Implicit conversion from float -41.62393162393164 to int loses precision