IBPS PO Prelims Mastery Course 2026

How Banking Quant Is Actually Tested — and the Maths of Scoring

Banking quant is not a maths test. It is a speed-and-selection test that happens to use maths. Two candidates with identical arithmetic ability routinely finish twenty marks apart, and the gap is almost never knowledge — it is the number of seconds each of them spent on questions they were never going to solve. This chapter builds the frame the rest of the book fills in: where the section sits in each of the six target exams, what it actually asks, and the exact arithmetic of scoring under negative marking, so that every "attempt or leave" decision you make later is a calculation rather than a feeling.

KarmSakha preparation topics covered

• The six target exams and the two stages: which of them calls the section Quantitative Aptitude, which calls it Numerical Ability, and which renames it Data Analysis & Interpretation at mains
• A question-type census: simplification, approximation, number series, quadratic comparison, quantity comparison, data sufficiency, DI sets and arithmetic word problems, mapped to the exam that favours each
• The scoring equation E = C − pW, the break-even accuracy a* = p/(1 + p), and the marks-kept-per-attempt table
• Seconds-per-question budgeting from your own notification's question count and clock, and the three-pass method for a prelims section
• The 45-second leave-it rule, and the exact point at which eliminating options makes a guess mathematically profitable
• Which figures are notification-controlled — question counts, marks, sectional timings, the negative-marking fraction — and where to read them

One Section, Six Exams, Two Names

This book targets six recruitment exams run by two bodies: IBPS PO (Probationary Officer / Management Trainee), IBPS Clerk, SBI PO, SBI Clerk (Junior Associate), IBPS RRB Officer Scale-I and IBPS RRB Office Assistant. All six run a preliminary stage followed by a main stage, and all six test the same underlying arithmetic. What differs is the label on the section and how much of the paper is data-driven.

At the preliminary stage the officer-level exams call the section Quantitative Aptitude; the clerical-level exams call it Numerical Ability. The names are not decoration. Numerical Ability papers lean towards direct calculation — simplification, approximation, straightforward series, single-topic word problems. Quantitative Aptitude papers assume that calculation is already automatic and spend their difficulty budget on interpretation instead.

The bigger shift happens at mains. In the PO-level mains papers the section is renamed Data Analysis & Interpretation (DA&I). That rename is the syllabus change: standalone arithmetic largely disappears and the same arithmetic returns wrapped inside tables, caselets and multi-set data. A profit-and-loss question does not vanish at mains — it becomes question 3 of a five-question caselet where the cost price has to be back-solved from a table first. This is why chapters 6 to 9 of this book (the arithmetic spine) are not optional for a mains candidate; they are the engine that Chapters 10 and 11 drive.

One practical consequence: a Clerk or Office Assistant aspirant should weight the calculation engine (Chapter 2) and simplification (Chapter 3) far more heavily than a PO aspirant, who can afford to assume that layer and invest in DI. Both, however, need Chapter 12's validity conditions, because a shortcut misapplied costs the same everywhere.

The Question-Type Census: What Each Exam Actually Asks

Banking quant is built from a small, closed set of question types. There are eight of them, and once you can name the type from the first line of the question you have already saved five seconds and avoided the commonest error, which is starting a DI question by reading the table instead of the footnote.

The table below records the typical emphasis each type receives at each level. Treat it as a study-weighting guide, not a guarantee: the composition of any one paper varies between shifts and cycles, and only the current notification and information handout define the official structure. What has been stable is the direction of the difference — clerical prelims reward raw calculation speed, officer prelims reward selection, and mains rewards the ability to reuse a derived value across four questions instead of recomputing it.

Read the table by column, not by row. Your column is the exam you have applied for. Everything marked Heavy in your column is where your practice hours should go, and everything marked Rare is worth exactly one revision pass so that the format does not surprise you on exam day.

The Maths of Scoring: E = C − pW

Every one of these exams applies a penalty for a wrong answer. Let C be the number of correct answers, W the number of wrong answers and p the penalty fraction of one question's marks that a wrong answer forfeits. Your net score is E = C − pW, and unanswered questions contribute nothing either way. The information handouts for these recruitments have commonly stated p as one-fourth of the marks assigned to that question — confirm the fraction in your own notification, because it is a notification-controlled figure and it has been stated differently at different times and for different posts. All the worked arithmetic below uses p = 1/4 on one-mark questions, so the method survives even if the number changes.

