A 23.73% range and double volume test whether a signal failed or the market did
Should the research team keep the signal live after a -6.90% week, or suspend it until calmer weeks show whether the loss came from the signal or from a 23.73% range?
The market facts are real; the decision maker is illustrative.
The week and the position
In the week of 6 October 2025, BTC fell 6.90%, from 123,482.32 to 114,958.80, across a 23.73% range on volume 2.05 times its prior 12-week average. The worst day, -7.31% on Friday, held the largest 1-hour move (-2.05% at 20:00 UTC) and the low of 102,000.00 at 21:00 UTC; Sunday gained 3.90%. The close sat in the middle third of the range, 8.77% below the 52-week high, in the 14th percentile of 399 weeks. The four prior weeks were mixed, ending 3.72%, -0.03%, -2.66% and 10.09%.
Judgment
One week cannot separate a broken signal from an unusually violent market. Keep the signal running in shadow mode at a fraction of normal size, logging every trigger and exit as if fully invested. This week's own facts argue against blaming the signal quickly: a 23.73% range on twice the 12-week volume, a Friday collapse followed by a 3.90% Sunday recovery, and near-zero funding (-0.0109% to 0.0100% per eight hours) describe two-sided turnover rather than a one-way trend any signal could have captured. Accept that the shadow record will judge attribution slowly.
What would change it
Suspend the signal if shadow results keep losing across several more weeks whose combined ranges stay well under this week's 23.73%, since volatility then no longer explains the losses. Return it to full size once the shadow record over a comparable stretch is flat or positive.
What sets this case apart
The case is about attribution, not forecasting: whether a -6.90% week on double volume broke the signal or exceeded the conditions it was built for.