Implicit conversion from float -40.424528301886795 to int loses precision