IBPS Clerk (CRP CSA-XVI) Pattern for Reasoning and English: What Is Tested and How to Play It
This chapter sets out the verified structure of the IBPS Clerk (CRP CSA-XVI) Prelims and Main for the two sections this book covers, then turns it into time, marks and risk arithmetic. It shows how much each section is worth, how wrong answers are penalised, and how to plan a separately timed section before you learn any topic. Later chapters teach the topics themselves.
KarmSakha preparation topics covered
- CRP CSA-XVI process, tiers and tentative dates
- Prelims tests, marks and sectional timing
- Main tests, marks and the absence of Computer Aptitude
- Seconds-per-question budgets for Prelims and Main
- Negative marking and derived Main deductions
- Break-even accuracy and expected value of guessing
- Eligibility, age relaxation and language of the test
- Topic-frequency evidence from Prelims 2026 and Mains 2025
- Triage, section time plans and error logging
The CRP CSA-XVI at a glance: two tiers, one merit list
The IBPS Clerk examination is formally the Common Recruitment Process for Customer Service Associates, abbreviated CRP CSA, and the cycle this book is built around is CRP CSA-XVI, which fills clerical-cadre vacancies of 2027-28 in participating public sector banks. Customer Service Associate (CSA) is the current name of the post that candidates still call the bank clerk. The detailed notification is dated 01.08.2026, and registration with fee payment ran from 01.08.2026 to 21.08.2026. A corrigendum dated 29.09.2026 changed only Annexure I, the indicative vacancy table, and left every other term untouched, so the exam structure described in this chapter was not altered by it.
The process has two online tiers: an Online Preliminary Examination and an Online Main Examination. Candidates who qualify in the Preliminary and are shortlisted sit the Main. The decisive fact for planning is that the two tiers play different roles. The Preliminary is a filter: its marks are used only to shortlist, and it is passed by clearing each of its three tests. The Main is the ranking stage: as per the notification, only Main marks are considered for the final merit list. A candidate who scrapes through Prelims with a comfortable margin gets no head start, which is why this book trains speed and accuracy for both tiers rather than treating Prelims as a warm-up.
| Item | What the CRP CSA-XVI notification says | Why it matters for Reasoning and English |
|---|---|---|
| Post and cycle | Customer Service Associates, vacancies of 2027-28 | The same pattern is the best guide for the next cycle, but re-check ibps.in |
| Tiers | Online Preliminary, then Online Main | Two separate sets of timing and penalty decisions |
| Notification date | 01.08.2026; registration 01.08.2026 to 21.08.2026 | Window closed; use it to check which edition of the pattern you are reading |
| Tentative dates | Prelims 10 and 11 October 2026; Main 27 December 2026 (IBPS calendar) | All dates are labelled tentative by IBPS |
| Corrigendum | 29.09.2026, Annexure I (indicative vacancies) only | Pattern unchanged |
| Merit | Final merit from Main marks alone | Main Reasoning and English decide ranks |
Prelims: three separately timed tests and what they leave you per question
The Preliminary Examination is objective and has three tests, each with its own clock: English Language with 30 questions and 30 marks in 20 minutes; Numerical Ability with 35 questions and 35 marks in 20 minutes; and Reasoning Ability with 35 questions and 35 marks in 20 minutes. The total is 100 questions, 100 marks and 60 minutes. Because the timing is sectional, a minute you save in Reasoning cannot be spent on English. When the 20 minutes for a test end, the test closes. Every Prelims strategy therefore starts inside a single test, and the question becomes how to turn 1,200 seconds into the best net score.
Dividing the time gives the average budget. English: 20 minutes is 1,200 seconds across 30 questions, which is 40 seconds a question. Reasoning: 1,200 seconds across 35 questions is about 34.3 seconds a question. Those averages hide a sharp spread. A series or inequality question can be finished in 20 to 30 seconds by a trained candidate, while a five-question seating or floor puzzle may consume three to four minutes for the whole set, that is 40 to 50 seconds per question even when you are fast. The average is therefore a ceiling for the whole test, not a target for each question, and Chapter 2 and Chapter 3 teach the set-up methods that make the longer sets fit.
English behaves differently because its time is spent reading rather than deducing. A Reading Comprehension passage with about ten questions may take six to seven minutes, which leaves roughly 13 minutes for the other 20 questions, close to 40 seconds each. Grammar-based formats such as error detection and word swap are quick when your rule base is solid, and slow when you are guessing between options; that is the reason Chapter 9 teaches rules rather than tricks. English is also the only Prelims test offered in English alone, whereas Numerical Ability and Reasoning are available in the state or UT version languages given in the notification.
| Prelims test | Questions | Marks | Time | Average seconds per question |
|---|---|---|---|---|
| English Language | 30 | 30 | 20 min | 40.0 |
| Numerical Ability | 35 | 35 | 20 min | 34.3 |
| Reasoning Ability | 35 | 35 | 20 min | 34.3 |
| Total | 100 | 100 | 60 min | 36.0 |
Main: four tests, and where Reasoning and English sit
The Online Main is where marks start to count for merit. As per the CRP CSA-XVI notification it has four objective tests that are separately timed and not in a fixed sequence: General/Financial Awareness with 40 questions and 50 marks in 20 minutes; General English with 40 questions and 40 marks in 35 minutes; Reasoning Ability with 40 questions and 60 marks in 35 minutes; and Quantitative Aptitude with 40 questions and 50 marks in 35 minutes. The total is 160 questions, 200 marks and 125 minutes. The notification lists no Computer Aptitude test in this Main, which is a change that older guides still get wrong.
