Implicit conversion from float -52.38410596026489 to int loses precision