In modular arithmetic, one defines classes of numbers based on the modulo. In other words, in modulo m arithmetic, a number n is equivalent (read: the same) to n + m, n - m, n + 2m, n - 2m, etc.
One defines m "baskets" and every number falls in one (and only one) of them.
Example: one can say "It's 4:30 pm" or one can say "It's 16:30". Both forms mean exactly the same time, but are different representations of it.
Thus both, the Python and C# results are correct! The numbers are the same in the modulo 5 arithmetic you chose. It would also have been mathematically correct to return (5, 6, 7, 8, 9) for example. Just a bit odd.
As for the choice of representation (in other words, the choice on how to represent negative numbers), that is just a case of different design choices between the two languages.
However, that is not at all what the % operator actually does in C#. The % operator is not the canonical modulus operator; it is the remainder operator. The A % B operator actually answer the question "If I divided A by B using integer arithmetic, what would the remainder be?"
— What's the difference? Remainder vs Modulus by Eric Lippert
Quick snippet to get the canonical modulus:
return ((n % m) + m) % m;
Test implementation:
Mono/C#:
machine:~ user$ cat mod.cs
using System;
public class Program
{
public static void Main (string[] args)
{
Console.WriteLine(Mod(-2, 5));
Console.WriteLine(Mod(-5, 5));
Console.WriteLine(Mod(-2, -5));
}
public static int Mod (int n, int m)
{
return ((n % m) + m) % m;
}
}
machine:~ user$ mono mod.exe
3
0
-2
Python:
machine:~ user$ cat mod.py
print -2%5;
print -5%5;
print -2%-5;
machine:~ user$ python mod.py
3
0
-2