The mark weights show why this book pairs the two sections. Reasoning carries 60 of 200 marks, which is 30 percent of the Main, the largest share of any test. General English carries 40 marks, 20 percent. Together they are 100 of 200 marks, exactly half of the Main, on 80 of 160 questions. Reasoning is also the most generously rewarded per question: 60 marks over 40 questions works out to 1.5 marks per question if the questions are equally weighted, against 1.0 in English, 1.25 in Quantitative Aptitude and 1.25 in General/Financial Awareness. A 2025 coaching analysis of the older Main reported puzzle questions at 2 marks each and the remaining Reasoning items at 1 mark each, so real per-question marks, and therefore the penalty per wrong answer, may differ from the equal-weight figures used in these worked examples. The notification gives section marks, not per-question marks, so treat the 1.5 and 1.0 figures as derived working, not quoted rules.
The time per question also differs from Prelims. Main Reasoning and English each give 35 minutes, that is 2,100 seconds, across 40 questions, which is 52.5 seconds a question. That is about 18 seconds more per Reasoning question than Prelims allows, yet Main puzzles are longer and carry more variables, and the paper includes a block of statement-based verbal reasoning. English in Main has 40 questions instead of 30, and a 2025 coaching analysis of the older Main reports about 12 Reading Comprehension questions, so passage reading is the largest single block of that test. Practise Main sets under the 52.5-second average, and Prelims sets under 34 seconds, so that neither clock surprises you.
| Main test | Questions | Marks | Time | Marks per question (derived) | Seconds per question |
|---|---|---|---|---|---|
| General/Financial Awareness | 40 | 50 | 20 min | 1.25 | 30.0 |
| General English | 40 | 40 | 35 min | 1.00 | 52.5 |
| Reasoning Ability | 40 | 60 | 35 min | 1.50 | 52.5 |
| Quantitative Aptitude | 40 | 50 | 35 min | 1.25 | 52.5 |
| Total | 160 | 200 | 125 min | 1.25 average | 46.9 |
Negative marking arithmetic and when guessing pays
The notification states that for each wrong answer one fourth of the marks assigned to that question is deducted, and that a blank carries no penalty. In Prelims every question is worth one mark, so each wrong answer costs 0.25. In Main the deduction scales with the question's marks. If Reasoning questions are equally weighted at 1.5 marks, a wrong answer would cost 0.375; for English at 1.0 it would be 0.25. That scaling is our inference from the stated rule and section marks, so keep it flagged as derived. Many coaching summaries state a flat 0.25-mark deduction in Main; the notification wording (one fourth of the marks assigned to the question) is the authority, so check Clause E on ibps.in. The net score formula is simple: net = (correct × marks per question) − (wrong × 0.25 × marks per question).
Because the penalty is a fixed fraction of the question's value, the break-even point does not depend on the stage. Suppose a question is worth m marks and you answer it correctly with probability p. Your expected gain is p × m − (1 − p) × 0.25m. Setting that to zero gives p = 0.25 ÷ 1.25 = 0.20. So the break-even accuracy is 20 percent in both Prelims and Main. With five answer options, which recent IBPS clerk papers have used, a pure blind guess has exactly 20 percent chance, so its expected value is zero. Blind guessing neither helps nor hurts on average; it only adds variance and consumes time that a better question could use.
Elimination changes the sum. If you can discard two options for certain, three remain and p = 1/3, so the expected gain per Prelims question is 1/3 − (2/3 × 0.25) = 0.167 marks. Discarding only one option leaves p = 0.25 and an expected gain of 0.0625. These gains are small, and each of them costs time, so guessing is a tool for the last half minute of a section, not a plan for a section. The larger lesson is that a wrong answer costs a quarter of a right one: ten careful attempts at 90 percent accuracy beat fourteen hurried attempts at 65 percent, as the table below illustrates with Reasoning-test scenarios: the careful candidate who attempts 28 questions and gets 26 right nets 25.50, while the fast candidate who attempts 35 and gets 25 right nets only 22.50.
| Prelims Reasoning (35 Q) scenario | Attempted | Correct | Wrong | Net marks |
|---|---|---|---|---|
| Careful | 28 | 26 | 2 | 25.50 |
| Balanced | 30 | 28 | 2 | 27.50 |
| Fast, loose | 35 | 25 | 10 | 22.50 |
| Fast, very loose | 32 | 24 | 8 | 22.00 |
| Very safe | 26 | 24 | 2 | 23.50 |
Eligibility and language details that shape preparation
Eligibility rules rarely change how you study, but a few of them change what you should practise. The notification requires a graduation degree in any discipline from a recognised university, with the result declared on or before 21.08.2026, and an age of 20 to 28 years as on 01.08.2026, meaning a General candidate born between 02.08.1998 and 01.08.2006 inclusive. Relaxations are 5 years for SC and ST, 3 years for OBC (non-creamy layer) and 10 years for persons with benchmark disabilities, with other categories as listed in the notification. Computer literacy is also required, shown by a certificate, diploma or degree, or by having studied Computer or IT at school or college.
