Why would it be slower? Not that performance matters when you are doing code golf.

You can omit the ; and replace && with & to shave a few bytes

The result variable in line 2 is a "suprise global". Just so you know

I think this is just missing a few newlines to be perfectly readable production-quality code.

Haskell translation addded. https://www.codewars.com/kumite/5eafa013e24b830032704262?sel=5eafa013e24b830032704262

If you are going to reverse it wouldn't it be better to use tail which is more efficient?

getMiddle (reverse (tail (reverse (tail s))))

Or equivalently,

getMiddle . reverse . tail . reverse . tail $ s

Of course reverse . tail . reverse is just a less efficient (and stricter) init

getMiddle . init . tail $ s

I just realized that it would be better to use an std::array of int instead of char. The generated assembly also looks a bit nicer.

charGroups is the same as group' from Data.List

TIL: Haskell has a special case so that (- 1) is not a section equivalent to \x -> x - 1, it is actually equivalent to negate 1

You forgot to use divBy 2 instead of flip (//) 2

The boobs operator!