Rejecting because no random tests, missing includes, and a couple of other problems.

What's the goal of this fork? It does not really change anything.

I like your testing approach, but I have two notes:

1. Major issue: The random tests rarely ever generate any 'integral' value. Even if you run the loop a million times, you won't get a 1.0.
2. Suggestion: I highly suggest replacing rand with utilities from C++11's <random> header (Here is a Codewars guide on how to do so as well as some documentation here). The utility you're probably looking for is std::uniform_real_distribution.

It was probably translated by someone who either wasn't a python coder or was trying to stick to the original as close as possible.

This made me laugh out loud.

I feel like this is a good challenge. It should probably not be a level 7 kata but it is indeed worthwhile.