you are on the clearnet. the addresses listed here only open inside the tor network - download the tor browser here »
AlphaBay.Market
last update: 0 min ago 255 onions tracked
home / news / site updates
21 August 2026 site updates 4 min read

Uptime Stats Now Use a Rolling 7-Day Window — Here's Why That Matters

A mirror goes down for six hours in March. Nobody notices, it comes back by April, yet its uptime percentage stays haunted for months. That single bad afternoon was enough to make our numbers useless — so we rebuilt how they are calculated.

The problem with forever averages

All-time uptime is a running average of everything that ever happened, weighted equally across eternity. A service that was flaky in January still looks unreliable in August, even if it has been rock-solid since spring. The number stops being a status report and becomes a museum exhibit. This is not a quirk unique to dark web trackers. The same math governs commercial SLAs: as the calculators at uptime.is show, "three nines" allows about 43 minutes of downtime per month but nearly nine hours per year — which is precisely why the reporting window you pick changes the story a percentage tells. So we changed our window.

What actually changed

As of this week, every uptime figure on our onion status checker and service listings is computed from a rolling 7-day window. Only checks from the last seven days count. Older results fall out of the math day by day until they carry no weight at all. The result is a number that tracks reality. A mirror that recovered from last month's outage shows green again today. A service quietly degrading this week sees its score slide within days, instead of the signal being diluted into invisibility by months of clean history. If history is your thing, nothing was deleted. Every individual check result remains visible in the recent sweeps list on each service page, so you can judge whether a green percentage rests on hundreds of clean probes or just a handful.

Gaps now count as online

One more change deserves an honest explanation. Our tracker cannot watch every address every minute, so brand-new entries and interrupted sweeps leave stretches with no data at all. The old method treated those gaps as failures — punishing services for our blind spots rather than their own. That felt backwards. Missing periods are now neutral: they count as neither downtime nor a black mark. The score reflects only what we actually measured. For readers, the practical effects:
  • Uptime reflects the last 7 days only, not lifetime history.
  • Data gaps count as online instead of counting against a service.
  • New entries start near 100% and earn their real score within days.
  • Full per-check history stays available on every service page.

How to read uptime numbers anywhere

This update doubles as a lesson in reading availability stats anywhere, not just here. Whenever a site advertises "99.9% uptime," the first question should be: measured over what period? A lifetime average hides recent problems; a daily snapshot can hide chronic ones. The standard formula — uptime divided by total time, as laid out in guides like Hyperping's SLA calculator — means nothing without knowing both sides of that fraction. Second question: who holds the stopwatch? Numbers generated from a service's own dashboard tend to look better than figures from an independent checker with its own vantage point and blind spots. We publish our methodology so you can judge it — and disagree with it. Third: what happens when data goes missing? Any monitor can be blinded by network trouble or an address that refuses automated visitors. Whether those gaps count as failures or get excluded says a lot about how candid a stats page wants to be.

A choice, not a law of nature

Seven days sits in a useful middle ground — long enough to smooth out one weird blip, short enough that real trouble surfaces fast. But the rolling window is a design decision, not an inevitability, and we would rather explain it than pretend otherwise. If a number looks wrong, or a service's score does not match your experience, tell us. The whole point of a status list is that it matches what users actually hit. More methodology notes will land in the site updates feed, and the faq will stay current as details evolve.

more notes

all news ›