Why we publish the worst depth in the window instead of the average
An average hides the exact event you are paying to avoid. Headline risk fields report the worst observed condition in the window, and the window is stated on every result.
Headline risk fields on every Crifine result report the worst observed condition in the window, not the mean. `lowest_depth_7d_usd` is the thinnest the book got over the trailing seven days, not what it typically was.
That choice looks pessimistic. It is really just an answer to the question actually being asked.
An average describes the days you did not need us
A depth series is valuable in proportion to how many stress events it contains. Someone paying for a pre-trade check is not asking what a normal Tuesday looks like — they are asking whether this specific order is about to clear badly. The middle of the distribution has no information about that. The tail is the entire product.
The same logic applies to protocol parameters. A liquidation bonus tuned to the average book is tuned to conditions under which the bonus was never going to matter. Tuning to the worst observed depth is the only version of the exercise that changes an outcome.
What we do not hide behind it
Reporting the stress case is not license to report an unfalsifiable one. `days_observed` sits next to every stress figure, because the worst depth in a seven-day window is a much weaker claim than the worst depth in two hundred days, and the reader is entitled to know which they are being handed. Confidence is a disclosed field here, not a marketing posture.