GATE Statistics PYQ: Topic-Wise Weightage 2023–2026 and Paper PDFs

The official GATE site gives you the papers and keys. It does not tell you which parts of the syllabus they test. We went through all 220 Statistics questions of the 2023, 2024, 2025 and 2026 papers (55 a year, 85 marks), put each into one of the 12 sections of the 2027 syllabus, and checked each answer against the final key. Three blocks come out on top. The mathematics (calculus and matrix theory) carried a steady 19–22 marks every year. Estimation and testing together carried 14–20, and probability with the two distribution sections 19–26. Everything else shares what is left.

Our classification of the official ST papers and final keys, 2023–2026.

Read the table as ranges rather than targets. The two mathematics sections have not moved: 7 questions of Matrix Theory every single year, and 6 or 7 of Calculus. Testing of Hypotheses has risen, from 5 marks in 2023 to 9 and 10 in 2025 and 2026. Probability is the most volatile, 13 marks in 2023 and only 5 in 2025. Non-parametric Statistics is the smallest section at 3–4 marks, which puts the new run tests and Kruskal–Wallis in perspective.

Negative marking touches only 42–52% of the Statistics marks, the MCQ share. 2025 was the most numerical paper, with 33 NAT marks, but there is no steady drift towards NAT: 2026 went back to 22. MSQ carried 16–25 marks. It has no negative marking, but you must select exactly the correct set of options, so partly right answers score nothing.

These are topics we found in three or more of the four papers, with question numbers so you can go straight to them in the official PDFs. The counts come from a keyword search of our classification notes, so treat them as a floor.

The pattern suggests where to put your hours. Testing of hypotheses, and above all the Neyman–Pearson and UMP family, is the single most repeated idea: nine questions in four papers by our search, five of them in 2026 alone. Estimation’s big three (MLE, sufficiency and Cramér–Rao) appear every year. In stochastic processes, Markov chains, the Poisson process and Brownian motion are almost guaranteed a question each.

The GATE 2027 download page (IIT Madras) carries every ST paper back to its introduction. Each year has its own file, so there is nothing to cut out. One oddity to know about: the 2023 key follows the master version of the paper, so its MCQ answers cluster on A and C. That is the master order, not a pattern candidates saw, since options were shuffled on screen.

When you write the older papers, mark them against the 2027 syllabus rather than the one they were set on. The 2023 paper’s delta-method question (Q61) is now outside the syllabus, while its Laplace questions (Q26 and Q28) are now inside it. Write 2026 last, closest to the exam, because it has the highest top-candidates’ mean of the four and is the best guide to the current setters.