Negative marking and attempt strategy in AFCAT

AFCAT awards 3 marks for a correct answer and deducts 1 mark for a wrong one. Unanswered questions score zero. That is the scheme in the official notification, and it has a simple consequence many candidates never work out: a wrong answer does not cost you one mark, it costs you four, compared with getting it right.

Understanding the arithmetic takes ten minutes. It changes how you guess, how many questions you attempt and how you use the last twenty minutes of the paper.

The arithmetic of guessing

Each question has four options. If you guess, your expected marks depend on how many options you can rule out first.

The break-even point. If p is your chance of being right, a guess is worth 3p − 1(1 − p) = 4p − 1 marks on average. This is positive only when p is more than 1/4. In plain words, a guess helps only when you are genuinely better than random.

Conclusion: a blind guess is, on average, worth nothing and adds risk. A guess after eliminating even one option has a positive expected value. After eliminating two, it is clearly worth taking.

Why accuracy matters more than attempts

Every wrong answer that becomes a right one gains 4 marks: the 3 you earn and the 1 you no longer lose. Every wrong answer that becomes a blank gains 1. So the fastest way to raise your score is usually to cut errors, not to attempt more.

These figures are made up, for practice.

Candidate B attempts 15 fewer questions than A but scores 43 marks more. Candidate C is even more accurate than B but attempts too few to catch up. The best position is B's: a high attempt count and high accuracy. Leaving questions blank is not the goal; avoiding careless errors is.

A useful formula. If you attempt N questions with accuracy a (as a decimal), your score is N × (4a − 1). At 85% accuracy, each attempt is worth 4 × 0.85 − 1 = 2.4 marks on average. At 70%, it is worth only 1.8.

Moving from 70% to 85% accuracy on the same 80 attempts adds 48 marks, more than adding 20 extra attempts at 70% would (20 × 1.8 = 36).

How many questions should you attempt?

There is no fixed right number. It depends on your accuracy, and your accuracy is something you measure in mocks, not guess on exam day.

1. From your last five mocks, work out your accuracy on questions you answered with confidence, and separately on questions you guessed after eliminating options.
2. Attempt every question you are confident about.
3. Add eliminated-option guesses only where your mock data shows they pay, which they usually do after removing two options.
4. Skip blind guesses.

Your target score then decides whether you need more attempts or better accuracy. See AFCAT cut-offs and merit for how to set that target.

A two-round method for the paper

Round 1 (about 75 minutes). Go through the whole paper once in your chosen order. Answer every question you are sure of. Mark doubtful questions for review. Skip anything that needs a long calculation or a second reading of a long passage. Do not let any single question hold you for more than about 90 seconds in this round.

Round 2 (about 35 minutes). Return to the marked questions. Solve the long ones you skipped, starting with the ones you know how to do. For the rest, answer only where you can eliminate at least one or two options for a clear reason.

Last 10 minutes. Check that you have not left any easy question unanswered by mistake. Check that marked-for-review questions have an option actually selected if you meant to answer them. Do not change answers on a hunch; change them only if you spot a definite error.

Section order

Many candidates start with General Awareness, because it is fast: you either know the answer or you do not. Then comes English, then Reasoning, and they finish with Numerical Ability, where time is most flexible. Others prefer to do Reasoning first while fresh.

There is no best order for everyone. Try two or three orders across your mocks, record the score and time for each, and keep the one that works. Our guide to mock tests shows how to record this.

Practice set

1. A candidate answers 68 correctly and 14 wrongly, and leaves the rest. What is the score?
2. With 85% accuracy, how many attempts are needed to score 180 on average?
3. A candidate has 12 doubtful questions and can eliminate two options in each. What is the expected gain from guessing all 12?
4. In a mock, a candidate had 20 wrong answers. Analysis shows 6 were careless errors on questions they knew. If those 6 had been right, how much higher would the score be?
5. A candidate eliminates one option and guesses. If they do this for 9 questions, what is the expected total?

Answers:

1. 68 × 3 = 204, and 204 − 14 = 190.
2. Each attempt is worth 4 × 0.85 − 1 = 2.4 marks. 180 ÷ 2.4 = 75 attempts.
3. Each guess is worth +1 on average, so the expected gain is +12.
4. Each wrong-to-right change gains 4 marks: 6 × 4 = 24 marks.
5. Each is worth +1/3 on average: 9 × 1/3 = +3.

What to do next

• In your next mock, mark every guess with a symbol on rough paper: "B" for blind, "E1" or "E2" for one or two options eliminated.
• After the mock, calculate your accuracy for each type.
• Stop blind guessing if your data confirms it earns nothing.
• Practise the two-round method in three consecutive mocks before deciding whether to keep it.
• Try two different section orders and keep the better one.