Bootstrap
A method that estimates how much a result would vary between samples by resampling the data with replacement thousands of times and recomputing the result each time. The middle 95% of those results gives a confidence interval.
In plain English
It works for results with no simple formula, such as a median or a percentile, and usually agrees with the formula when one exists. It can't create values beyond the largest or smallest in the data, it doesn't fix a biased sample, and data that come in runs over time need versions that resample whole blocks.
Uncle Gary explains
Fifteen commutes with a median of 31 minutes: resampling them 200,000 times gives a 95% interval for the median of about 27 to 41 minutes.
Why it matters for your money
Bootstrapping a backtest's daily returns one day at a time ignores streaks and volatility clusters, and can make a strategy look steadier than it was. Market data need block resampling.
Where we tested it
Related terms
All glossary terms. Educational content, not financial advice.