Skip to content
Get started

Why NeetQuant · Standards

A wrong answer key teaches you something false. So we test ours.

How the 1,456 questions are written, what every one of them includes, and the automated checks that run before anything reaches you. A failed check stops the release - it is not a warning someone can ignore.

What every question has

An original prompt. Written for practice, in the style of one or more interview formats, as described in public candidate reports. Never copied from a real interview, and never labelled as coming from a firm (our content policy explains why).

Two or three progressive hints. The first points you towards the idea; the last usually leaves only the execution. You take one at a time, so you get exactly as much help as you needed.

A complete worked solution, typeset in LaTeX, that shows the reasoning and not just the number - often with the shortcut an interviewer would hope to see.

Tags for topic, difficulty, the formats it resembles and the roles that meet it - which is what lets tracks, analytics and plans pick the right question for you.

Variety over volume. A bank padded with the same problem in different numbers looks big and teaches little. New questions are judged by their key idea, and the duplicate check below backs that up.

A grader that accepts right answers

A grader that only accepts one spelling marks correct candidates wrong, and you stop trusting it. Ours accepts the forms a right answer is naturally written in and rejects near-misses. The table below is not a mock-up: it is produced by the production grader each time this page is built.

Correct answerYou type → verdict
17/3 ≈ 5.6667
  • 17/3 ✓
  • 5.667 ✓
  • 5.7 ✓
  • 6 ✕
  • 4.53 ✕
0.4
  • 2/5 ✓
  • 40% ✓
  • .4 ✓
  • 0.32 ✕
  • 0.5 ✕
1000
  • 1,000 ✓
  • 1000.0 ✓
  • 999 ✕

Why is 5.7 right but 6 wrong? A rounding to two significant figures shows you computed the value; a whole number for a non-whole answer is the classic trap, so it must be exact.

Try the graderLive

Expected sum given a six

Two fair dice are rolled. Given that at least one of them shows a 6, what is the expected value of their sum? Give a decimal to four places.

Or try: · ·

Runs the grading function the server uses, in your browser - nothing is sent. See the question with its hints.

The checks that gate every release

These run automatically before every build of the site. The build - and so the release - stops if any one fails.

01
Structure
Malformed questions, answers that cannot be graded, broken tags, duplicate slugs and near-identical titles.
02
Answer gradability
Replays every numeric answer through the real grader in the spellings people type - the key, a trailing zero, correct roundings, the exact fraction, a percent, comma-grouped digits - and confirms each is accepted while decoys 20% off are rejected. (It proves a key is gradable, not that the key is right.)
03
Duplicate detection
Compares every question's prompt with every other's, with numbers masked, so the same problem with new numbers is caught. Pairs a person has judged genuinely different are recorded.
04
Formula rendering
Renders every prompt, hint, solution and option through the same maths pipeline the app uses, and fails on any formula that would show up broken.
05
Probability OA by simulation
For every generated OA question type, reads the numbers back out of the question text and simulates the scenario independently. The marked answer must match the simulation.
06
Drill answers re-derived
For every generated arithmetic item, recomputes the answer from the prompt the candidate sees, so a correct answer can never be marked wrong.
07
Simulator variety
Simulates many runs and fails if any single question template dominates a run.
08
Game invariants
Drives the game engines over many random seeds: no hidden information leaks, scores stay in bounds and deterministic, and - for Kelly Betting and Make a Market - optimal play earns a rating near 100 regardless of luck.
09
Links
Fails on any internal link to a page that doesn't exist or only works through a redirect.

Common questions

Are NeetQuant's questions real interview questions?
No. They are original practice problems written in the style of each interview format. We never publish questions taken from real interviews, and we never ask candidates for them.
What happens if I find a mistake?
Tell us through the contact page. A confirmed error is fixed in the question bank itself, and because the checks run on every release, the fix is verified before it goes live.
Why do some questions have no answer box?
15 of 1,456 ask for a proof or an explanation, which a machine cannot fairly grade. For those you compare your reasoning with the worked solution.

Try a question and see for yourself

Prompts and hints are public. Create a free account to check answers and read every worked solution.

Read next