Why this page exists

This is a companion to Dealing with randomness.

Most personal websites are highlight reels. This one mostly is too. This page is a small correction.

It lists a few things that didn’t work out — closed loops from ~6–8+ years ago — to document that trajectories are noisy and outcomes are not monotonic.

This is mainly for:

  • juniors who might be over-indexing on a single exam / competition / workplace / outcome
  • people trying to infer ability from one data point

Treat these as counterexamples.

Why only older failures?

  • Signal vs noise: Recent failures are still entangled with current ability. Older ones read more clearly as variance.

  • Adverse selection: Making recent misses highly visible can be net-negative in the short term.

This is a retrospective slice, not a real-time log.

Selected entries

  • 2017 — JEE Mains (~50k rank)

    Didn’t prepare seriously.

    Outcome: went to NUS instead

  • 2019 — Jane Street Dev Internship (No Return Offer)

    Didn’t convert.

    Outcome: moved on DRW → Citadel → current role

Closing note

None of these felt small at the time.

They mattered, just not as much as they felt.