How scoring works

Scaled score (100–900)

CompTIA reports results on a scale from 100 to 900. Each exam has its own cut score, listed with its blueprint below. AmpUp uses a simple linear approximation of that scale:

scaled = clamp(100, 900, round(100 + (earned / possible) * 800))

CompTIA does not publish the real conversion — theirs is item-response weighted so a hard question is worth more than an easy one, and the cut score is stable across forms. We show the raw percentage next to the scaled score so you always see the underlying ratio. Treat the scaled number as a reasonable proxy, not a promise. Pass or fail is judged against that exam's own passing score, read from the database — nothing is hardcoded.

Blueprint-weighted question selection

AmpUp picks questions so each domain gets its blueprint share, using the largest-remainder method (the same one used in proportional-representation elections). If your question bank is thin in a domain, the engine caps the count at what is available and shows a warning banner.

Within each domain the engine randomises and slightly deprioritises questions you got right recently, so re-taking an exam doesn't drill the same items.

Readiness score

Readiness estimates what you'd score right now on a real exam. For each domain we compute mastery as your correct-answer rate over the most recent 30 questions in that domain, across all modes. Domain mastery is combined by blueprint weight and mapped onto the 100–900 scale:

readiness = 100 + (Σ mastery_d × weight_d) × 8

Readiness stays honest about thin coverage. Any domain with fewer than 10 answered questions is flagged as low-confidence, and the dashboard verdict falls back to "Not enough data yet" until every domain crosses that bar — even if the raw number looks good. The verdict compares readiness with that exam's own passing score.

Flashcard scheduling (SM-2)

Flashcards use the classic SuperMemo-2 algorithm. Each card has an ease factor (starts at 2.5) and an interval in days. After you rate a card, the interval grows by ease if you got it right (1 day → 6 days → interval × ease…), and resets to 1 day if you missed it. Ease drifts up on easy ratings and down on hard ones, floored at 1.3.

Cram mode is different on purpose. It walks through a deck without touching your scheduling state, so a pre-exam blitz doesn't blow up your intervals or ease factors. Nothing you do in cram is written to flashcard_states.

Performance-based questions (PBQs)

Real CompTIA exams open with a handful of hands-on questions — drag-to-order sequences, matching items to targets, and mini scenario forms. They appear first, they count inside the question cap (so more PBQs means fewer multiple-choice items), they can be flagged and revisited like any other item, and they carry more weight than a single multiple-choice question. Partial credit applies: getting two out of three matches right is not the same as getting them all wrong.

Each PBQ is worth 3 points and each multiple-choice item 1 point, so an exam's total points depend on its PBQ count — for a 90-question exam with 4 PBQs that is 98 points. Your raw score is points earned divided by points possible, then mapped onto the 100–900 scale with the same formula above. CompTIA does not publish its real PBQ weighting, so this is our approximation — not a claim about the actual exam.