Implicit conversion from float -15.50373134328359 to int loses precision