• 10 random tests are not enough to prevent false positive solutions.
• Edge cases should be specifically tested.

Python: the random test assumes the class must have a public property named balance (or its getter), but it's not specified anywhere in the Kata description. The tests should not rely on the implementation.

Python: tests have dependencies between test cases, Fixed tests depend on the state created by Sample tests.

The method that you named as "check" is not really understandable. Please can you change the name with one more describable like transfer_money or something like that?

• Python new test framework should be used (Refer this & this for more detail)

You should ban a lot more than "import" and "sys" - https://www.codewars.com/kata/reviews/5a484c87425057ac80002891/groups/5cd46dc1d4c4470001f665c4