Revision notes

Handling Negative Marking: Practice

The key points, rules to remember, tables and diagrams from all 9 lessons, on one page for quick revision. To keep a copy, use your browser's Print and choose "Save as PDF".

Lesson 1

Expected value of a guess: the arithmetic step by step

Expected value of an attempt = (chance of being right x marks for right) - (chance of being wrong x penalty).

Four options, no option ruled out
Scheme (illustrative)WorkingExpected value per blind guess
+1 / -1/31/4 x 1 - 3/4 x 1/3 = 1/4 - 1/40
+1 / -1/41/4 x 1 - 3/4 x 1/4 = 4/16 - 3/16+1/16 (0.0625)
+2 / -0.51/4 x 2 - 3/4 x 0.5 = 0.5 - 0.375+0.125
+1 / -1/21/4 x 1 - 3/4 x 1/2 = 2/8 - 3/8-1/8 (-0.125)
+1 / -11/4 x 1 - 3/4 x 1 = 1/4 - 3/4-0.5
Pallavi's 16 blind guesses (+2 / -0.5, illustrative)
Right out of 16WrongMarks earnedPenaltyResult
21447-3
41286+2
610125+7
  • Expected value = chance right x mark - chance wrong x penalty; a blank is worth exactly 0.
  • Under an illustrative +1 / -1/3 scheme, a four-option blind guess is worth 0; under +1 / -1/2 it is worth -1/8.
  • A blind guess breaks even when the penalty is 1/(k - 1) of the mark, for k options.
  • Even a positive expected value swings; spend time on checking before guessing.
Read the full lesson

Lesson 2

Eliminating options: how ruling out one or two changes the odds

With k options left and one of them right: expected value = (1/k) x mark - ((k - 1)/k) x penalty.

Expected value per attempt after eliminating options
Scheme (illustrative)4 left (blind)3 left (one eliminated)2 left (two eliminated)
+1 / -1/31/4 - 3/4 x 1/3 = 01/3 - 2/3 x 1/3 = 1/3 - 2/9 = 1/91/2 - 1/2 x 1/3 = 1/2 - 1/6 = 1/3
+1 / -1/41/4 - 3/16 = 1/161/3 - 2/3 x 1/4 = 1/3 - 1/6 = 1/61/2 - 1/8 = 3/8
+2 / -0.50.5 - 0.375 = 0.1252/3 - 2/3 x 0.5 = 2/3 - 1/3 = 1/31 - 0.25 = 0.75
+1 / -1/21/4 - 3/8 = -1/81/3 - 2/3 x 1/2 = 1/3 - 1/3 = 01/2 - 1/4 = 1/4
+1 / -11/4 - 3/4 = -1/21/3 - 2/3 = -1/31/2 - 1/2 = 0
  • Expected value with k options left = (1/k) x mark - ((k - 1)/k) x penalty.
  • Under +1 / -1/3: 0 blind, 1/9 after one elimination, 1/3 after two.
  • Harsher penalties need more eliminations before an attempt pays.
  • A wrong elimination drops your chance on that question to zero, so measure how reliable yours are.
Read the full lesson

Lesson 3

Break-even accuracy: the minimum hit rate that makes an attempt worthwhile

Break-even accuracy = penalty / (mark + penalty). Attempt a group of questions on average only if your accuracy on that group is above it.

Break-even accuracy for illustrative schemes
Scheme (illustrative)WorkingBreak-even accuracyWrong answers that cancel one right
+1 / -1/3(1/3) / (4/3)1/4 = 25%3
+1 / -1/4(1/4) / (5/4)1/5 = 20%4
+2 / -0.50.5 / 2.51/5 = 20%4
+1 / -1/2(1/2) / (3/2)1/3 = about 33.3%2
+1 / -11 / 21/2 = 50%1
  • Break-even accuracy = penalty / (mark + penalty).
  • Only the ratio of penalty to mark matters, so a fraction-based penalty gives one break-even point for all question values.
  • Compare it with your measured accuracy on a group of similar questions.
  • Close to the break-even point, an attempt gains almost nothing; prefer time for checking.
Read the full lesson

Lesson 4

Attempts versus accuracy: finding your most profitable number of attempts

Keep adding attempts while the marginal accuracy of the extra questions stays clearly above your break-even accuracy. Stop when it reaches it.

