Hi, got one question. Sure it has a evident answer, but I'm failing to see it right now. We have the following configuration: time_bucket: period: hour count: 1 seasonality: hour_of_week detection_period: period: day count: 1 training_period: period: day count: 4 We are expecting to see that the sensitivity range is calculated using the last 4 samples (without the current sample which is the one where we want to check any anomaly). Thing is, we've found that it is using 4 samples, but including the current, which it induces sometimes to false negatives. Are we missing something? Any configuration detail not set properly? Or is it the way it usually works? Thanks a lot for your comments