Rewrite the equation in terms of accuracy, because that is the variable you can actually control. If you attempt n questions with accuracy a (the fraction of attempts you get right), then C = an and W = (1 − a)n, so E = n[a − p(1 − a)] = n[a(1 + p) − p]. The bracket is your marks kept per attempt. At p = 1/4 it equals 1.25a − 0.25.

Set that bracket to zero and you get the true break-even accuracy: a* = p/(1 + p) = 0.25/1.25 = 0.20, or 20 per cent. Below 20 per cent accuracy your attempts destroy marks; above it they add marks. Since a blind guess among five options is exactly 20 per cent, blind guessing is worth precisely zero in expectation — not a catastrophe, but not free marks either. This is the single most misquoted number in banking preparation, so state it carefully: 20 per cent is the mathematical break-even, while 80 per cent is the practical target, because 80 per cent is where you still keep three-quarters of every attempted mark.

The consequence is brutal and worth memorising: it takes 1/p = 4 wrong answers to erase one correct answer. A candidate who fires off twelve hopeful attempts and gets four of them right has gained 4 and lost 8 × 0.25 = 2, netting 2 marks for perhaps six minutes of work. The same six minutes spent on ten questions from a strong topic at 90 per cent accuracy nets 9 − 0.25 = 8.75.

Seconds Per Question, and the Three-Pass Method

Your time budget is arithmetic, not instinct. If your notification gives the section Q questions in T minutes, the average budget is s = 60T/Q seconds per question. Compute this once for your own exam and write it on the first page of your notebook, because every later decision in this book is measured against it. If, for illustration, your paper gives 35 questions in 20 minutes, then s = 1200/35 ≈ 34 seconds.

The average is a budget, not a rule — question types do not cost the same. A simplification takes 15 to 25 seconds. A number series takes 25 to 40. A DI set costs a fixed 45 to 75 seconds just to read the table, footnotes and units, after which each question inside it costs only 20 to 30 seconds. That fixed reading cost is why a half-read DI set is the single worst investment in the paper, and why you should never begin a set you do not intend to finish.

The three-pass method turns this into a plan. Pass 1, roughly the first third of the clock: sweep the free marks — simplification, approximation, easy series, quadratics — without pausing on anything that resists. Pass 2, the middle third: take one or two DI sets, chosen by glancing at the numbers first (round figures and clean percentages mean a fast set; four-digit values with odd percentages mean a slow one). Pass 3, the final third: return for arithmetic word problems from your strong chapters, and only then for anything you skipped in Pass 1.

Passes matter because they protect Pass 2. Candidates who work strictly in question order routinely reach the DI sets with four minutes left, having spent that time on two hard arithmetic questions worth 2 marks, when the sets were worth 10.

The 45-Second Leave-It Rule, and When a Guess Actually Pays

The rule is simple: if 45 seconds have passed and you cannot see the next concrete step — not the answer, just the next step — mark the question for review and move. The justification is arithmetic. In a 20-minute section, six questions each held for 45 seconds and then abandoned consume 270 seconds. That is 270/1200 = 22.5 per cent of the whole section, and at a 34-second budget it is 270 ÷ 34.3 ≈ 7 questions you could have solved instead. Six blanks bought you nothing and cost you seven marks.

Guessing is a separate decision, and it also has an exact answer. If k options remain plausible after elimination, the expected marks from one guess are (1/k) − p(k − 1)/k, which simplifies to [1 − p(k − 1)]/k. At p = 1/4 this gives exactly zero for k = 5, so a blind guess is worthless; +1/16 = 0.0625 for k = 4; +1/6 ≈ 0.167 for k = 3; and +3/8 = 0.375 for k = 2. Elimination, not optimism, is what makes a guess profitable.

Read those numbers as a time cost too. A k = 3 guess is worth about 0.167 marks. A question you can actually solve at 80 per cent accuracy is worth 0.75 marks. So spending 30 seconds to eliminate two options is a bad trade if those 30 seconds would have bought you a whole solvable question. Guess only from eliminations you already have, made on the way past — never go back to manufacture them.

Approximation and option-elimination (Chapter 3) and quadratic sign-reading (Chapter 5) are the two skills that most often collapse five options to two without solving. That is the real reason they are high-yield: they convert the 0.000 line of the table below into the +0.375 line.