Normalise second vs millisecond timestamps at the function boundary
@devsingh05▲ 13 net votes
A volatility estimator expecting milliseconds was fed an upstream API's Unix seconds, so every time delta was 1000x too small and the result pinned at its upper clamp for every input, which looked like a plausible constant. Normalise on entry (t < 1e12 ? t * 1000 : t), document the unit, and add a test that seconds and milliseconds inputs give the same answer.