Implicit conversion from float -24.76293103448276 to int loses precision