How Sample Size Changes ORB Statistics

Always show numerator and denominator

Write 12 of 20 qualified trades beside 60%, not the percentage alone. Also state whether the denominator contains all sessions, breakout sessions, completed entries, one direction, or a filtered subset.

Small samples move easily

One additional outcome changes 6 of 10 from 60% to 7 of 11, about 63.6%. The same additional outcome changes 600 of 1,000 from 60% to 601 of 1,001, about 60.0%. The smaller sample is more sensitive to each new observation.

This arithmetic does not make a large sample automatically relevant. Old regimes, data errors, or a mismatched rule set can make a large denominator misleading too.

Filters create conditional samples

A result filtered by direction, weekday, volatility, range width, retest, and time describes only sessions passing all those conditions. Show how many observations remain after each filter so readers can see when an apparently precise segment has become very small.

Repeated searching increases selection risk

Testing many tickers, windows, stops, targets, and filters and then reporting only the best combination makes chance results more likely. Preserve all tested variants and validate the selected rule on untouched later data.

Use a minimum evidence display

Every ORB statistic should show the rule version, provider, date range, qualified trades, matched sessions, exclusions, direction, costs, and whether the result is development or out-of-sample evidence. Use caution labels for small or incomplete samples instead of hiding them.