Product · Analytics
Know what to practise next, not just how much you have done.
Every answer and simulator run feeds one view of where you stand: what you have solved, how accurately, where you are weakest, and whether your speed is improving. It turns “I should do more probability” into “I should do conditional expectation, and stop guessing Bayes questions.”
What it looks like
Sample data These are the real panels from the app, filled with an invented candidate four weeks into prep.
By topic
Questions solved per topic
- Expected Value88/215
- Probability142/397
- Brainteasers57/180
- Combinatorics61/305
- Statistics23/170
- Stochastic Processes6/159
Weak topics
Where to focus next
- Stochastic Processes6/159
- Statistics23/170
- Combinatorics61/305
- Brainteasers57/180
- Probability142/397
Mental Math Blitz trend
Standard-format runs, newest on the right
Probability OA - by topic
Accuracy and realized EV per question, weakest topic first
Across 6 sessions: net +44 over 90 questions - 0.49 EV/question, 81% accuracy when answering.
- Bayes' theorem4/11EV -0.09
- Markov chains & states4/8EV +0.25
- Cards9/14EV +0.50
- Dice15/19EV +0.74
Reading it
This candidate has put most of their time into expected value and probability, and the arithmetic trend is climbing. The weak-topics panel lists everything they have started but solved less than half of, least-covered first - and stochastic processes, at 6 of 159, tops it. For a research role that is the first thing to fix.
The OA panel shows a different problem. On Bayes questions they got 4 right out of 11 seen and still finished below zero: they answer when they should skip. The fix is not more Bayes questions; it is the rule on the right.
So next week: the Stochastic Processes & Random Walks track, and a Probability OA run where every Bayes question under 50% confidence is skipped.
Answer or skip?
- EV of answering
- −0.20
- EV of skipping
- 0.00
- Over 15 such questions
- −3.0
Skip it. Answering pays only above 50%: EV = 2p − 1.
Exactly what is measured
- Solved, attempted, accuracy
- Accuracy uses your latest answer to each question, so a problem you got wrong and later fixed counts as fixed - the number tracks what you know now, not your history.
- Coverage by topic and difficulty
- How much of each topic and each difficulty you have actually solved, against everything the bank holds.
- Weak topics
- Topics you have started but solved less than half of, lowest coverage first. Topics you have never touched are left out on purpose - otherwise a new account would see every topic flagged.
- Mental Math Blitz trend
- Your score on standard-format runs over time. Custom settings are excluded, so the trend compares like with like.
- Probability OA decision quality
- Accuracy and realised points per question for each topic, weakest first. Under +1 / 0 / −1 scoring this tells you where you are guessing too much and where you are skipping questions you could win.
- Streak
- Consecutive days with at least one attempt, counted in your own time zone.
Questions about analytics
Is progress tracking free?
Who can see my progress?
See your own numbers
Create a free account and your analytics start with the first question you answer.
Read next