GATE Statistics 2027: An 18-Week Study Plan, October to February

GATE 2027 runs over six days, 6, 7, 13, 14, 20 and 21 February 2027. The ST paper’s day is not yet published; in 2026 it was on Sunday 8 February, in the forenoon. From 1 October that leaves about 18 weeks. Registration closes on 5 October 2026 (12 October with the late fee), so do that first. This plan gives your hours to each section in proportion to the marks it has carried, puts the steadiest marks first, and ends with the five official papers.

GATE 2027 brochure and important-dates page; section averages and marks targets are our arithmetic on the official papers and reports.

Pick the programme you are aiming at, then work back to marks. The table uses the 2026 field, the hardest of the four years to score on. Aim for the marks in the right-hand column in your timed full papers from January onwards.

A realistic reading: 35–40 marks puts you among the top 200 of about 2,100, and 50+ marks puts you in the top 30. In 2026 only 75 candidates who appeared reached 40 marks and 31 reached 50, by our count of the official table. This is a paper where a disciplined plan beats a large field.

Start where the marks are steadiest. Matrix Theory gave 7 questions every year and Calculus 6 or 7. Do rank and nullity, systems of equations, eigenvalues (asked every year), Cayley–Hamilton, diagonalisation, definite matrices and quadratic forms. For calculus, cover sequences and series (asked every year), continuity and differentiability, Taylor’s theorem, maxima and minima with Lagrange multipliers, and double integrals, which you will need again for joint distributions. Alongside, begin Probability: axioms, conditional probability and Bayes, random variables, expectation and variance (with the properties the 2027 syllabus now names), the MGF and the new PGF.

Work through the univariate distributions, including the two newly named ones, Laplace and Cauchy, and both kinds of beta. Then joint distributions: marginals and conditionals, conditional expectation, transformations, order statistics, the bivariate normal and the χ², t and F distributions. Finish with convergence: the four modes, Slutsky, Borel–Cantelli, the laws of large numbers and the CLT. The delta method is out of the 2027 syllabus; an hour on it is still worth it, as it follows from Taylor and the CLT.

This is the heart of the paper and of every recurring topic list. Estimation: sufficiency and the factorisation theorem, minimal sufficiency, completeness, Basu, UMVUE through Rao–Blackwell and Lehmann–Scheffé, the Cramér–Rao bound, method of moments and MLE, pivotal intervals. Testing: the Neyman–Pearson lemma, MLR, UMP and UMPU tests and the likelihood ratio test. Neyman–Pearson and UMP questions appeared nine times in four papers by our search, five of them in 2026. Give these four weeks your best hours.

Now the four sections that carry 4–7 marks each. Stochastic processes: Markov chains (classification and stationary distributions), the Poisson process and Brownian motion, each asked in every paper. Multivariate: the multivariate normal, MLEs of the mean vector and dispersion matrix, Hotelling’s T² and Wishart. Regression: simple and multiple, R², Gauss–Markov and tests on coefficients; add ridge regression, which was asked in 2025 although not named. Non-parametric: the usual tests plus the new run tests and Kruskal–Wallis.

Write the five official papers (2022–2026) under exam conditions, three hours each, and mark each against the final key. Our previous-papers guide lists the files. Start from week 14 with one paper a week, then two a week in the last fortnight. After each one, sort every wrong answer by section and go back to that section the next day. Convert your mark into a score on the 2026 field with the calculator in our marks-vs-score guide, and keep the last three days for revision and sleep, not new topics.

Four habits cost ST candidates the most. The first is guessing two-mark MCQs. MCQs carry −2/3 on a two-mark question, and they made up 36–44 of the 85 Statistics marks, so skip one you cannot narrow to two options. MSQ and NAT answers cost nothing when wrong. The second is half-answering MSQs: an MSQ scores only if you select exactly the right set, so check every option rather than stopping at the first correct one. The third is slow arithmetic on NATs, which carried 20–33 marks a year; practise with the GATE on-screen calculator, not a handheld one. The fourth is leaving General Aptitude to the end. It is 15 marks shared with every GATE paper, and the easiest marks on the paper to secure.