How is the ViJEEta JEE readiness score out of 300 computed?

readiness = 300 × Σ over covered chapters of (weight_c / Σ weights) × mastery_c. Each chapter's weight is its measured share of past JEE Main papers; the covered set is rescaled to 300 so the score reads on the exam's own scale for any stage of preparation.

How does ViJEEta correct for guessing in multiple-choice questions?

Chapter mastery is a guessing-corrected IRT true score: m = mean over the chapter's items of max(0, (P(θ−z·SE, b_i) − c) / (1 − c)), where θ is the student's ability estimate, SE its uncertainty, b_i the item difficulty, and c the guessing floor. Ability is evaluated conservatively (θ minus z times its standard error), so lucky streaks decay and demonstrated skill persists.

Why do some chapters show a dash instead of a score?

Below 3 answered items in a chapter, mastery is reported as unknown (—). With a single observation, a Bayesian ability estimate is dominated by its prior — the model's assumption about an average student — which once produced readings like 23% mastery from one wrong answer. ViJEEta refuses to present the prior as a measurement.

What is the evidence ladder in ViJEEta?

Different work proves different amounts, enforced as caps on the ability scale: book work ≤ 0.60, unproctored practice and quizzes ≤ 0.75, tests uncapped. A chapter counts as mastered at ≥ 0.80, which only test evidence can reach. Opening a worked solution before answering earns zero ability credit.

How does ViJEEta compute study velocity?

velocity = Δreadiness ÷ days × 7 — the change in the /300 readiness over a Monday-anchored week, normalised to marks per week, compared against the target velocity implied by the student's goal and days remaining to the exam.

How does ViJEEta decide which chapter to study next?

marks_gainable_c = exam_weight_c × 100 × (1 − mastery_c) per subject. Chapters are ranked by recoverable marks, so a heavy weak chapter outranks a light strong one.

How are JEE Advanced mocks scored in ViJEEta?

Out of 360 (two papers × 180) with the real marking scheme, a per-subject qualifying floor and an aggregate cutoff, and a published marks→rank lookup table, so a mock score maps to an approximate rank.

Does ViJEEta repeat questions you have already answered?

No. A question answered correctly is retired from practice, quizzes, tests and generated mocks, unless the student bookmarks it. Real past papers are never filtered — a PYQ paper minus your answered questions would no longer be that paper.

How does ViJEEta schedule revision?

Per-chapter memory is tracked with an FSRS-style spaced-repetition model (stability and retrievability). Chapters whose predicted retention is decaying surface in the revision queue before they slip.

Methodology · published in full

How ViJEEta computes your JEE readiness score

Every number in the product is computed by the formulas on this page. We publish them because a score you can't audit is a score you shouldn't trust — and because honest measurement is the product.

How is the readiness score /300 computed?

readiness = 300 × Σc∈covered ( wc / Σw ) × masteryc

Each chapter's weight w is its measured share of past JEE Main papers — derived from a tagged bank of previous-year questions, keyed to a fixed curriculum tree. Only chapters you have covered enter the sum, and their weights are rescaled to 300: a Class-11 student five months in and a dropper in December both read readiness on the exam's own scale, for their syllabus.

This is deliberately not a predicted all-India score or percentile. A prediction needs the full syllabus and exam-day conditions; pretending otherwise is how apps flatter students into false security.

How is chapter mastery measured — and guessing corrected?

mastery = meani max( 0, ( P(θ − z·SE, bi) − c ) / ( 1 − c ) )

Mastery is an Item-Response-Theory true score: the expected fraction of the chapter's real questions (difficulties bi) you would answer correctly at your estimated ability θ. Two corrections keep it honest: ability is evaluated conservatively (θ minus z standard errors, so uncertainty lowers the claim, never raises it), and the multiple-choice guessing floor c is subtracted and rescaled — a blind 25% guess rate earns exactly zero mastery.

Why do untested chapters show “—” instead of a number?

nevidence < 3 → mastery = unknown (rendered “—”)

With one observation, a Bayesian estimate is mostly its prior — the model's belief about an average student, not about you. In testing, a single wrong answer produced “23% mastery”: roughly 80% prior, 20% student. Below three answered items in a chapter, ViJEEta reports unknown. This rule can only ever lower a claim, never inflate one.

What can each kind of work prove? (the evidence ladder)

Surface	Ability cap	Why
Book exercises (Tatva)	0.60	The worked solution is printed on the same page — weakest evidence.
Practice & daily quiz	0.75	Real evidence, but unproctored and untimed.
Tests, PYQ papers, mocks	uncapped	Exam-condition performance — the only evidence that can certify.

A chapter counts as mastered at ≥ 0.80 — reachable only through tests. The caps are derived from one another in the engine (book = practice − 0.15), so they can never silently drift apart. And the reveal-gate: open a worked solution before answering and that attempt earns zero ability credit — it still counts as reading coverage, but proof requires answering first.

How is study velocity computed?

velocity = Δreadiness ÷ days × 7 (marks / week, Monday-anchored)

Your weekly plan derives a target velocity from your goal score and the days remaining to your exam; the app measures your actual velocity against it. Progress becomes a rate you can steer, not a feeling.

How are chapter priorities ranked?

marks_gainablec = wc × 100 × ( 1 − masteryc )

Per subject (100 marks each in JEE Main), every chapter shows the marks still recoverable in it. The focus list ranks by recoverable marks — a heavy, weak chapter outranks a light, strong one automatically.

How are JEE Advanced mocks scored?

score /360 → rank ≈ lookup(marks) · per-subject floor + aggregate cutoff enforced

Advanced mocks use the real structure (two papers × 180), the real marking scheme including partial and negative marking, the per-subject qualifying floor, and a published marks→rank table — so a mock result answers “what rank is this?” instead of shrugging.

Does the app repeat questions? (question memory)

No. Every graded surface writes to one per-student question memory; a question you've answered correctly is retired from practice, quizzes, tests and generated mocks — unless you bookmark it to revisit. One deliberate exception: real past papers are never filtered, because a PYQ paper minus your answered questions is no longer that paper.

How does spaced revision work?

Per-chapter memory is modelled with an FSRS-style spaced-repetition system (stability × retrievability). When predicted retention decays toward the forgetting threshold, the chapter surfaces in your revision queue — before the exam discovers it for you.

The honesty rules, as constraints

Unknown is never rendered as zero. A revealed-solution attempt never earns ability. Every figure shown — ranks, weekly deltas, accuracies — is computed from your data or not shown at all. These aren't editorial policies; they are enforced in the scoring engine and locked by its test suite.

Open ViJEEta free →