You should be able to determine by the string's length how many iterations it will take before it's back to it's original sort order. Then take the modulus of the iteration count and the input count, and only iterate that many times.
For example, a string that is three characters will be back to it's original sort order in 2 iterations. If the input count was to do 11 iterations, we know that 11 % 2 == 1, so we only need to iterate one time.
Once you determine a formula for how many iterations it takes to reach the original sort order for any length of string, you can always reduce the number of iterations to that number or less.
Coming up with a formula will be tricky, however. A string with 14 characters takes 12 iterations until it matches itself, but a string with 15 characters only takes 4 iterations.
Therefore, a shortcut might be to simply start iterating until we reach the original sort order (or the specified count, whichever comes first). If we reach the count first, then we return that answer. Otherwise, we can determine the answer from the idea in the first paragraph - take the modulus of the input count and the iteration count, and return that answer.
This would require that we store the values from our iterations (in a dictionary, for example) so we can retrieve a specific previous value.
For example:
public static string ShuffleChars(string source, int count)
{
string s = source;
var results = new Dictionary<int, string>();
for (int i = 0; i < count; i++)
{
s = string.Concat(s.Where((item, index) => index % 2 == 0)) +
string.Concat(s.Where((item, index) => index % 2 != 0));
// If we've repeated our original string, return the saved
// value of the input count modulus the current iteration
if (s == source)
{
return results[count % (i + 1) - 1];
}
// Otherwise, save the value for later
else
{
results[i] = s;
}
}
// If we get here it means we hit the requested count before
// ever returning to the original sort order of the input
return s;
}