Implicit conversion from float -33.094170403587455 to int loses precision