Better sleep analysis. When?

True story. Also should be possible to see right in the fenix3 some basic stats.

Unfortunately, the reality is that sleep trackers that claim they are actually analyzing how much sleep you get per night are guessing: they assume low movement == deep sleep, moderate movement == light sleep, and much movement == awake. This theory of measuring sleep quantity by wrist movement has been THOROUGHLY debunked, as it VASTLY over-estimates actual deep sleep time. Garmin's approach is reasonable for a movement-based wrist tracker IMHO: it only claims to monitor your movement during sleep.

The only wrist-based trackers that have a prayer of getting even within a one standard deviation of the accuracy of sleep studies are the ones that measure heart rate.

From where I sit, Garmin is actually doing the right thing from an engineering and liability standpoint: rather than giving you the pleasant lie of claiming that a wrist-based movement tracker can give you accurate sleep statistics, they are claiming what it actually can do. Give you accurate MOVEMENT statistics during the time you claimed to be sleeping.

That may be true txg_sync but a simple graph of movement during the hours of sleep nominated by the user is useless. It really does not tell you anything. You cannot even compare between different sleeps. If it is not possible to tell exactly (or even approximately) how much sleep was light / deep etc., then we at least need a number that is somehow representative of the amount of movement during sleep that can be compared day to day.

For example, why not just integrate the area under the graph to work out a 'score', at least I can say last night I moved a lot but the night before I didn't move much. You could include a disclaimer that the score may not be indicative of actual sleep quality, if necessary.