The negative-marking maths for CAPF Paper I, the attempt-or-skip rule, timing and pacing, and concrete accuracy targets, authored guidance
This note turns the CAPF Paper I mark scheme into decisions you can make in real time: when to attempt, when to skip, how fast to move, and what accuracy to aim for. It is the operating manual for the mocks in full mock 01 and full mock 02 and for the live paper. The maths below is general and durable; the live design (about 125 questions, 250 marks, 2 hours, one-third penalty) should still be re-checked against the current UPSC notification, because UPSC can revise it.
Authored guidance, not a verbatim PYQ. The mark scheme used here is the standard UPSC objective scheme; verify the latest before the exam.
Each question carries +2 for a correct answer, minus one-third of 2 (precisely minus 2/3, that is minus 0.66) for a wrong answer, and 0 for a blank. There is no penalty for leaving a question, and no extra penalty for a "more wrong" option; every wrong option costs the same 0.66.
The single most useful idea is expected value (EV): the average marks a decision earns if you repeated it many times. For a guess, EV = (probability correct x +2) plus (probability wrong x minus 0.66).
So a pure blind guess is essentially EV-neutral. Across a whole paper of blind guesses you would hover around zero, but with real variance (you could lose a chunk on a bad run). Conclusion: do not blind-guess as a habit, because the upside is nil and the variance is real, but a stray blind guess is close to harmless.
| Situation | EV | Decision |
|---|---|---|
| You know the answer | plus 2 | Attempt |
| You can eliminate two options | about plus 0.67 | Attempt (educated guess) |
| You can eliminate one option | about plus 0.22 | Attempt (lean in) |
| You can eliminate nothing | about zero (break-even) | Skip by default; a single blind guess is near-harmless but adds variance |
The crisp version: if you can rule out even one option, attempt; if you can rule out none, skip. Elimination, not raw recall, is what makes CAPF negative marking pay.
You can almost always shave at least one option without full knowledge:
The clock, not the difficulty, is the binding constraint on CAPF Paper I.
Translate the EV maths into numbers you can chase in the mocks.
| Metric | Target | Why |
|---|---|---|
| Attempt rate | 85 to 95 percent of the paper | Skipping more than 15 percent usually means leaving positive-EV guesses on the table |
| Accuracy on attempted | 75 percent or higher | Below this, the penalty erodes too much of the gain |
| Wrong-answer count | Keep well below your correct count | Each wrong costs 0.66; runaway wrongs sink the net |
| Blind guesses | Near zero | Convert blind guesses into one-elimination guesses or skips |
A clean shape for a strong attempt: attempt about 90 percent, get about 80 percent of those right, leave the genuinely unknowable blank. On a 125-question paper that is roughly 112 attempted, about 90 correct, about 22 wrong, 13 blank, for a raw score near (90 × 2) minus (22 × 0.66) = 180 minus 14.5 = about 165 out of 250 before considering Paper II. Use this only as a self-benchmark; the actual qualifying total depends on the year and on Paper II, so verify the latest cut-offs.
Grade each mock with the penalty (see the scoring method in Index), then triage every wrong and blank into knowledge gap, silly error, elimination miss, or time-out. The fix for a silly error is reading discipline, not more content. The fix for an elimination miss is rereading this note and forcing the attempt-or-skip rule on the next mock. Over four mocks, your attempt rate should rise toward 90 percent while your accuracy holds above 75 percent.