Bhavna's ladder (+1 / -1/3, illustrative)
Group of attemptsQuestionsRightWrongMarginal accuracyMarks from this groupRunning total
First6057395%57 - 1 = 5656
Next159660%9 - 2 = 763
Next103730%3 - 2.33 = 0.6763.67
Last1531220%3 - 4 = -162.67
More attempts help only when accuracy holds
StudentAttemptedRightWrongAccuracyNet (+1 / -1/3)
Simran7263987.5%63 - 3 = 60
Rohit906624about 73.3%66 - 8 = 58
Kavita90721880%72 - 6 = 66
  • Marginal accuracy is the accuracy of the extra questions you add, and it falls as you attempt more.
  • Add attempts while marginal accuracy is clearly above break-even; stop near it.
  • Rushing for attempts can raise careless errors on easy questions too.
  • Raise your attempt count by building speed in practice, not by forcing it in the exam.
Read the full lesson

Lesson 5

Sectional thinking and common traps: attempt targets, time sinks and panic guessing

Ananya's sections (+1 / -1/3, illustrative)
SectionAccuracyWorkingExpected marks per attempt
Reasoning90%0.90 - 0.10 x 1/3about 0.87
Quantitative Aptitude85%0.85 - 0.15 x 1/30.80
English75%0.75 - 0.25 x 1/3about 0.67
General Awareness50%0.50 - 0.50 x 1/3about 0.33

Set rules for how you decide, not targets for how many you attempt.

  • Work out expected marks per attempt for each section from your own accuracy.
  • Where a section has a minimum, raising accuracy there matters more than attempting more.
  • Attempt targets, time sinks, panic guessing and pattern myths all turn into blind guessing.
  • Change an answer only for a definite reason, and leave time to mark answers carefully.
Read the full lesson

Lesson 6

Net score drills: a faster formula and checking whether a score is possible

Net score = M x R - P x (A - R) = (M + P) x R - P x A.

Net = (mark + penalty) x right - penalty x attempts. To go backwards, add penalty x attempts and divide by the swing.

  • The fast formula needs only right answers and attempts.
  • To go backwards, R = (net + P x A) / (M + P).
  • R must be a whole number; otherwise the figures are inconsistent.
  • Always use the scheme of the section you are scoring.
Read the full lesson

Lesson 7

Confidence and calibration: using your honest chance of being right

The same chance, different schemes
Invented schemeBreak-even pEV at p = 0.3EV at p = 0.5
+1, -0.2520%0.1250.375
+3, -125%0.2001.000
+2, -133%-0.1000.500
+1, -150%-0.4000.000

Do not trust the feeling; trust the record. Attempt when your real hit rate at that confidence is above the break-even rate.

  • Break-even p = penalty / (mark + penalty).
  • Compare your real hit rate for a confidence level with the break-even rate.
  • Overconfidence is common; the log corrects it.
  • Small groups give unreliable rates.
Read the full lesson

Lesson 8

Marks per minute: ordering questions when time is short

Expected marks per minute = expected gain of the attempt / minutes it needs.

Gain = p x 1 - (1 - p) x 0.25
QuestionpMinutesExpected gainGain per minute
a0.910.8750.875
b0.730.6250.208
c0.620.5000.250
d0.510.3750.375
e0.320.1250.0625
f0.840.7500.1875

In the last minutes, fill the time with the questions that give the most expected marks for the minutes they use.

  • Compare questions by expected gain per minute, not by gain alone.
  • Rank by rate, then test an alternative when minutes do not fit.
  • Use the idea as a habit: sure and quick first, long and doubtful last.
Read the full lesson

Lesson 9

Luck and variance: judging your answering policy over many papers

Net score of 20 blind guesses (chances rounded)
Right out of 20Net scoreChance
2-4.007%
3-2.6713%
4-1.3319%
5+0.0020%
6+1.3317%
7+2.6711%
8+4.006%

A single paper is a small sample. Judge a policy by its reasons and its averages over several papers, not by one score.

  • Guessing results vary by chance from paper to paper.
  • The typical spread of results from a few guesses can be several marks.
  • Averages over several papers are steadier than one paper.
  • Judge a policy by the reasons for each answer as well as by the score.
Read the full lesson

Privacy · Terms · Refunds · Delivery