A 99th-percentile rally with a -7.15% hour strains a mean-reversion desk's fixed daily loss limit
Should the desk cut trade size through a week this extended and this volatile, or keep full size and let its fixed daily loss limit decide when to stop?
The market facts are real; the decision maker is illustrative.
The week and the position
In the week of 28 December 2020, BTC rose 25.56%, from 26,281.54 to 33,000.05, with realised volatility at 98.0% annualised and a 34.38% range. The return ranked in the 99th percentile of 150 weeks. Yet Saturday 2 January carried a -7.15% hour inside a week whose best day gained 9.7% and whose worst day still closed up 0.17%. Funding averaged 0.059% per eight hours and reached 0.155%. An illustrative intraday mean-reversion desk fades sharp moves and stops at a fixed daily loss limit, sizing each trade for ordinary volatility, not this week.
Judgment
This week breaks the desk's premise: there was no down day, so reversion trades fade moves that kept going, while a -7.15% hour can hit a fixed loss limit before the reversion arrives. Cut per-trade size so the daily limit survives a repeat of that hour, and concentrate on the extreme moves rather than routine intraday dips. Keep the loss limit unchanged; it is the sizing, not the rule, that no longer fits.
What would change it
If hourly volatility and the largest one-hour moves return to the levels of the prior four weeks, with funding back near the 0.01% minimum, full size can resume. If a daily loss limit is actually breached this week, size must be cut immediately rather than at the week's close.
What sets this case apart
This case turns on trend persistence defeating mean reversion, not on the size of the rally itself.