Python / pandas function to replace '-' negative sign but not negative sign in negative numbers ( '-' in '-1')

There is no need for .replace() at all. Find the cells that contain the dash, and update them:

df[df['Values'] == '-'] = np.nan

Bear in mind that 'NaN' is not a NaN: it is a string that looks like a NaN. A "real" NaN is np.nan from numpy.

You can cast the value of the column as numeric. Then maybe - will be replaces by 0, not sure if you want it :

pandas.to_numeric(arg, errors='raise', downcast=None)

Reference

If it doesn't work, astype may be an alternative

DataFrame.astype(dtype, copy=True, errors='raise')
# Example :
df.astype({'values': int})

Reference

Nevertheless the most elegant solution, and the most often used is to use th pandas selector:

df[ df['Values'] == '-' ] = np.nan

Also you can choose to replace the value - by anything you want: 0, Nan (np.nan), etc.