StudyLumina - The exam readiness score. Not a feeling β a number
byβ’
Every study app answers the same question: what should I memorize?
StudyLumina answers a different one: am I actually ready?
It turns your real quiz answers into a 0β100 Exam Readiness Score per chapter, computed by a deterministic algorithm, never guessed by AI. Then it tells you exactly what to study today.
Replies
How does the Exam Readiness Score actually hold up when someone crams the night before and just guesses well on practice questions?
@kranur594186Β Love this question β it's the thing that makes or breaks a "readiness" score.
Two design decisions handle it directly:
Coverage gates the score, so lucky guessing can't fake it. Accuracy alone isn't the score β it's accuracy weighted by how much of the material you've actually been tested across. Ace a handful of questions and the ERS stays low and flags "Not Ready," because you haven't demonstrated breadth. You have to show accuracy across the whole chapter before it says "Ready," so a few lucky guesses on a thin sample can't get you there.
Cramming decays, and the score accounts for it. Recency follows a forgetting curve, with a separate retention signal from your spaced-repetition history. A night of cramming spikes short-term, but the score reflects that it won't hold, and because it's smoothed against your history, one hot quiz doesn't send it flying β it takes a consistent pattern.
The whole score is deterministic and computed in code β the AI generates and explains, it never decides your number. So it measures distributed mastery, not a good night or a good guess.
Finally, a study app that doesn't pretend to know me better than I know myself. The chapter readiness score felt oddly accurate after one quiz.
@erdalkoalaecjcΒ Really glad to hear that β "oddly accurate after one quiz" is honestly the best kind of feedback I could get. That's the whole bet behind the ERS: base it on your real answers instead of asking you to guess how ready you feel.
If you keep using it, curious to hear whether that accuracy holds up as you go deeper into a chapter (it should get sharper, not drift) β always happy to hear what breaks it too.
Curious how the readiness score handles mixed question types like essays vs multiple choice, does the algorithm weight them differently or treat every wrong answer the same?
@bernatokpn2jcuΒ Good question β they're not treated the same, no.
MCQ / short-answer are binary at the question level (right or wrong), but essays and written responses go through AI grading that scores reasoning quality, not just a final answer match β so a partially-correct proof or a well-argued-but-incomplete essay doesn't get crushed to a flat "wrong." That nuance feeds into the same underlying accuracy signal, just calibrated per format instead of pass/fail.
Where they do get treated the same is at the coverage level: every question, whatever the type, counts toward "how much of this chapter have you actually been tested on." The ERS won't call you ready off a handful of easy MCQs any more than a handful of essays β it needs breadth across formats before confidence climbs.