📊 Prelims PYQ Analysis — Decoding What UPSC Actually Asks
Based on observed patterns across roughly the last decade of GS Paper I. Use this to prioritize effort; the mocks in this folder are built to mirror these patterns.
Subject-wise weightage (approximate, GS Paper I, out of 100 questions)
| Subject | Typical range | Trend |
|---|---|---|
| Polity & Governance | 12–20 | Increasingly application/scenario-based; fewer rote articles |
| Economy | 12–20 | Conceptual + current (schemes, RBI, budget); analytical statements |
| Environment & Ecology | 12–18 | High weight; often merged with S&T & current affairs; species/conventions |
| History (Ancient+Medieval+Modern) | 12–17 | Art & Culture rising; Modern still core |
| Geography | 8–15 | Physical geo conceptual; India mapping; increasingly integrated |
| Science & Tech | 7–12 | Almost entirely current-affairs driven (space, biotech, IT, defence) |
| Current Affairs (standalone) | 12–18 | Woven through every subject; reports/indices/appointments |
| Art & Culture | 3–7 | Rising trend within History bucket |
Takeaway: No subject can be skipped, but Polity + Economy + Environment + Current Affairs form the ~60-question backbone. Master these first.
Question format trends (this is what you must train for)
- "How many of the above statements are correct?" — the dominant modern format. UPSC shifted from "1, 2 and 3" style to "Only one / Only two / Only three / All four." This kills partial elimination. You must know each statement independently.
- Multi-statement (2–4 statements), choose correct combination — classic; elimination still works if you're sure of one statement.
- Matching / pairing ("Consider the following pairs" — how many correctly matched).
- Assertion–Reason (less frequent now but still appears).
- "Which of the above is/are correct?" with 3 statements.
- Direct single-fact (rare, mostly in S&T/current affairs).
- Odd-one-out / correctly-describes type in environment & economy.
Implication: Guessing is riskier than ever. The "how many are correct" format defeats the old trick of knowing 2 of 4 statements. Depth per fact > breadth of facts.
Recurring themes UPSC loves (build focused notes on these)
- Constitutional bodies vs statutory vs regulatory vs executive — perennial confusion tested every year.
- Committees & their recommendations (economy & polity).
- International conventions/protocols & their focus (environment: Ramsar, CITES, CMS, Basel, Stockholm, Rotterdam, Minamata, Cartagena, Nagoya).
- Government schemes — nodal ministry, objective, beneficiary, features (2–3 statements per scheme).
- Species in news — habitat, IUCN status, protection schedule.
- Places in news — location (which country / which state), why in news.
- Indices & reports — publishing agency (World Bank vs WEF vs UN body vs private).
- Historical: Buddhist/Jain councils, Gupta-era science, Sangam literature, temple architecture styles, INC sessions & who presided.
- Economy: monetary policy instruments, types of deficits, GST structure, indices (CPI/WPI), external sector terms.
- Geography: climatic phenomena (El Niño/La Niña/IOD), local winds, straits & their connecting water bodies, national parks & their states.
The "elimination & probability" strategy (topper method)
- Attempt in 3 rounds. Round 1: 100% sure (aim 40–55). Round 2: educated (eliminate ≥1 option). Round 3: leave or calculated guess.
- Negative marking math: −1/3 per wrong. A blind 4-option guess has +1/4 chance right; expected value slightly negative. But if you eliminate 1 option (3 left), EV turns positive. Guess only when you can eliminate at least one option.
- Optimal attempts: Historically, ~80–90 attempts with ~75%+ accuracy clears comfortably. Don't chase all 100.
- "Extreme word" caution: statements with "only," "never," "all," "must," "first," "largest" are more often wrong — but verify, don't reflex.
- Two-statement questions: if both look right and are logically linked, UPSC often makes both correct; if one is an extreme claim, suspect it.
Cut-off intuition (GS Paper I, general category)
Varies year to year (roughly high-80s to low-100s out of 200). Target 110+ in your mocks to build a safety buffer. CSAT only needs 66.67/200 — but never neglect it; it has ended many campaigns.
How to analyze YOUR mocks (more important than the mock)
For every question, after scoring, tag it: - ✅ Right & sure — good. - 🍀 Right but guessed — treat as a gap, add to error notebook. - ❌ Wrong (concept) — revise the source topic fully. - ❌ Wrong (silly/misread) — process error; slow down, underline the negative/"NOT" words. - ⬜ Left but was answerable — you under-attempted; recalibrate risk appetite.
Track the ratio each mock. Rank 1 comes from shrinking the 🍀 and silly-❌ buckets to near zero.
Now attempt the 8 mocks in 8-Full-Prelims-Mocks/ under strict timed conditions.