Fairly often it happens that a problem occurs in production that we do not have a proper trigger in place to catch.  When that happens, very often it becomes a thought experiment as to what the trigger logic to detect the issue might look like.

If that thought experiment gets into things like "if the value of some item as compared to itself time shifted one week ago and averaged over the span of an hour varies by more than X...",  the path becomes  sub-optimal.  That's not something one can really properly test using the trigger constructor tool that Zabbix currently provides.

Instead what ends up happening is that  you think through the problem as best you can, put together some trigger logic for it, and hope you got it right - because you can't test it "in context".

It would be really nice if Zabbix had a tool that let one construct trigger logic and test that logic against historic data.  So  you  could say things like "the value of this item one week ago averaged over 3 hours" and actually test to see if your logic would have caught the event that happened.  It'd sort of be like being able to "replay" the historic event through some work-in-progress trigger logic to see what would happen.