The rolling average: smoothing out the noise
Launch library · evergreen read
A single day's figure can swing for reasons that have nothing to do with any underlying trend: a public holiday, a reporting delay, or simple random variation can all push a number up or down without meaning anything has actually changed. A rolling average smooths that noise out by blending several recent points together.
The trade-off is a short delay: a rolling average reacts to a genuine shift more slowly than a single day's number would, because it takes a few data points before the new pattern outweighs the old ones already baked into the calculation. That lag is usually a fair price for a much steadier picture.
When a chart looks jagged and alarming, checking whether it shows daily figures or a smoothed average is often the fastest way to tell whether something has genuinely changed or whether the apparent spike is simply noise that a slightly longer view would have absorbed.