Just had a look at some metrics that I haven't looked at for ages, and this looks odd:

I think this is coming from the wrist (I don't have a HR strap that provides running metrics), and I can't begin to understand how on earth one could construct an algorithm to get this right from wrist movement, with arm swing going on, so I guess this is just not entirely optimal.
It is probably time to dust off my Running Dynamics pod.
Most of you will probably use a HR strap with running dynamics, but if you don't, do you see the same thing?

