Implicit conversion from float -24.489795918367335 to int loses precision