Implicit conversion from float -24.439655172413808 to int loses precision