Implicit conversion from float -55.65420560747663 to int loses precision