Data Snooping Bias
Data Snooping Bias — Multiple testing inflating apparent significance.
Definition
Data Snooping Bias refers to multiple testing inflating apparent significance. Keep that definition fixed when comparing series, managers, or regimes — renaming the same tape does not create a new signal.
Why it matters
It is an operating object: if the definition drifts, routing, risk limits, and audit trails drift with it. When multiple testing inflating apparent significance shifts, related hedges, limits, and narratives usually need an explicit update rather than a quiet assumption.
Case
Suppose a desk is positioned for the opposite of what data snooping bias is saying. If multiple testing inflating apparent significance moves against that book, the first question is not “is the story clever?” but whether size, hedges, and stop logic still match the observation.
How to read it
Verify ownership, inputs, and failure alerts the same way you would for a production control. Prefer a short written null hypothesis for Data Snooping Bias: what would falsify the current reading in the next window?