So that S and Z return an integer, wich seemed cleaner to me.

The sorting part of my first solution is a ugly hack but it still passes most of the time :
https://www.codewars.com/kata/reviews/617af30f6e47f30001186417/groups/62ed7e18fd3e760001ede112

This also could be a more advanced sequel to this kata.

Missing random tests.
Please read and apply https://docs.codewars.com/authoring/guidelines/kata/.

Beautiful solution.

Indeed, good catch !

Fun kata !

May I suggest that you raise this as an issue, not a suggestion.
EDIT : It would also be useful to mention the language ... the error seems to be only in python.

Fun kata, harder than first expected (as often with g964 katas ...)

But a minor issue for python :
My solution is wrong for short strings and still passes most of the times.
I suggest adding the following test :

test.assert_equals( moving_shift("zokieciaxul", 24) , ['xnk', 'jgf', 'mfd', 'bt', ''] )

Fun kata

Indeed, apart from recursion, I'm summing over and over, I should sum once and subtract both ends... or start from the middle as you did.

I don't understand why we should return false for an empty array...
Am I the only one to think that true would make more sense here ?

Python :
Static test added.
Random generation slightly modified to avoid exceptions ( zero division and invalid input like "052" ) when generating a floating point number below 0.5.
Looks fine to me now.
Feel free to reraise if issue is not resolved.

Edit :
Static test also added for javascript.