Confidence-Calibration Practice for Estimation Questions
Give a 90% confidence interval, not a point answer - and learn to be honestly uncertain.
Free account required to play.
Estimation questions - the 'Fermi' problems interviewers love - aren't really about the number; they're about whether you know how uncertain you are. The Calibration Trainer asks you for a 90% confidence interval rather than a point estimate, then scores you on honesty: if you say 90%, the truth should land inside your interval about 90% of the time.
Most people are badly overconfident, giving intervals far too narrow. Repeated calibrated practice fixes that - teaching you to widen intervals to match your real knowledge, which is exactly the self-aware reasoning trading interviews reward.
What it's modelled on
Estimation and confidence-interval interview questions
Rather than one named test, this trains a theme that recurs across trading interviews: estimation ('Fermi') questions and requests for a range instead of a point answer, which probe whether you know how uncertain you are. Well-calibrated reasoning is what interviewers are really checking for.
NeetQuant is an independent study platform. It is not affiliated with, endorsed by, sponsored by, or connected to any firm named on this site. All company names and trademarks are the property of their respective owners. Trademark notice.
What it trains
- Estimation and Fermi-style reasoning
- Quantifying uncertainty with confidence intervals
- Calibration - matching stated confidence to real accuracy
How it's scored
Each round you give a 90% interval; you're scored on calibration (how often the truth lands inside) and tightness, not on hitting a single number. Free plays the fixed default at 90% intervals; Premium sets the confidence level (50, 80, 90, or 95%) and the length.
Related guides
Frequently asked questions
- What is a calibration question in an interview?
- Instead of a single estimate, you give a range you're (say) 90% confident contains the answer. It tests whether you understand your own uncertainty - a well-calibrated candidate's 90% intervals contain the truth about 90% of the time.
- How do I stop being overconfident on estimation questions?
- Practise giving explicit confidence intervals and check how often the truth actually falls inside them - most people need to widen their intervals substantially. The Calibration Trainer gives you that feedback loop directly.