LeetMath

An Avianna lab prototype · Track 1: Probability

Elite math training for adults with real-world intuition and limited time.

Master probability through short sessions, adaptive feedback, visual intuition, and hard problems that reveal exactly what kind of hard they are. Not a textbook, not a school course — a training arena.

One sitting, four outcomes

Open LeetMath for 20–30 minutes and leave with one of four things: a concept became intuitive, a known weakness was diagnosed, a problem-solving move became reusable, or a hard problem became less mysterious. Adults don’t need math made shallow. They need math made legible, compressed, recoverable, emotionally safe, and hard in the right order.

Solve the problem

40 original problems from Easy to Boss, each with a solution autopsy and a transfer variant. Tags are hidden until you commit, so the wording can’t cue the method.

Diagnose the miss

Wrong answers trigger specific feedback: the tempting wrong path, what assumption failed, and the repair drill that fixes it. Errors are diagnostic signals, not failure.

Climb the hint ladder

Hints preserve learning: first the diagnostic question, then the named move, and only then the solution. Never a jump from confusion to answer.

Build mastery

Every attempt schedules a spaced review. Clean solves wait a week; misses come back tomorrow. The goal is recognizing the idea later, when the surface changes.

Tagged by the kind of hard

Most platforms tag problems by topic and a difficulty number. LeetMath tags every problem by why it is hard — and each hardness dimension comes with the diagnostic question that dissolves it.

H1

Sample-space

Are you counting favorable and total in the same representation?

H2

Conditioning

What exactly became known, and how did it become known?

H3

Dependence

If one event happens, does it change the probability of the other?

H4

Symmetry

Is there an exchangeability hidden under the wording?

H5

Linearity of expectation

Can I count expected local successes instead of the global distribution?

H6

Recursion

After the first move, what smaller version of the problem remains?

H7

Invariant / martingale

What quantity has no drift from one step to the next?

H8

Bounding

What failure event can I union-bound, dominate, or concentrate?

H9

Existence

If I choose randomly, is the expected number of defects less than one?

H10

Representation

What is this problem secretly isomorphic to?

H11

Adversarial wording

What assumption did the wording tempt me to make?

H12

Proof control

Which statement would a skeptical grader reject?

Why probability first

Probability touches adult reality: hiring uncertainty, investing and downside risk, A/B tests, reliability and outages, medical base rates, games and adversaries, AI evaluation under noise. And it has the right elite arc — from counting and conditioning up to random walks, concentration, coupling, entropy, and the probabilistic method. Your real-world intuition is the raw material; the track converts it into formal reasoning.

How a session works

  1. 0–2 min
    Recall warmuptwo old cards from the review queue — retrieval plus spacing
  2. 2–14 min
    Concept and worked examplepredict first, then the visual model, then the formal move
  3. 14–21 min
    Main problemsolve with the hint ladder — deliberate practice at the edge
  4. 21–24 min
    Feedback autopsyclassify the error or explain the move — metacognition
  5. 24–25 min
    Commitone-sentence takeaway becomes a spaced memory card

In this prototype the arena, hint ladders, miss diagnosis, and spaced review are live. Lessons, visual models, and the diagnostic are on the roadmap.