Thanks for the numbers and you picked a great example. I plotted the values on a Parkes Error Grid, something used when the FDA evaluates CGMs, and your example is in Zone B: “Altered clinical action, little or no effect on clinical outcome.” As an example lets say you didn’t check with the BGM and just ate something to raise your glucose 30mg/dl. If the BGM was exactly right, you’d end up between 110-120.
So the accuracy is good enough but the alarm going off when it doesn’t need to is a problem. I hate it too, enough to sometimes quit the app so it can’t bother me. In the morning I have regrets. Every time. Having a quirky alarm is better than no alarm. Give it a few more years, you’ll see what I mean. In the meantime lets try to reduce the number of alarms due to something other than blood sugar.
You said “no arrow”, do you mean literally no arrow or 
? No arrow means the app thinks the number might be wrong, but the alarm has to sound after the second low reading. I was going to ask for a graph but then I remembered I have graphs… is it like this?
Or jumpy like this
Also can you grab that great illustration of an arm Joe posted and draw where you wear your G7 on it and maybe other places you have tried? Location can make a big difference.