Implicit conversion from float -24.03688524590163 to int loses precision