Data study · every CAT paper, 2017–2025

CAT, decoded by the numbers.

We analysed all 23 modern CAT papers — question by question — to map exactly how the exam has shifted: section weights, the quiet rise of negative marking, and the topic-by-topic weightage of QA, DILR & VARC. Then we turned it into a plan to maximise your percentile.

23 papers analysed
9 years (2017–25)
1,400+ questions mapped
3 sections, topic-level
The 6 findings that change how you prep

What 9 years of CAT actually tells you

24 / 22 / 20
The new section split
From 2024, DILR (22) overtook QA (20). Plan for this, not the old 24/20/22.
~50%
of DILR is now TITA
~11 of 22 DILR questions carry no negative marking (2024–25). A gift — attack it.
4 → 5
DILR sets per paper
More sets = more choice. Selecting the right 3 is now the core DILR skill.
~92%
of VARC is MCQ
VARC TITA collapsed to 2–4. Accuracy is everything; a blind guess costs you.
~40%
of QA is Arithmetic
Every single year. Master Arithmetic and half of QA is already yours.
4 × 4
VARC RC, unchanged 5 yrs
4 passages × 4 questions + 8 VA. The most predictable block in the exam.
01 · Format evolution

The exam keeps shrinking — and just re-balanced

CAT went from a 100-question, 3-hour marathon to a tight 66-question, 2-hour sprint with 40-minute sectional locks. In 2024 it also quietly swapped the weights of its two hardest sections.

Section split over the years (VARC / DILR / QA)

Each bar = one era's paper. Notice DILR (gold) grow past QA (green) from 2024.

2017–19
VARC 34DILR 32QA 34
2020
262426
2021–23
242022
2024–25
242220
VARCDILRQA
EraYearsQuestionsTimeSectional lockV / DILR / QA
Classic2017–2019100180 min60 min34 / 32 / 34
Covid202076120 min40 min26 / 24 / 26
Current‑A2021–202366120 min40 min24 / 20 / 22
Current‑B2024–202566120 min40 min24 / 22 / 20
What to do: Build your test temperament around 66 questions in 3 locked 40-minute windows, and treat DILR as a 22-question section — it is now bigger than QA.
02 · The negative-marking revolution

Where the −1 penalty hides has completely shifted

Overall ~27–33% of questions stay TITA (type-in-the-answer, no negative marking). But which section those safe questions sit in has flipped — and it changes your entire attempt strategy.

TITA questions per section, by year

Taller = more no-negative-marking questions. VARC (blue) fell to almost nothing; DILR (gold) doubled.

2017
2018
2019
2020
2021
2022
2023
2024
2025
VARC TITADILR TITAQA TITA
DILR — attack freely. ~10–11 of 22 questions have no negative marking. Once you commit to a set, answer every TITA — even an educated guess is free upside.
VARC — protect your score. Only 2–4 TITA means ~20+ questions carry −1. Never blind-guess; skip the 1–2 hardest inference questions instead.
03 · VARC deep-dive

The most predictable section in the exam

For five straight years, VARC has been the same shape: 4 reading passages × 4 questions (16) + 8 Verbal Ability questions. Predictability is an edge — you always know exactly what's coming.

Composition · 24 questions

Reading is ⅔ of your VARC score

Reading Comprehension (4 passages)16
Verbal Ability8

RC passages are abstract prose — economics, sociology, philosophy, science, history, arts — never current affairs. The skill is dense inference, not knowledge.

Verbal Ability · the 8 questions

What the VA 8 actually are

  • Para-summary — ~3 (MCQ). The reliable, scoreable ones.
  • Para-jumble — ~2–3 (TITA). Order the 4 sentences.
  • Odd sentence out — ~2 (TITA). Remove the misfit.
  • Para-completionextinct 0 across 2021–25.

VA turned more MCQ in 2024–25 (TITA fell to 2–4), so guesswork now costs you here too.

Strategy: Do all 4 RCs first (16 marks, highest value), aim 85%+ accuracy. In VA, bank the para-summaries; attempt jumbles/odd-out (TITA — no risk); skip, don't guess, the MCQs you're unsure of. Target ~18–20 attempts, not 24.
04 · DILR deep-dive

The section that now decides your percentile

DILR grew (20→22), gained a set (4→5), and became ~50% penalty-free. It is the most volatile section — which means it's also the biggest opportunity to leap from a 90 to a 98+ percentile.

