A trend signal posted its 89th-percentile week, but everything nearby rose too
Should the research team certify the momentum signal for live capital on this week's result, or keep it in paper trading because the week cannot separate skill from a rising market?
The market facts are real; the decision maker is illustrative.
The week and the position
In the week of 2 November 2020, BTC rose 12.45% to close in the upper third of its range, 2.45% below its 52-week high, with volume 2.19 times its prior 12-week average and a return at the 89th percentile of 142 weeks of history. Four prior weeks were all up. ETH rose more, beating BTC by 2.08 percentage points, and hourly correlation was 0.71. The worst day was Saturday at -4.89%. A research team asks whether its trend signal generated this result or merely followed the market.
Judgment
The week shows the signal does not break in an up market, but it does not show the signal works. Every listed prior week was up, ETH outperformed BTC, and the return sat near the top of 142 weeks of history, so a signal long BTC would have won regardless of its own merit. Keep the signal in paper trading. Spend the week instead on a relative test: whether the signal's ranked picks beat a simple BTC-and-ETH equal weight, and whether it avoided Saturday's -4.89% day better than that benchmark.
What would change it
If a week arrives in which BTC falls or trades flat and the signal's ranked picks still beat the equal-weight benchmark by a stated margin, that evidence certifies it. If the signal's live result diverges sharply from its paper result, its execution assumptions need re-testing before any size.
What sets this case apart
This case is about attributing a good week: a strong signal week and a strong market week look identical unless a relative benchmark is specified in advance.