BharatNotes There is no magic number; it is a function of your accuracy. With 85%+ accuracy attempt 90+. With 75-85% accuracy attempt 75-85. Below 70% accuracy, attempting more than 70 actively destroys your score. Toppers cluster in the 80-90 attempt band with 75-80% accuracy.
The simple maths every aspirant should memorise
Let A = attempts, r = accuracy (fraction of attempts that are correct).
Net score = A × r × 2 − A × (1−r) × 0.66
Simplifying:
Net score = A × (2.66r − 0.66)
Notice the break-even: when r = 0.66/2.66 = 24.8%, your score is zero. So unless you can get above 25% right, attempting hurts you. The real question is how high above 25% you can climb.
Score grid: what 87.98 cutoff (CSE 2024 General) actually demands
| Attempts | 60% accuracy | 70% accuracy | 80% accuracy | 90% accuracy |
|---|---|---|---|---|
| 60 | 56.16 | 71.04 | 85.92 | 100.80 |
| 70 | 65.52 | 82.88 | 100.24 | 117.60 |
| 80 | 74.88 | 94.72 | 114.56 | 134.40 |
| 90 | 84.24 | 106.56 | 128.88 | 151.20 |
| 100 | 93.60 | 118.40 | 143.20 | 168.00 |
Look at the 70% accuracy column — 70 attempts barely clear cutoff, but 90 attempts hit a comfortable 106. Now look at the 60% column — even 100 attempts cross cutoff, but only just (93.60). And at 50% accuracy (not shown), even 100 attempts only score 67 — a fail. Accuracy is the master variable.
Worked scenario 1 — the 'cutoff hugger'
You attempted 75 questions in CSE 2024. Of those, 38% are correct → 29 right, 46 wrong. Net = 29 × 2 − 46 × 0.66 = 58 − 30.36 = 27.64. You miss the 87.98 cutoff by 60 marks. The takeaway is brutal: when accuracy drops below 40%, even a heavy attempt count cannot save you. The fix is not "attempt more next time" — it is to train accuracy upwards before stepping into the hall.
Worked scenario 2 — the 'sweet spot'
You attempted 82 with 76% accuracy → 62 right, 20 wrong. Net = 62 × 2 − 20 × 0.66 = 124 − 13.20 = 110.80. This is exactly the topper-band number — well above CSE 2024's 87.98 and CSE 2025's 92.66, with safety buffer for a difficult year.
Worked scenario 3 — the 'aggressive elimination wizard'
You attempted 94 with 72% accuracy → 68 right, 26 wrong. Net = 136 − 17.16 = 118.84. Clears most years comfortably but the risk is real — a single bad day where accuracy slips to 65% on 94 attempts gives 124 − 22 = 102, still safe but margin thinning. Aggression rewards consistency, punishes off-days.
Worked scenario 4 — the 'over-attempter'
You attempted 100 with 55% accuracy → 55 right, 45 wrong. Net = 110 − 29.70 = 80.30. Just below CSE 2024's 87.98 cutoff, fails CSE 2025's 92.66 cleanly. The over-attempter loses because they over-attempt — every extra blind guess at 55% accuracy bleeds 0.34 marks.
Three honest profiles
- Aggressive elimination wizard (test-series veteran, conceptual depth, calm temperament): attempt 90-95, target 75-80% accuracy → 130-145 marks.
- Steady eliminator (most serious aspirants): attempt 80-85, target 75% accuracy → 110-125 marks.
- Conservative first-timer or risk-averse repeater: attempt 65-75, target 80%+ accuracy → 95-110 marks. Safely clears most years.
Topper voice — Shubham Kumar (AIR 1, CSE 2020)
"My target was around 95 questions because I always used to do a minimum of 20 wrong regardless of paper difficulty. My mock scores ranged from 56 to 110; mean and median around 80. Even where I had eliminated one option, I used to attempt by using some other logic."
Shubham failed Prelims on his first attempt in 2018, then took 70-75 mocks for CSE 2019 and 40-45 for CSE 2020. The discipline of knowing your own number came from data, not instinct.
How to discover your number
- Take the last 10 mock tests of your test series.
- For each, record attempts, correct, wrong, blanks, and net score.
- Calculate accuracy = correct ÷ attempts.
- Plot net score vs attempts. The peak of your personal curve is your sweet spot.
- Most aspirants discover their peak lies 5-10 attempts below what they instinctively want to do. Respect the data.
Common psychological traps in the hall
- "I've already attempted 60, may as well keep going" — sunk-cost fallacy. Stop when elimination dies.
- "Last 10 minutes, paper feels easy now" — easy-question illusion; UPSC plants traps in the last quarter.
- "I cleared Mains-level Polity, surely I can guess this" — Mains depth does not equal Prelims precision.
Mentor's rule of thumb
If you cannot eliminate at least two options, leave it. That one discipline raises most aspirants' scores by 15-20 marks overnight. With CSE 2026 Prelims on 24 May 2026, you have time to drill it on the last 15 years of PYQs — do it.
