A 37.46% range and one -11.74% hour judge a long ETH signal
Should the research team keep its long ETH signal live at full size, or suspend it until this week's result can be separated from the single -11.74% hour inside a 37.46% range?
The market facts are real; the decision maker is illustrative.
The week and the position
In the week of 3 February 2025, ETH fell 8.45% to close at 2,627.18, inside a 37.46% range between 2,921.00 (Monday 22:00 UTC) and 2,125.01 (Monday 02:00 UTC). The single largest hour, starting 01:00 UTC on Monday, lost 11.74%. Volume ran 1.41 times the prior 12-week average, realised volatility reached 140.8% annualised, funding averaged -0.0017% per eight hours, and ETH lagged BTC by 7.18 percentage points at a 0.81 hourly correlation. Three of the previous four weeks were down. An illustrative research team runs a long ETH signal and must judge whether it survived.
Judgment
A week containing one 11.74% hour cannot by itself confirm or reject a signal on a longer horizon, even though -8.45% sits in the 15th percentile of 364 weeks. Keep the signal's rules unchanged, re-run them on the week with that hour and the 2,125.01 low isolated, and hold live size down meanwhile. ETH closed at 63.1% of its range, in the middle third, after three down weeks in four. The team accepts an unconfirmed signal and cannot yet separate decay from shock.
What would change it
If the re-run without the 01:00-02:00 UTC Monday window still loses beyond the signal's historical drawdown tolerance, suspend or rewrite it. If a later week closes above the 2,869.68 open with volume below the 1.41 multiple, restore full size.
What sets this case apart
The call is about attribution, not direction: whether one -11.74% hour may decide a signal's fate when ETH moved with BTC at 0.81.