A signal posted a 9.55% week, but 5.05 points of it came from one Monday
Should the research team mark the signal validated after an 85th-percentile week, or keep it provisional because most of the week's return came from a single day inside a four-week winning run?
The market facts are real; the decision maker is illustrative.
The week and the position
In the week of 4 December 2023, BTC rose 9.55%, from 39,972.26 to 43,789.51, ranking in the 85th percentile of 303 weeks of history. Best day Monday returned 5.05%; the worst day, Thursday 7 December, returned -1.12%. The four prior weeks were all up (5.86%, 0.8%, 0.23%, 6.74%). Volume ran 1.86 times the prior 12-week average, funding averaged 0.0142% per eight hours, and the largest one-hour move was only 1.55%. An illustrative research team must decide whether its signal is validated.
Judgment
The week clears the team's bar on size and volume, but the evidence is concentrated. One Monday supplied 5.05 of the 9.55 points, and the run already includes four consecutive up weeks, so the signal has not been tested against adversity. Validate the signal's direction, not its expected return, and keep position size at the level set before the run rather than raising it on this percentile reading. Report the result with Monday separated from the rest.
What would change it
If the signal's weekly return remains positive when Monday's 5.05% is excluded, the concentration objection weakens and sizing can rise. If a later week closes below this week's open of 39,972.26 during the run, treat the streak as finished and re-test.
What sets this case apart
A high-percentile week can validate direction while failing to validate return size; the loss is a mis-scaled conviction, not a bad forecast.