• Advanced
  • About 15 min
  • 5 practice questions

Testing attempt strategies across mocks, one change at a time

Run planned experiments on your attempt strategy (section order, number of attempts, time caps) across mocks, change one thing at a time, and judge the result on more than the score.

If something looks wrong, please tell us.

By now your analysis tells you what you get wrong. The next level is to use mocks to decide how you attempt the paper: which section first, how many questions to attempt, when to move on. Most students settle these by habit or by copying someone else. A better way is to treat them as experiments, test them over several mocks, and keep what works for you.

What counts as an attempt strategy

  • Section order: which section you start with and which you leave for last.
  • Attempt target: roughly how many questions you aim to attempt, given your accuracy.
  • Time caps: the maximum time per question on first look, or per section.
  • Rounds: for example, a first pass for questions you can do quickly, then a second pass for the rest.

How to run an experiment

  1. Record a baseline: your usual strategy over two mocks.
  2. Write down one change and what you expect it to improve, for example 'starting with reasoning will raise my reasoning attempts without lowering accuracy'.
  3. Use the new strategy for at least two mocks. Change nothing else.
  4. Compare the score and the inside numbers: attempts and accuracy by section, time per section, number of time-sinks.
  5. Keep the change, drop it, or test it once more if the result is unclear.

Changing two things at once, say section order and time caps, leaves you unable to tell which one helped. And because mock papers vary in difficulty, one mock is never enough to judge a strategy.

Worked example 1: Nikhil tests section order

Nikhil's strategy log
MockStrategyReasoning attemptsReasoning accuracyNet score
8Usual order (reasoning third)1889%112
9Usual order (reasoning third)1883%108
10Reasoning first2286%118
11Reasoning first2286%114

Baseline average = (112 + 108) ÷ 2 = 110. New average = (118 + 114) ÷ 2 = 116, a rise of 6. The inside numbers support it: reasoning attempts went from 18 to 22 with similar accuracy, which is what he predicted, so the gain is not just an easier paper. He adopts reasoning first and tests nothing else for two weeks so the new order settles.

Worked example 2: Divya tests a higher attempt target

Suppose a paper gives +2 for a right answer and deducts 0.5 for a wrong one. Divya usually attempts about 70 questions. In one mock she attempts 70 with 60 right and 10 wrong: net = 120 − 5 = 115. In the next she pushes to 85 attempts with 63 right and 22 wrong: net = 126 − 11 = 115.

  1. Extra attempts: 85 − 70 = 15. Extra correct: 63 − 60 = 3. Extra wrong: 22 − 10 = 12.
  2. Marks from the extra 15: 3 × 2 − 12 × 0.5 = 6 − 6 = 0.
  3. Marginal accuracy (accuracy on the extra questions only) = 3 ÷ 15 = 20%.

Under this assumed scheme, a question answered with a 20% chance of being right gains nothing on average: 0.2 × 2 = 0.4 gained against 0.8 × 0.5 = 0.4 lost. The extra 15 attempts added risk and no marks. Her next experiment is about 78 attempts, chosen more carefully, to see if the marginal accuracy rises. (The arithmetic of guessing under negative marking is covered in the Handling Negative Marking course.)

Change one thing, test it on at least two mocks, and judge it on the inside numbers as well as the score.

Key takeaways

  • Section order, attempt targets, time caps and rounds are strategies you can test.
  • Record a baseline, change one thing, and run it for at least two mocks.
  • Check that the inside numbers moved the way you predicted.
  • Look at marginal accuracy: what the extra attempts actually earned.

Check your understanding

  1. 1. Why should only one part of the attempt strategy be changed at a time?

    1. Otherwise you cannot tell which change caused the result
    2. Because mocks allow only one change
    3. Because two changes always lower the score
    4. Because strategies cannot be combined in the real exam
    Show answer

    Answer: A. With two changes at once, any rise or fall cannot be traced to either one, so you learn nothing reliable.

  2. 2. Baseline nets are 104 and 100; with a new strategy the nets are 109 and 107. What is the rise in average?

    1. 5 marks
    2. 6 marks
    3. 7 marks
    4. 12 marks
    Show answer

    Answer: B. Baseline average = (104 + 100) ÷ 2 = 102. New average = (109 + 107) ÷ 2 = 108. Rise = 108 − 102 = 6.

  3. 3. Under +2 / −0.5, a student's extra 20 attempts gave 6 right and 14 wrong. What did they add to the net score?

    1. 5 marks
    2. 12 marks
    3. 0 marks
    4. −2 marks
    Show answer

    Answer: A. 6 × 2 − 14 × 0.5 = 12 − 7 = 5 marks.

  4. 4. What is 'marginal accuracy' in this lesson?

    1. Your accuracy in the last section only
    2. Your lowest accuracy in any mock
    3. The accuracy on the extra questions attempted beyond your usual number
    4. The average accuracy of all the other test-takers who sat that same mock
    Show answer

    Answer: C. It looks only at the additional attempts, showing whether pushing for more questions actually earned marks.

  5. 5. Nikhil's score rose with a new section order. Why did he also check reasoning attempts and accuracy?

    1. Because the score alone is always wrong
    2. To confirm the gain came from the predicted change, not an easier paper
    3. Because the answer key for the reasoning section of that mock was incomplete
    4. Because rank depends only on reasoning
    Show answer

    Answer: B. If the inside numbers moved as predicted, the gain is more likely to be real rather than caused by paper difficulty.

Privacy · Terms · Refunds · Delivery