Implicit conversion from float -10.240384615384613 to int loses precision