Implicit conversion from float -23.766233766233753 to int loses precision