom of good selection.
How many questions should I attempt in IBPS PO prelims?
There is no single correct number, because it depends on the difficulty of your paper. As a working reference, strong candidates typically attempt roughly seventy to eighty of the hundred questions with accuracy above eighty-five per cent. If you find yourself attempting ninety-five, you are almost certainly guessing into the negative marking.
Does leaving questions blank hurt my score?
A blank scores zero, which is better than a wrong answer's minus a quarter or minus a third. There is no penalty for unattempted questions and no benefit to attempting everything. The only cost of a blank is the opportunity you would have had if you actually knew the answer.
Read next
General Science
General science for RRB NTPC: the twelve marks with the best return on study time
General science is about twelve of the forty general awareness marks in CBT 1, drawn from class ten physics, chemistry and biology. This is what actually gets asked, and what you can leave out.
General Awareness
How to build a study plan that survives a job, college or a bad week
A realistic framework for preparing for Bank PO or Railway exams alongside other commitments, built around weekly review rather than a rigid daily timetable that collapses on the first missed day.
Quantitative Aptitude
Data interpretation: how to read a set before you solve it
Data interpretation is the largest single block in the quantitative section. The skill being tested is reading and selection, not arithmetic, and this is the routine that gets a set done in three minutes.
Banking Awareness
RBI and monetary policy: the banking awareness that actually gets asked
The Reserve Bank of India, the policy rates, the Monetary Policy Committee and the regulatory framework, organised the way bank exams question it rather than the way textbooks explain it.