Hi!
The way that updates work *in all systems* (polling small lists of recent changes at intervals and hoping that this leads to a complete change history), it seems quite possible that such systems will sometimes miss an update, at least in the long run and under varying conditions (high server load, network troubles, update script down for a while, whatever). Insufficient update frequency is maybe not the biggest problem here (it should be in the range of one to a few minutes for all of the services).
Very important point with which I agree - it is completely possible that update polling misses an update, WDQS is no exception and it usually does not treat it as a problem, as the next update can fill up the missed one. However the ultimate truth source is on the wikidata site only. Beware of the caches though - if you ask for the same data on the same URL twice, I think you can get the same result even if the underlying data changed in the meantime.