Stop calculating the moment you know
This is a multiple-choice paper, and that changes what a maths question even is. You are not being asked to produce a number. You are being asked to identify one from four. Those are different tasks, and the second is very often much easier — which most candidates never exploit, because they were trained for twelve years to compute the answer and then look for it.
Look at the options before you calculate
The habit worth building, and it inverts what school taught you.
Read the options first. They tell you how much precision the question actually requires — and the answer is usually much less than you assume.
If the options are 12, 47, 280 and 1500, they are orders of magnitude apart, and a rough estimate separates them instantly. There is no reason to compute exactly. If the options are 282, 284, 286 and 288, you need the exact value and you know that before starting. Either way, you have learned something in two seconds that governs the whole approach.
This single habit — options first — routinely turns a ninety-second calculation into a fifteen-second elimination, and it costs nothing. Under 108 seconds, on a paper you cannot revisit, that saved time is not a luxury; it is the marks on the questions after it.
The sanity check that costs two seconds
The second habit is asking whether an answer is physically absurd, and it is worth more here than in any other exam because the subject is architecture.
A door 30 metres tall. A room of 4 square millimetres. A building weighing 2 kilograms. Each of those is arithmetically reachable from a plausible slip — a misplaced decimal, a scale applied twice — and each is instantly wrong to anyone who pauses to picture it.
So picture it. Does this answer describe a thing that could exist? That question takes two seconds and catches the errors that matter most, because decimal-point errors are large and they are exactly what the distractors are made from.
This is also, quietly, what the bulletin means by mathematics and its association with creative thinking. Not maths in a vacuum — maths tied to whether the answer makes sense as a space. That is an architect's instinct and it is being tested, so use it.
Know when to stop, and when to leave
Two decisions, both about time rather than maths.
Stop when the estimate decides. The moment your rough number picks one option, commit. Every further second of refinement buys certainty you already had and costs you a question you have not seen yet. This is the hardest habit here, because finishing the calculation feels responsible. It is not — on a one-way paper it is a transfer of marks from later questions to this one.
Leave when it will not come. Numerical questions are bimodal in the same way patterns are: either the route appears fairly quickly or it does not appear at all. If you are at sixty seconds with no path, eliminate what you can — by order of magnitude, by absurdity, by the scaling law — and commit. Blind guessing is not free (nobody has told us whether wrong answers are penalised), but a reasoned elimination is not blind.
That is the whole of Part B strategy landing on one area again: decide fast, commit, move.
The rules behind this
Sourced to the official brochure rather than restated here, so there is one place to correct when the Council revises it.
Part B allows 108 seconds per question, presented one after another, on an adaptive engine.
90 minutes across 50 questions. The adaptive structure dates to NATA 2025 per the President's foreword in V2.0, which states that NATA 2026 continues it.
Source · verified 2026-07-16
Part B questions appear one after another at 108 seconds each. There is no evidence of a review screen.
The skip-flag-and-return habit that works in JEE does not transfer. Budget the 108 seconds and commit.
Source · verified 2026-07-16
Whether Part B carries negative marking is not stated in any official document.
The widely-repeated "no negative marking" claim appears only on coaching sites. We could not source it to COA.
Read this carefully: This one matters for strategy — whether to guess depends on it. Since it is unpublished, do not build a guessing policy on the assumption that wrong answers are free.
No instruments are permitted — no compass, no set squares — and no calculators, phones, or wet media.
Appendix-II states "Don't bring any instruments". Also barred: Bluetooth devices, slide rules, log tables, electronic watches with calculators, and any textual material. Numerical Ability is examined without a calculator.
Source · verified 2026-07-16
What almost everyone believes
“I should calculate the exact answer and then find it among the options.”
You are being asked to identify an answer, not produce one. Very often an estimate identifies it in a fraction of the time.
Twelve years of school maths trains you to compute and then match, and multiple choice quietly inverts that. Reading the options first tells you how much precision is actually required — usually far less than assumed — and often turns a ninety-second calculation into a fifteen-second elimination. Under a 108-second one-way clock, the time saved is not a nicety; it is the marks on the questions that come after, which you cannot return to.
Depending on how long you have
Foundation
Understand the skill. Months out, or starting from zero.
Build the estimating habit in daily life, where it costs nothing. Guess the bill before it arrives. Guess the height of a building, then check. Guess how many people are in the room. You are training a sense of magnitude, which is what the sanity check runs on and what an architect needs anyway.
Drill
The practice protocol. What to repeat, how often, how to score it.
Read the options first on every single numerical question, without exception, until you stop noticing you are doing it. Then track how often the exact calculation was actually needed. For most people the honest answer is under half — and that gap is free time on a paper where time is the binding constraint.
Exam-Day
What to actually do under the constraint — 108 seconds, no instruments, one pass.
Options first. Estimate. Stop the moment one option is picked out. Ask whether the answer could physically exist. And at sixty seconds with no route, eliminate by magnitude and absurdity, commit, and move — the next question is worth exactly as much as this one and you have not read it yet.
Try it
Fifteen minutes. The discipline is reading the options before the question.
- 01Take any set of numerical MCQs. For each one, read the OPTIONS first and write down how much precision you actually need.
- 02Now answer using the least precision that separates them. Note how often exact calculation was unnecessary.
- 03For each answer, spend two seconds asking: could this thing exist? Flag anything absurd.
- 04Under a 108-second timer, practise committing the instant your estimate picks an option — even though it feels wrong to stop.
- 05Count how many you finished. Then count how many needed exact arithmetic. The gap between those numbers is the time this habit gives you back.
The short version
Multiple choice asks you to identify a number, not produce one, and reading the options first tells you how little precision you actually need — often turning ninety seconds into fifteen. Ask whether the answer could physically exist; a 30-metre door is instantly wrong and decimal errors are exactly what the distractors are built from. Stop the moment your estimate decides, because refining an answer you already have is a transfer of marks to this question from ones you have not read.
That completes Numerical Ability. Take it to the mock, where the clock is real and there is no calculator.
Questions people actually ask
- Do I need to calculate exact answers in NATA?
- Often not. It is a multiple-choice paper, so you are identifying an answer rather than producing one. Read the options first: if they are orders of magnitude apart, a rough estimate separates them in seconds. Precision beyond the point where one option is picked out is time stolen from later questions you cannot return to.
- What is a sanity check in a maths question?
- Asking whether the answer describes something that could exist. A 30-metre door, a 4-square-millimetre room, a 2-kilogram building — each is arithmetically reachable from a misplaced decimal and instantly wrong to anyone who pictures it. It takes two seconds and catches the largest errors, which is what the distractors are made from.
- Should I guess if I cannot work out a numerical question?
- Eliminate rather than guess blindly — by order of magnitude, by physical absurdity, by the scaling law. Whether wrong answers are penalised is not published anywhere, so do not assume guessing is free. A reasoned elimination is not a blind guess, and at sixty seconds with no route it is the right move.