My observation with two watches with wrist heart rate (Forerunner 235 and Fenix 5+) shows that the problem with the wrist heart rate is not the “absolute” accuracy when, e.g., running with a roughly constant pace, but the delay when my heart rate rises. What I mean is, fast heart rate changes, such as very short spikes (when, e.g., climbing, weight lifting or doing interval training with short intervals) are just not recognized by the wrist heart rate sensor. Here I have two examples:
With chest heart rate strap:
With the optical sensor from the watch:
it’s both indoor climbing and I’ve just tracked it for fun. But anyway, you see on the elevation graph that the watch recognized the “altitude” change even indoors (which amazed me). But you can then easily correlate whenever I was climbing, because my heart rate went up, whereas on the one which I tracked with the wrist sensor you see nothing. Also the summary below shows 107 – 176bpm (when measured with the chest heart rate strap) vs. 87 – 118bpm (when measured with the optical sensor). Of course these were two different sessions, but I’m pretty sure my heart rate in the second session was above the given value.
And as you can see from both graphs, by “short intervals” and “fast heart rate changes” I mean roughly 1.5-2 minutes, which is a lot! Especially, when the experts here (Scott and Steve) are talking about “do an interval of 8 seconds” xD