Implicit conversion from float -161.68831168831167 to int loses precision