Structure shift

From 4 tidy sets to 5 sharper ones

YearsSetsQuestionsTITA
2021420~5
2022420~6
2023420~8
2024522~10
2025522~11
Set types (2023–25)

Two families, roughly 60/40

  • Logic puzzles — seating/arrangement, tournaments & games, grouping, truth-teller/tapping scenarios.
  • Data Interpretation — and increasingly novel charts: candlestick plots, GDP tables, nutrient grids.
  • Rising: quant-reasoning sets — tariffs, cross-border trade, currency exchange, ratings systems.
Strategy: Spend the first 3–4 minutes reading all 5 sets and ranking them — set selection is the whole game. Commit to 3 sets, go deep, and farm every TITA (no negative marking). Leave the ugliest 2 sets untouched. Target 14–16 accurate attempts.
05 · QA deep-dive

Arithmetic wins the Quant section

Across every year, Arithmetic is ~40% of QA — more than Algebra and Geometry combined. It's also the fastest to solve. A student strong in Arithmetic clears the QA cut-off almost on that alone.

QA by topic family (5-year average, % of the section)

Approximate — from our classification of solved 2021–25 papers. The ranking is stable every year.

Arithmetic core~40%
Algebra~20%
Geometry & Mensuration~14%
Number System~14%
Modern Math (P&C · Prob · Progressions)~10%

Topic-by-topic weightage & priority

Share = approximate % of the QA section. Priority tiers tell you where to spend your prep hours.

TopicShare of QAPriority
ARITHMETIC · ~40%
Percentages~8%Tier 1
Profit, Loss & Discount~7%Tier 1
Time, Speed & Distance~6%Tier 1
Averages & Mixtures~7%Tier 1
Ratio & Proportion~4%Tier 1
Time & Work~4%Tier 1
Simple & Compound Interest~3%Tier 1
ALGEBRA · ~20%
Linear & Quadratic Equations~8%Tier 1
Inequalities & Maxima–Minima~5%Tier 2
Logarithms~4%Tier 2
Functions~3%Tier 2
GEOMETRY & MENSURATION · ~14%
Triangles~6%Tier 2
Mensuration (area/volume)~5%Tier 2
Circles~2%Tier 2
Coordinate Geometry~1%Tier 3
NUMBER SYSTEM · ~14%
Factors, HCF & LCM~5%Tier 2
Digits & Integer properties~5%Tier 2
Remainders & Divisibility~4%Tier 2
MODERN MATH · ~10%
Progressions & Series~5%Tier 2
Permutation & Combination~3%Tier 3
Probability & Set Theory~2%Tier 3
Strategy: Make Arithmetic + Linear/Quadratic Equations bulletproof (Tier 1 ≈ 55–60% of QA). Add Geometry, Number System & Progressions (Tier 2). Do Logs, Functions, Coordinate, P&C & Probability last — small footprint, high time-cost. With QA now more MCQ, prize ~13–15 clean attempts at 90% over volume.
06 · The game plan

Your CAT 2026 blueprint, on one screen

SectionAttempt targetAccuracyThe one move that matters
VARC (24)18–2085%+All 4 RCs first; skip — never guess — hard MCQs
DILR (22)14–16high on TITARank all 5 sets, commit to 3, farm the ~11 penalty-free TITA
QA (20)13–15~90%Arithmetic → Equations first; don't blind-guess MCQs
1
DILR is the swing section
Bigger, more sets, half penalty-free. Percentiles are won here.
2
VARC rewards accuracy
Predictable shape, ~92% MCQ. Reward precision, punish guessing.
3
QA rewards Arithmetic
~40% of the section, every year. Master it, select ruthlessly.

Train on the exam you just read about.

30 full-length CAT mocks built to this exact blueprint — the real TCS-iON interface, 40-minute sectional locks, and deep analysis that shows your leaks by topic. Take a full one free, no signup.

Try a full mock — free →
Methodology & honesty

How this was built

We processed all 23 modern CAT papers (2017–2025, all slots). Structural figures — section splits, RC/DILR set counts, and MCQ vs TITA counts — are extracted directly from every paper and are exact. Topic weightage is our own classification of the solved papers; family-level shares (Arithmetic, Algebra, etc.) are reliable, while individual topic percentages are close approximations (±1–2%) and can vary a little slot to slot. Use them for priority, not to the decimal. This is analysis and guidance, not an official document; CAT is conducted by the IIMs.