It rules out your device and your general connection speed, since VOD would struggle too if either were the real problem. Live delivery arrives as a continuous run of roughly six-second segments requested one after another, with no room to pre-load ahead the way an on-demand file can. That makes live far more sensitive to a session stalling or a source feed having trouble, and much less sensitive to raw bandwidth than most people assume. If someone tells you to run a speed test first, this split is the reason that advice usually wastes your time: the number it returns rarely explains what you're actually seeing on the live side.

That specific split points at the individual channel's source feed rather than at Smarters, your device, or the service as a whole. Switch away and back to confirm; if it clears, the previous session had stalled and nothing structural was wrong. If it buffers again immediately on return, the source feed itself has a live problem, and no setting inside Smarters reaches that layer. Try a neighboring channel to double-check the split before doing anything else. There is nothing to gain from reinstalling the app or resetting your router in this case, since the fault never reaches either of those layers, and time spent there is time not spent waiting out the one feed that is actually affected.

That window is peak viewing time, and buffering that clusters there points to contention: either your local network is more loaded at that hour, or the service has more concurrent viewers than its capacity comfortably serves. Test the same channel on a mobile hotspot at the same time. If it stalls there too, your setup is exonerated, and the honest next question is what uptime figure the service will actually commit to, not another settings tweak. Keep a simple log for two or three evenings: the channel, the time, and whether it cleared on its own, because a consistent window is much easier to act on than a vague sense that things are worse at night.

Two settings move the needle for most people: drop buffer depth to none or minimum, since a deep buffer slows recovery more than it prevents stalls, and switch to a public DNS resolver to shorten the lookup ahead of each segment. Hide channel groups you don't use, since Smarters holds the full visible list in memory and large lists strain lower-powered devices above roughly 18,000 channels. Apply all three once, note the date, and treat any buffering that returns after that as a fresh symptom to scope again rather than proof the fix did not work.