A fixed daily loss limit meets a -14.96% day and the +12.95% rebound behind it
Should the mean-reversion desk cut position sizes so a day like Thursday's -14.96% fits inside its daily loss limit, or keep full size and accept being stopped before Friday's +12.95% rebound?
The market facts are real; the decision maker is illustrative.
The week and the position
In the week of 2 February 2026, ETH fell 7.95%, from 2270.16 to 2089.74, with a -14.96% day on 5 February followed by the week's best day, +12.95% on 6 February. Realised annualised volatility was 119.6%, the range was 37.12%, the largest 1-hour move was 4.27% late on 3 February, and volume of 19.95 billion dollars was 2.18 times the prior 12-week average. The close sat in the middle third of the range (52.7%). An illustrative desk buys dips expecting reversion and is stopped by a fixed daily loss limit.
Judgment
The reversion arrived, but a fixed daily loss limit sized for calmer weeks would have cut the position on 5 February and forfeited 6 February's +12.95%. Reduce per-trade size until a -14.96% day stays inside the daily limit, and enter in smaller tranches so the 4.27% hours do not force a full stop at one level. Accept the missed rebound as the price of staying funded; a limit that fires on the market's own worst day is not a rule failure but a sizing failure.
What would change it
If the worst-day and largest-hour magnitudes contract for several consecutive weeks and realised volatility falls well below 119.6%, sizes can be restored towards the original level. If the desk's records show it survived the week's low of 1747.80, reached early on 6 February, without being stopped, and still held size into the +12.95% day, the sizing case weakens.
What sets this case apart
One week of data cannot prove a reversion edge; the decision is about sizing a fixed loss limit to a 15% day, not about forecasting direction.